Commission Staff Working Paper:The EU Economy:2004 Review - Hoofdinhoud

COUNCIL OF Brussels, 3 November 2004 THE EUROPEAN UNION

14191/04 ADD 1

ECOFIN 352 SOC 521 EF 46

COVER NOTE

from : Secretary-General of the European Commission,

signed by Mrs Patricia BUGNOT, Director

date of receipt : 29 October 2004

to : Mr Javier SOLANA, Secretary-General/High Representative

Subject: Commission Staff Working Paper: The EU Economy: 2004 Review

Delegations will find attached Commission document SEC(2004) 1368.

________________________

Encl.: SEC(2004) 1368

COMMISSION OF THE EUROPEAN COMMUNITIES

Brussels, 26.10.2004 SEC(2004) 1368

COMMISSION STAFF WORKING PAPER

The EU Economy : 2004 Review

ACKNOWLEDGEMENTS

The “EU Economy 2004 Review” was prepared under the responsibility of Klaus Regling, Director General for Economic and Financial Affairs. Executive responsibilities were attached to Servaas Deroose, Director for the Economy of the euro area and the Union, and Jürgen Kröger, Director for Economic Studies and Research.

Primary contributors to this report include Alfonso Arpaia, Salvador Barrios, Manfred Bergmann, Guiseppe Carone, Nuria Diez Guardia, Marie Donnay, Carole Garnier, Martin Hallet, Mark Hayden, Dermot Hodson, Fabienne Ilzkovitz, Joost Kuhlmann, Kieran Mc Morrow, Gilles Mourre, Gaetan Nicodeme, Karl Pichelmann, Bartosz Przywara, Werner Röger, Aino Salomaki, Jean-Bernard Sauner Leroy, Michael Stierle, Michael Thiel, Johan Verhaeven, Reinhilde Veuglers, James Watson, Maxwell Watson and David Young.

The report benefited from contributions and comments by staff of the Directorate General for Economic and Financial Affairs, including Sophie Bland, Carsten Brzeski, Declan Costello, Adriaan Dierx, At Draaisma, Elena Flores Gual, Stefan Henseler, Lars Jonung, Filip Keereman, Sven Langedijk, Andrea Montantio, Aude Neuville, Heikki Oksanen, Moises Orellana, Dario Paternoster, Eric Ruscher, Jan H. Schmidt, and Klaus Wälde.

Statistical assistance was provided by Danila Conte, Cécile Denis, Karel Havik, Jouko Kuosmanen, Jérôme Saulnier, and André Verbanck. Secretarial support was provided by Daphné Corne, Cecilia Mulligan and Marianne Vermeulen.

Comments on the report would be gratefully received and should be sent to:

Directorate-General for Economic and Financial Affairs

Unit Co-ordination of the economic policies of the Member States and the euro area

European Commission

Rue de la Loi 200

B-1049 Brussels

Or by e-mail to Joost.Kuhlmann@cec.eu.int

TABLE OF CONTENTS

Summary and main conclusions .......................................................................................................................5

• 1. 

  Macroeconomic development in the euro area .........................................................................................7

• 2. 

  Catching-up, growth and convergence of the new Member States .........................................................55

• 3. 

  Labour markets in the EU: an economic analysis of recent performance and prospects ......................111

• 4. 

  The Lisbon strategy and the EU’s structural productivity problem ......................................................155

• 5. 

  The link between product market reforms and productivity: direct and indirect impacts .....................189

• 6. 

  Protecting the environment and economic growth: trade-off or growth-enhancing structural

adjustment .............................................................................................................................................217

• 7. 

  Ongoing issues in EU economic surveillance ........................................................................................247

SUMMARY AND CONCLUSIONS 1

1

   Communication from the Commission to the Council and the European Parliament "EU Economy: 2004 Review", COM(2004) 723 i, adopted on 26 October 2004.

• 1. 

  MACROECONOMIC DEVELOPMENTS IN

  THE EURO AREA

Summary

Following three years of weak growth, the euro area economy is now recovering. After turning the corner in the summer of 2003, the economy strengthened during the first half of 2004, growing close to potential. Growth of a similar magnitude is expected to continue during the remainder of this year and into 2005. While exports have led the way, an increase in domestic demand will be crucial to sustain and increase the momentum of the recovery.

There is little evidence to support the argument that the weak domestic demand in recent years can be attributed to macroeconomic policies. The monetary conditions remained accommodative as was reflected in the historically low short and long-term interest rates and fiscal policy was marked by the cushioning impact of the automatic stabilisers. Thereby the policy-mix provided a sound macro-economic environment conducive to a resumption of domestic demand.

Belated and sluggish recovery puts the spotlight on the area’s economic resilience. From a global perspective, the rebound of the EU economy took place belatedly and sluggishly. In comparison to other industrial countries, it takes more time in the euro area before potential GDP is reached. This has prompted queries into the euro area’s economic resilience and more specifically into whether the euro area is more sensitive to adverse economic shocks or whether its economic structures are less favourable to economic resurgence.

Activity is little affected by adverse economic news. In this context, the question arises as to whether events that hit the news headlines, such as the euro’s appreciation and mounting oil prices, may have dipped economic activity in the euro area. The analysis conducted on this point shows that though they have impacted on economic confidence indicators, their effect on industrial production was short-lived and not particularly deep. This finding is in line with comparable events since the start of EMU. Economic shocks have been harmful if they unveiled underlying economic imbalances. In this context, the strong growth in world trade in the late 1990s seems to have spurred the euro area’s reliance on external demand. This dependency has not been worked out during the growth slowdown and implies a risk for the economic rebound, particularly in view of the high oil price, which may reduce the strength of world trade and therewith the demand for euro-area exports.

But structural rigidities may be at the origin of sluggish adjustment. Rather, it would appear that the existence of structural rigidities can help to explain the late and sluggish cyclical adjustment in the euro area. Model simulations suggest that structural rigidities mainly impact on investment activity. This implies that the pace of the return to potential output will be crucially determined by how much rigidities weigh on investment once the cyclical impact of a lack of demand and financial constraints that held back investment growth is worn out. Wage rigidities and imperfect competition, in particular in the service economy, feature high when it comes to understand sticky prices in the euro area. Relative prices do not adjust quickly in the euro area, implying the need for more costly quantity adjustments in the form of scaling down production and employment. It can also not be excluded that the lack of resolve in addressing budgetary consolidation, structural reforms and pension reforms have raised uncertainty and thereby adversely affected consumer confidence and spending. Forceful resolution of these outstanding policy issues is liable to support consumption during the upswing. While rigidities help to cushion the effects of economic shocks, they do so at the expense of slowing the recovery. For example, while employment has been resilient in the slowdown, to the extent that it was due to labour hoarding, the corollary is that job creation may turn out more muted in the current upswing.

Countries with the strongest growth in the euro area benefited from strong domestic demand whereas growth in the laggards was mainly driven by external demand. Those countries which performed well in previous upturns have not grown strongly in the current upswing. Performance in the latest upswing appears to be more dependent upon performance during the previous trough, with some countries – notably the Netherlands and Portugal still experiencing restricted growth as a consequence of previous overheating. Inflation differences are broadly in line with growth differences. Low inflation countries benefited from an improvement in price competitiveness. They recorded a stimulus from external demand but remained at the lower end of the growth spectrum. The effect of inflation on real interest rates seems to have over-compensated the improvement in price competitiveness. In high growth countries higher inflation reduced real interest rates, thereby boosted domestic demand and amplified the differences. The limited role for the first relative to the second effect in reducing growth and inflation differences during the last year reflects structural rigidities that reduce the information content of relative prices as well as the adjustment of demand and output to price signals.

TABLE OF CONTENTS

• 1. 

  I NTRODUCTION ..............................................................................................................................................11

• 2. 

  R ECENT ECONOMIC DEVELOPMENTS ..............................................................................................................11

2.1 Growth pick-up since autumn 2003 ...................................................................................................................... 11

2.2 The anatomy of the early rebound ........................................................................................................................ 13

• 3. 

  T HE IMPACT OF ADVERSE ECONOMIC SHOCKS ...............................................................................................18

3.1 The impact of adverse economic news on confidence and economic activity: The early years of EMU .............. 18

3.2 Shocks and economic imbalances ......................................................................................................................... 20

3.4 Conclusions .......................................................................................................................................................... 23

• 4. 

  F ACTORS SHAPING THE RETURN TO POTENTIAL GROWTH ..............................................................................23

4.1 The impact of rigidities on economic recovery: results from model simulations ................................................. 23

4.2 The recovery’s employment-content .................................................................................................................... 25

4.3 Labour productivity growth, investment and potential output growth .................................................................. 28

4.4 Determinants of households’ savings behaviour ................................................................................................... 31

4.5 Factors behind prices stickiness ............................................................................................................................ 40

4.6 Conclusions .......................................................................................................................................................... 45

• 5. 

  E CONOMIC DIFFERENCES ...............................................................................................................................45

5.1 Measuring differences in timing and strength of the last upturn ........................................................................... 45

5.2 Can country-specific factors explain growth differences? .................................................................................... 47

5.3 The working of adjustment mechanisms .............................................................................................................. 48

5.4 Conclusions .......................................................................................................................................................... 51

R EFERENCES ............................................................................................................................................................52

MACROECONOMIC DEVELOPMENTS IN

THE EURO AREA

• 1. 

  Introduction adverse shocks. But if shocks require structural

  adjustment, they reduce the pace of recovery.

Following three years of weak economic growth, the Section 2 presents a brief review of economic euro-area economy is now recovering. After turning the developments in the current year, main developments in corner in the summer of 2003, the economy macroeconomic policy and presents the evidence on strengthened during the first half of 2004. Growth of a why the rebound in the euro area has been perceived as similar magnitude is expected to continue during the being sluggish and overdue. The exposure of the euro remainder of this year and into 2005. In 2004, economic area to adverse economic shocks is analysed in perspectives brightened against the background of a Section 3. Section 4 deals with the forces that shape favourable international economic environment while recovery. It starts by showing the impact of rigidities on domestic demand in the euro area gained pace slowly economic recovery by means of macroeconomic but steadily. The contribution to growth of private simulations and then deals with explanations for four consumption and investment, which are the two main structural puzzles that became apparent during the last components of domestic demand, however, remained slowdown. Section 5 elaborates on intra-area differences too unsteady to speak of an already secured recovery of among Member States, trying to identify countrydomestic demand. An improvement on both counts specific factors and the effectiveness of adjustment would be necessary to keep higher growth on a mechanisms behind differences in the pace of recovery.

sustainable footing.

In comparison with other industrial countries, the 2. Recent economic developments

economic recovery in the euro area was sluggish and

late starting. This chapter elaborates on possible reasons 2.1 Growth pick-up since autumn 2003

for the lack of dynamism with the focus being on two

main avenues. First, the euro-area economy is often Summer 2003 witnessed a turnaround in economic considered to lack resilience in the face of adverse activity in the euro area. After no growth in the first half economic shocks, in the sense that unforeseen events of 2003, real GDP growth picked up in the second half have a severe impact on economic activity, pushing the of the year and firmed in the first half of 2004. The pace of economic growth below potential output growth acceleration in the annual rate of growth from 0.6 per and delaying economic recovery. The second avenue cent in 2003 to 2.1 per cent in 2004, however, looks at the role of structural rigidities. Rigidities could overestimates the underlying dynamics because of delay the rebound, leaving growth below potential for a statistical and calendar effects.

2 While the period of the

longer time than in flexible economic areas. growth slowdown has been overcome, entry into a highgrowth setting is still pending.

Both strands of explanations relate to the concept of economic resilience, which has two different connotations. It can be understood as the short-fall in

growth caused by adverse economic shocks. A second 2

application of resilience denotes the speed of recovery About 0.3 percentage point of the acceleration is due to a from past shocks. Apparently, the role of structural base effect, which arrives mechanically from the

rigidities differs strongly in both perspectives. Rigidites acceleration of growth in the course of 2003. A further

could delay and cushion the impact of temporary estimated 0.3 percentage point is added by a calendar effect because there were more working days in 2004 than in 2003

(leap year, more holidays on week-ends).

Graph 1: GDP growth in the euro area Graph 2: Output gap estimates, euro area

4.5 2

4 1.5

European Commission

3.5 Annual growth 1

IMF 3 quarterly growth

D P 0.5 OECD . p.

• a. 

  2.5 ia l G 0

 c hg 2 te

nt

% -0.5

1.5 o f po

% -1

1 -1.5

0.5 -2

0 -2.5

1999 2000 2001 2002 2003 2004 2005 2006 1999 2000 2001 2002 2003 2004 2005

Note: Commission Autumn 2004 forecast for years 2004-2006. Note: OECD data is the spring forecast., based on potential GDP.

Source: Commission services. Source: IMF, OECD, Commission services.

Although actual GDP developments since summer 2003 In comparison to its peers, the euro area is projected to surprised to the upside, expectations remain prudent as witness the latest and slowest recovery. The OECD regards the area’s capability to initiate a strong recovery predicted the euro-area output gap to start closing only and a quick return to potential GDP. The notion of a in 2005. Table 1 shows that it will have a considerably particular sluggish recovery in the euro area rests on the larger negative output gap than any of the other observation that growth has proven so far to be stronger countries in that year. The speed with which the output in other economic entities than in the euro area. In gap is predicted to change from 2004 to 2005 in the euro addition to the comparison of proper GDP growth rates area is also rather modest compared to the speed with in the euro area and elsewhere, this view finds some which the output gap in for example Japan, Norway and support from the development of output gaps in different the USA is calculated to close in the first year after the countries over the next years as projected by trough. It is, however, about the same pace as in Canada, international organisations. Despite the uncertainty Switzerland and the UK. surrounding the calculation of output gaps, this concept

is seen as superior to the comparison of actual growth Graph 3 suggests that the current projections of a slow rates because differences in rates of potential output closing of the output gap are in line with the experience growth are explicitly corrected for. Estimates suggest in the past. In the 1980s and 1990s, the output gap was that these differences are quite substantial between for closed only in year 5 after the trough. This was true for instance the USA and the euro area. Taking Commission the euro-area aggregate as well as for the UK. Of the

calculations, potential output growth in the USA has been close to 3 per cent on average 2001-2005 compared

to around 2 per cent for the euro area. Graph 3: Closing of output gaps in major economic areas in 1980s and 1990s

According to the European Commission autumn 2004

forecast, the euro-area output gap is minus 1.2 per cent 1

of potential GDP in 2004 and will gradually close to a 0

negative 1.0 per cent in 2006. Projections from IMF and -1

OECD also point to a rate of GDP growth in the euro area over the next year that hardly exceeds potential -2

GDP growth, meaning that the output gap would only Euro area 83-88

marginally close in 2005. While the Commission -3 Euro area 93-98

forecast is more recent than the OECD projections, -4 UK 81-86

which were published in early summer this year, the UK 92-96 -5 latter ones are used for the cross-country comparison in USA 82-85 USA 91-97

Table 1. The reason is simply that the OECD calculates -6 output gaps for more countries than the Commission Trough 1 2 3 4 5 6 does, which allows a larger panel for international Years since trough comparisons. The use of different methods in both

institutions mainly affects the level of the output gap, Source: Commission services.

yielding a very similar profile over time.

Table 1: Cyclical developments in industrial countries

Previous output Recent trough of the output gap

gap peak (2000 Expected output

unless indicated) Magnitude Time gap in 2005

Comments

Australia 2.2 (1999) 0.6 2004 0.3 Gap never negative Canada 2.3 -0.3 2004 0.0 Denmark 2.2 -1.4 2004 -0.9 Iceland 4.1 0.1 2002 1.1 Gap never negative Japan -1.0 -3.5 2002 1.5 Gap closed in 2004 New Zealand 1.3 (2002) 0.5 2001 0.0 Gap never negative Norway 2.8 (1998) -1.5 2003 0.7 Sweden 2.3 -1.3 2003 0.1 Switzerland 1.8 -2.1 2003 -0.8 United Kingdom 1.3 -0.7 2003 0.2

USA 2.4 (1999) -2.0 2002 and 2003 0.2 Euro area 1.1 -2.3 2004 -1.8

Source: OECD.

two rebounds witnessed in the USA, one was faster, the other hand, the strongly counter-cyclical

namely the 1982-85 recovery. 3 According to macroeconomic policy in the USA might have prevented

Commission estimates, the US output gap will close full adjustment to the economic imbalances that have even faster between 2002/03 to 2005 than in the early built up during the previous boom period. US savings 1980s. have been lagging behind investments in the USA for

In complement to the analysis of how the output gap is more than a decade now with the consequence of a large expected to close, namely over the longer term of 4 to 6 current account deficit.

years, it is informative to compare the speed of the

recovery in its early stage, for instance in the first two 2.2 The anatomy of the early rebound

years after its trough in the output gap. Experience in Unsteady development of demand components. The industrialised countries since the mid-1980s suggests drivers of growth were variable in the early phase of the that the speed of recovery, for instance measured by the rebound and in particular the contribution of domestic change of the output gap in the first two years after the demand was too unsteady to speak of a secured trough, is generally loosely related to the magnitude of recovery. Net exports were the main contributor to the trough. Looking at the early phases of the rebound growth in the first quarter of the rebound. Investment after the recession of the early 1990s and the Mexico and inventories fulfilled this role in the subsequent crisis of 1995/96, it appears that the output gap closed quarter and private consumption contributed strongly in faster in non-euro area OECD countries than in eurothe third quarter of the rebound. While the sequence of area Member States even if the difference in the depth of demand forces is as expected, the speed of their the output gap is controlled for. However, only few alternation is not, causing some difficulties in assessing observations are available and differences between both the robustness of the rebound. The quick spill-over of groups of countries are not statistically significant. the external impulse to domestic demand is a reassuring

In comparison with the historical precedents, it seems sign, because it reduces the recovery’s reliance on that the rapid speed of the current recovery in the USA external demand. A strengthening of private is the unusual event rather than the slower speed in the consumption had long been anticipated. A worrying euro area. To some extent, the considerable easing of development, however, is that each factor’s contribution monetary and fiscal policy in the USA is a factor behind was relatively small and is therefore unlikely to trigger its quick economic recovery. Countries with close trade, strong knock-on effects on the other demand

financial and exchange rate ties such as the UK or components.

Canada may have benefited from this, therefore

experiencing a faster rebound now than in the past. 4 On

3 The slightly longer duration of a negative output gap in the

USA in 1991-97 seen in Graph 3 is not evident in OECD figures, which indicate a closing of the gap one year earlier.

4 Countries that aim at stabilising their exchange rate towards

the US dollar not only benefit from an indirect depreciation downward pressure on interest rates initiated by the low US relative to for instance the euro, but also from the interest rate.

Graph 4: Contribution of external demand to growth in the Graph 5: Contribution of domestic demand to growth in the early phase of cyclical rebounds, euro area early phase of cyclical rebounds, euro area

2003Q3 2002Q1 1996Q2 1993Q2 2003Q3 2002Q1 1996Q2 1993Q2

1.2

0.6 1

0.4 0.8

0.2 q 0.6

q 0 s

qo 0.4

pt

 p t qo %

0.2

% -0.2

0 -0.4

-0.2

-0.6 -0.4

-0.8 1 2 3 4

Quarters since trough Quarters since trough

Source: Commission services. Source: Commission services.

Graph 4 reveals an atypical V-shape in the contribution Employment and prices still reflect signs of economic of net exports to growth in the current rebound. The weakness. Employment growth came to a standstill from contribution of net exports was smoother and positive the second half of 2002 onwards as a lagged throughout the early phase in previous rebounds with the consequence of the prolonged economic slowdown notable exception of the aborted recovery of 2002. Since 2001-2003. Though the previous vigour in job creation the rebound in 2003 took place against a brightening of was lost, the overall development is in sharp contrast the global economic outlook, with continuously high or with past experience. For example, although the accelerating economic growth in most industrial magnitude of the output gap was similar in 1993 and countries, including Japan, the negative contribution 2004, employment growth in the early 1990s was from external demand in the final quarter of 2003 is markedly negative in the year before, at and after the surprising. It may have been caused by changes in price trough of the output gap (see Graph 6). The rate of competitiveness due to the appreciation of the euro unemployment also increased more modestly in the witnessed at that time. recent slowdown, i.e. from 8 per cent in 2001 to 8.9 per

A relatively strong contribution from domestic demand cent in 2004 compared to an increase from 7.8 to 10.8 to growth in the euro area is the mirror image of the per cent 10 years before. Section 4.2 takes a closer look early weakening of the contribution from external at the factors behind the resilience in labour market demand (see Graph 5). The weakening of domestic performance during the slowdown and the potential demand after a first initial peak is a pattern that seems to implications of past resilience for the pace of recovery.

be common in recoveries in the euro area. The most

reasonable explanation points to the impact of pent-up Graph 6: Employment growth before and after the trough of economic activity, euro area

demand as a temporary driving force at the early stage of recovery. A further strengthening of domestic 2.5

demand, which would be crucial for both the vigour and 2

sustainability of the recovery, is not yet visible in GDP 1.5

data.

1 a.

. p. 0.5

hg

 c 0

% -0.5

-1 2004 = Year 0

-1.5 1993 = Year 0

-2

-3 -2 -1 0 1 2 3 Years since trough

Note: The year with the largest negative output gap is year 0 in the graph. Commission autumn 2004 forecast for 2004-06. Source: Commission services.

Graph 7: Consumer price inflation in the euro area Graph 8: Monetary conditions index, euro area

3.5 -5 Contrib. of real exchange rate

3 -4 Contrib. of real interest rate

Loosening MCI

2.5 -3

.a . 2 p -2 c hg

% 1.5 -1

1 HICP

Core 0

0.5 1

0 Tightening

2 Jan 00 Jan 01 Jan 02 Jan 03 Jan 04

Jan-99 Sep-99 May-00 Jan-01 Sep-01 May-02 Jan-03 Sep-03 May-04

Note: Core inflation is HICP excluding energy and unprocessed food.

Source: Commission services. Source: Commission services.

Inflation remained close to, but above, the 2 per cent cyclically-adjusted primary balance, which is considered margin. This is higher than many expected given the the best available tool to measure the stance of fiscal pronounced period of weak growth and the downward policy, was marginal in both 2003 and 2004 for the pressure exerted by the appreciation of the euro. euro-area aggregate. Although the direction changed Increasing oil prices strongly determined the pattern of from 2003 to 2004, it, nevertheless, indicates a neutral headline inflation in the course of the year, though they policy stance.

caused little concern as regards potential second-round Steadiness in key policy parameters does not indicate effects, probably because the impact of the previous oil policy inactivity. Discussions on fiscal policy were price hike in 2000 remained limited to energy prices and vivid. This was in particular so in Autumn 2003 when petered out smoothly. Furthermore, increases in tensions rose regarding the application of the EU fiscal regulated prices and indirect taxes in some countries framework as the Council decided not to adopt a shifted the rate of consumer price inflation upward. Commission recommendation to France and Germany Wage growth seems to have eased in 2004 and on new measures to reduce the budget deficit and especially against the background of the expected remedy the situation of an excessive deficit. 5 The ECB cyclical pick-up in labour productivity should not weigh was faced with demands to cut interest rates in early on inflation prospects in the short-term. Inflation is 2004 when the euro exchange rate climbed to close to expected to come down to below 2 per cent in the course 1.30 USD/EUR and many foreign exchange market of 2005 only. observers feared a continuous firming. Policy easing did

In a strict sense, price stability as defined by the ECB as not take place. an annual rate of consumer price inflation in the euro The monetary stance is assessed as supportive to area of below 2 per cent did therefore not prevail. In the economic activity in 2004. Monetary indicators like M3 absence of strong apparent upside risks to the inflation growth suggest the availability of sufficient liquidity in outlook throughout the slow growth period, the central the financial system and nominal interest rates are on a question is why the apparent downward pressures had low level, in absolute terms and relative to the Taylor no stronger impact. The weakening of demand and the rate (see Graph 9). Real short-term and long-term decline in import price inflation due to the euro interest rates remained close to their historic lows. With appreciation were expected to push consumer price the real interest rate remaining relatively stable for more inflation below 2 per cent. This did not materialise, than a year now, movements of monetary conditions as questioning the role of lower inflation in helping measured by the MCI were only caused by changes of cyclical adjustment. The analysis of price stickiness in the euro exchange rate. Since the euro exchange rate has Section 4.5 sheds some further light on this issue. only a small weight in the MCI, reflecting the relative

Macroeconomic policy stance remained accommodative. importance of real interest rates and real effective No tangible adjustment of key macroeconomic policy exchange rates for economic activity, the euro parameters took place between summer 2003 and appreciation caused only a minor tightening of the MCI, summer 2004. This is most visible for monetary policy which has become in consequence more favourable to since the ECB’s last interest rate cut dates from 6 June domestic relative to external demand.

2003, when policy rates were reduced by 0.5 percentage point, yielding a rate of 2.0 per cent of the minimum bid

rate in the main refinancing operation. The change in the 5 For a detailed account of the discussion, see European

Commission (2004).

Graph 9: Taylor rate, euro area Graph 10: Fiscal stance and cyclical conditions, 1999-2005

6 1

Taylor rate corridor

5 B ) Pro-cyclical fiscal

Counter-cyclical

A P C 0.5 tightening fiscal tightening

 in 4 es

ng 1999

ha 2003 2005 0

% 3 e (C ta nc

2 l s 2004 2000

Actual 3 m onths interes t rate -0.5

2002 2001

F is

ca

Counter-cyclical

1 Pro-cyclical fiscal fiscal loosening

loosening -1

0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Cyclical conditions (Output gap)

Source: Commission services. Source: Commission services.

While the ECB kept policy rates on hold for more than a current stage of the economic recovery. In contrast, the year, forward-looking financial market prices reflected gap was considered to be much wider in the USA, signs of speculation on possible changes in the monetary requiring the US central bank to remove some policy stance at several occasions. One incident was the stimulus at an early stage. expectation of a further interest rate cut in early 2004

motivated by the feared negative consequences the Looking at the developments of the fiscal stance in the appreciation of the euro could have on the economic euro area, Graph 10 plots the change in the cyclicallyrecovery and possible negative effects on economic adjusted primary balance (CAPB) against the output growth caused by geopolitical uncertainties. The first gap. The SGP favours a neutral budgetary stance for speculation that the ECB might tighten the monetary countries which have achieved a medium-term position stance emerged when rising oil prices lifted consumer of ‘close-to-balance or in surplus’, which would be price inflation in early summer and economic indicators visible in a movement along the horizontal axis. signalled higher than expected growth in the euro area. Countries that have not yet achieved the medium-term Expectations among financial market participants of an target of the SGP should aim for a positive change in the increase in euro-area interest rates were muted in CAPB, indicating a restrictive fiscal stance. Small summer. The rise in oil prices was interpreted as being changes in the CAPB such as those seen in 2003, 2004 more disinflationary than inflationary, given its taxing and projected for 2005 are considered as indicating a effects on growth. broadly neutral fiscal stance.

6

The major story behind market perceptions of future Table 2 summarises some of the euro area’s key interest rate decisions relates to the question of when macroeconomic policy parameters, illustrating that longand how quickly central banks have to withdraw their term interest rates have remained on a low level. The accommodative stance. The concept of a neutral interest widening budget deficit caused public debt to exceed rate, which should prevail in neutral cyclical conditions, 71 per cent of GDP, which is far above the 60 per cent while theoretically appealing, is difficult to implement Maastricht criterion and too high in view of the fiscal in practice. Calculations on the basis of the Taylor rule burden of ageing populations. Stepping up the pace of are often used as a proxy, indicating that euro-area budgetary consolidation would be adequate if economic interest rates were low relative to the Taylor rule. Muted activity recovers as expected.

interest rate expectations at a time when the US central bank started tightening its monetary policy stance could be interpreted as market participants’ assessment that the gap between actual and neutral interest rates in the euro area would not represent a risk to price stability at the

6

   See 5.3 for a comparison of Member States fiscal stances and their relationship with the output gap.

Table 2: Key macroeconomic policy parameters, euro area

2003 2004 2005 2006

Short-term interest rate (3 months) 2.3 2.1 n.a. n.a.

Long-term interest rate (10 years) 4.1 4.26 n.a. n.a.

Real effective exchange rate (GDP deflator, 1999 = 100) 104.0 106.8 n.a. n.a.

Budget deficit, % of GDP -2.7 -2.9 -2.5 -2.5

Cyclically-adjusted deficit, % of GDP -2.2 -2.5 -2.1 -2.2

Cyclically-adjusted primary deficit, % of GDP 1.3 0.9 1.2 1.1

Public debt, % of GDP 70.7 71.1 71.1 70.9

Note: 2004 figures denote the first semester for interest and exchange rates, REER vs 34 industrial countries, 2004 to 2006 budgetary data based on Commission Autumn 2004 forecast.

Source: Commission services.

Box 1: The reason for slow cyclical recoveries – A look at the recent academic literature

The academic literature is generally relatively silent on the question of why economic recoveries materialise at different speeds. In the past, this kind of discussion focused on differences in the macroeconomic policy stance. Some recent contributions have highlighted the importance of supply side issues.

Hausmann et al. (2004) analyse the determinants of growth accelerations, which they define as an increase in per capita GDP growth of at least 2 percentage points. They also require the post-acceleration growth rate to be at least 3.5 per cent per year sustained over the next 8 years. Based on these criteria, they identified around 80 episodes since the 1950s. While most of the observations stem from former developing countries and the number of episodes declines from the 1970s onwards, EU countries also witnessed growth accelerations in the past two decades (United Kingdom 1982, Spain 1984, Ireland 1985, Portugal 1985, and Finland 1992). The comparison of macroeconomic trends in these episodes suggests that increases in investment, trade and real exchange rate depreciations seem to be correlated with growth accelerations. The detailed empirical analysis of these episodes, however, revealed that most of them were not related to expected exogenous determinants. So the authors find financial liberalisation and positive external shocks to have a temporary impact. The estimation also suggests that despite the finding that the variable capturing economic reform is rarely significant, economic reform has a statistically meaningful impact, at least on the probability of experiencing a sustained acceleration in growth.

Applying the concepts from growth theory on the analysis of the business cycle, Prescott (2002) attributes the most important role for explaining differences in economic prosperity to productivity. Crucial policy determinants are trade integration, an efficient financial system and competition among firms. As regards labour utilisation, he considers the tax structure to have an important distorting impact, which could account for the difference in labour supply in the USA and continental Europe. Bergoeing et al. (2004) present evidence in a cross-country setting that regulation is negatively related to the economy’s ability to recover from shocks. They argue that economic recovery requires the re-allocation of resources, i.e. from declining to growing industries, and regulation makes this adjustment process costly.

Some evidence on the impact of market rigidities on economic performance has been derived from incorporating rigidities into macroeconomic models and comparing the performance with a more flexible model environment. For example, using the IMF- GEM model Bayoumi et al. (2004) show that differences in parameters that capture the impact of competition can account for half of the difference in per capita GDP between the USA and the euro area. Reforms that raise the level of competition in the euro area to the US level would lead, in the model, to an increase of GDP by more than 12 per cent in the euro area. They would also improve price flexibility and therewith allow for a more active role for monetary policy in stabilising output. Following a similar approach, Drew et al. (2004) compare model outcomes if rigidities on different markets are introduced. Structural rigidities on product and labour markets reduce the impact of negative demand shocks. However, output and unemployment need longer to return to their equilibrium levels in the simulations shown, by about 2 to 3 years.

Galí et al. (2003) approach the slack in an economy by constructing an inefficiency gap, which is the inverted sum of the aggregate price mark-up and the aggregate wage mark-up in an economy. This indicator is highly pro-cyclical in the euro-area, driven by counter-cyclical behaviour of the wage mark up, defined as the difference between the real wage and the disutility of work. This could be interpreted as a desired adjustment among firms and workers in an environment with flexible prices and wages. However, the authors’ alternative and preferred interpretation is that the change in the mark-up is driven by wage rigidities.

Some economists in investment banks conjectured that the slack in the euro-area economy could be smaller than estimated because the rate of potential growth could have become lower over time. JP Morgan (2004) argues that some of the peculiarities observed in the recent slowdown, namely the slow pass-through of the euro appreciation to prices, the limited decline in capacity utilisation and the small increase in unemployment, would be less odd, if the rate potential growth and therewith the (negative) output gap were smaller than thought. A reason could be that economic performance in the 1990s was spurred by some favourable supply shocks, i.e. low energy and commodity prices, trade and financial liberalisation, productivity in ICT. According to Goldman Sachs (2004), these shocks influence the cycle in which they occur, but not necessarily the next one.

• 3. 

  The impact of adverse economic shocks impact on economic confidence in addition to any

  materially justified impact.

This and the subsequent section cast some light on the A later step relates the results of the event study to those possible reasons for the sluggish recovery in the euro economic shocks that used to be considered of particular area with the focus on two, not mutually exclusive relevance for explaining the euro area’s dismal growth strands of explanations. The first set of explanations is performance in 2001-03. Somewhat pre-empting the built on the perception of the euro area being conclusions, adverse economic events have been particularly vulnerable to adverse economic shocks. The frequent in the euro area. However, their effect via pronounced weakness in economic growth could be due deterioration in economic confidence is transitory unless to the impact of the shocks that hit the economy in 2001- they unveil some underlying economic imbalances.

• 03. 

  This section asks whether the euro area is particular

sensitive to adverse economic shocks and if so why. In Since 1998 the euro area has been hit every year on this context, the effects of adverse economic news on average by two adverse economic events, which are households’ and investors’ sentiment are often quoted as defined here as unexpected events that are presumed to being of particular importance. A second line of be potentially harmful to economic activity. Admittedly, reasoning focuses on the existence of structural rigidities any method of identifying the occurrence and timing of in the euro-area economy that could delay recovery and these events encompasses a discretionary element and a inhibit the return to potential output growth. This second degree of arbitrariness. The preferred strategy in this approach is taken up in Section 4. exercise consisted in matching all major changes in the

Commission’s industrial confidence indicator to a

The occurrence of adverse economic shocks is widely chronology of market events. 7

mentioned as a reason for the brisk interruption of the

strong pace of economic growth that prevailed in the late Graph 11: Changes in industrial confidence

1990s up to 2000. The shocks that were highlighted in previous issues of the Review concerned (i) the oil price

hike in 2000, (ii) the bursting of the ICT investment 4

bubble and (iii) the slump in world trade. They all have 3

in common that they acted on a global scale, i.e. dented nt

s

 p oi 2

growth not only in the euro area but also in many other th in 1

economic areas. The observation that growth picked up on

later in the euro area than elsewhere suggests that the s

m 0

ou

euro area was particularly vulnerable to these shocks. re vi -1 er p

However, this perspective could give a biased view on -2 ov e

the resilience of the euro area. The reason is that it ng -3

highlights the shocks that had a severe impact but leaves C ha -4

out all shocks from which the impact did not materialise. -5

For instance, three special incidents in the current year Feb 98 Feb 99 Feb 00 Feb 01 Feb 02 Feb 03 Feb 04 were widely thought to present a risk to economic

recovery. These are the appreciation of the euro Source: Commission services.

exchange rate, the hike in commodity and especially oil prices and the terrorist attack in Madrid. Though they

impacted on economic confidence indicators, their effect It turned out that each fall in the industrial confidence

on the recovery has been muted. indicator of 2 or more points was accompanied by a major news event. Table 3 provides a list of events and

3.1 The impact of adverse economic news on Graph 11 marks the timing of changes in the industrial

confidence and economic activity: The early confidence indicator. There were, however, a number of events that would a priori be judged as potentially

years of EMU severe, and that found no reflection in the change of the

This section looks at the adverse economic shocks in the form of an event study, i.e. it identifies in a first step unexpected events that were considered to be potentially harmful for economic activity and subsequently tries to 7

establish their relationship with economic sentiment and An alternative strategy used consisted in deriving the timing

short-term hard economic indicators. The intention is to of events from an autoregression of production expectations in industry. An event was defined as any observation when

get insights into the kind of shocks the euro area had the residual of this estimate was unusually low. Both

been exposed and the severity of their impact. methods yield a high degree of, but no perfect, overlap. Moreover, the research design permits to establish some Moreover, while the alternative approach is more

preliminary evidence of the importance of confidence sophisticated from a quantitative point of view, some of the effects, i.e. the significance of economic shocks via their observations are difficult to reconcile with “economic

news”. As a consistency test, the same exercise was repeated for the Reuter’s PMI, yielding comparable results.

Table 3: Adverse economic shocks

Change in Negative growth Time industrial in industrial

confidence production, mom

Russia, LTMC August to October 1998 -2/-2/-4 October to December

NATO air strikes on Serbia March 1999 0

Oil price exceeded 30 USD March 2000 2

Stock market collapse April 2000 -3 June

Downward revision of already slow GDP growth in

USA December 2000 -2 January

Turkish financial crisis, foot and mouth disease February 2001 -3 March and April

Indication that the global economy was in recession April/Mai 2001 -2/-2

Enron scandal August 2001 -2

Terrorist strikes in USA September/October 2001 -4/-2 September to

Afghanistan conflict, GDP release confirms December Germany to be in recession November 2001 -2

Enron collapse, financial crisis in Argentina December 2001 3

World com scandal July 2002 -2 July

Floods in D, E, F, A August 2002 3

Iraq conflict March 2003 -2 March and May

Strong increase in US bond rates July 2003 -2 August and September

Euro exceeded 1.20 December 2003 -2 January

Terrorist strike in Madrid March 2004 0

Oil price exceeded 40 USD May 2004 0

Shocks are related to 16 out of 28 observations for negative monthly investment growth

Source: Commission services, BBC, Macro-Dev.

industrial confidence indicator. Notably the military consistent with the consensus among economists that the conflict in Serbia in spring 1999, the increase of the oil terrorist strike was not the cause of economic slowdown.

price above 30 USD in February 2000 and the floods in

some euro-area Member States in summer 2002 were Interestingly, the direction of changes in industrial

not followed by a drop in industrial confidence. confidence used to be clustered, but this pattern has dissipated in the recent past. For instance, increases

Over the period 1998 to 2003, the euro area was exposed were prevalent in 1999 and declines in 1998 and from to 12 adverse economic events, measured as a decline in autumn 2000 to autumn 2001. A deviation from this the industrial confidence of at least 2 points. The pattern is visible for the time since summer 2002, when average of 2 shocks per year also continued in 2004, periods of plusses and minuses changed in small with both the terrorist strike in Madrid on 11 March and intervals. All the negative changes that were equal to or the increase in oil prices qualifying as economically exceeded 2 points in this time can be traced back to meaningful events. The financial crises in South-East worrying economic news. This could suggest that either Asia and Russia in 1998 as well as 11 September 2001 economic shocks have become more frequent or that initiated the most marked drop in industrial confidence. agents have become more sensitive to adverse economic With hindsight, they can be considered the most severe news, i.e. responding more strongly in their assessment

8

economic shocks in the panel. It is, however, in surveys. remarkable that industrial confidence had already

deteriorated before 11 September 2001, which is The right column in Table 3 indicates that industrial production growth became negative whenever industrial

confidence fell by 2 or more points either in the same or

8 Note that a fall in the Commission’s industrial confidence the subsequent month. Of the 26 observations of indicator of similar magnitude was only recorded once since negative monthly industrial growth that were recorded

the start of the series in 1985. This was in October 1992,

following the ERM crisis (black Wednesday) that led Italy between 1998 and 2004, 16 occurred at or shortly after

and the UK to leave the ERM. an economic shock. This result is not a major surprise

since the confidence indicator is calculated from The timing of deteriorations in consumer confidence is answers to, amongst others, questions regarding actual only weakly correlated to downward changes in and expected industrial production. Nevertheless, it industrial confidence. The same method of relating large should be noted that econometric estimates suggest that monthly changes in the consumer confidence indicator these economic events have some predictive power for with external events does not reveal a consistent pattern. industrial production. Only three out of five observations occurred in times

Although significant, the impact of these adverse when the change in industrial confidence indicated an economic events on industrial production seems to be adverse economic event. It is, however, remarkable that rather short-lived. This can be shown by including them the decline in consumer confidence is often spread over into an empirical estimate of monthly industrial 2 or even more months. The duration of the deterioration production growth. Dummy variables signifying the in consumer confidence as well as its occurance after timing and severity of bad economic news turn out some of the major economic shocks suggests that significant in auto-regressions of industrial production, consumer confidence depends more on the state of the but only with a lag of 1 and 2 months. Higher-order lags general economic situation than on events that could of the shock variable are not significant and the trigger any deterioration in the economic outlook. Each estimated coefficients of lagged investment growth observation of a severe deterioration in consumer imply that shocks fade out quite quickly. According to confidence coincided with weak growth in private the empirical estimates, industrial production falls for consumption in the national accounts.

two months when a shock that causes a decline in

industrial confidence by 2 or more points occurs. 3.2 Shocks and economic imbalances

Afterwards, growth in industrial production oscillates There is a discrepancy between the adverse economic around zero, and although industrial production will events referred to above and the economic shocks that remain on a lower level, Graph 12 suggests that the are held responsible for the dismal economic impact of the shock is practically invisible in growth performance in the euro area. Moreover, the short term rates after 6 months. impact of adverse economic events on industrial

production, in combination with the imperfect Graph 12: The impact of shocks on industrial production,

growth rate correlation between the subsequent impact on consumer confidence and households’ spending suggests shocks to

% mom be relatively short-lived. The impact of the oil price hike

0.3 of 2000, the bursting of the ICT bubble and the slump in

0.2 world trade are, however, perceived to have had a lasting impact on economic activity, 9 with all three

0.1 shocks having a long-lasting impact on demand. Even if

0 the economic disturbances can be related to adverse economic events, 10 the duration of their impact seems to

-0.1 be a major difference between both concepts.

-0.2 bc := value (1) It is very likely that the period of strong growth in the

-0.3 bc := dummy if shock (2) euro area was not cut off by the occurrence of economic

bc := value if shock (3)

-0.4 shocks per se, but by underlying economic imbalances that surfaced at the time of, or because of, these shocks.

-0.5

0 1 2 3 4 5 Months 6 7 8 9 10 11 12

Note: (1) includes all values of the change in industrial confidence,

irrespective of sign and magnitude. (2) includes a dummy equal to 1 9 For an attempt to identify the economic shocks of 2000 by whenever the industrial confidence declined by 2 or more points. (3) is means of VAR analysis, see Peersman (2003). It finds that

similar to (2) but also reflects the severity of the decline in the industrial

confidence indicator. the shocks were more pronounced in the USA than the euro

Source: Commission services. area but cautions that the results are sensitive to the empirical strategy.

10 For example, the stock market collapse in 2000, when

In terms of magnitude, a decline in industrial confidence NASDAQ share prices fell by 20 per cent from March to by 2 points reduces industrial production by 0.4 to 0.8 April, heralded the end of the ICT bubble. Investment

percentage points over two months, depending on the growth became negative shortly thereafter, namely in the third quarter in the USA and in the fourth quarter in the

specification of the empirical estimate. Accounting for a euro area, marking the beginning of a period of weak share of industrial production of 20 per cent in GDP, this investment in both economic entities. The severity of this

would translate into an annual income loss of less than disturbance was probably unknown until the release of US 0.05 per cent. Unless activity in other economic sectors national accounts data in late autumn 2000. Similarly, the

is also affected by the shock, this would be too small to decline in industrial confidence in December 2000 might

expect a significant deterioration in aggregate demand. not just have been triggered by evidence of weakening economic activity in the USA, it also portended the fall in

world trade growth that materialised in 2001.

Graph 13: Financing and investment of non-financial Graph 14: Industrial production, euro area (1996Q1=100) corporations, euro area

20 160

15 150

10 Manufacturing

5 140

D P 0 Office machinery and

 o f

G 00 130 computers

% -5 1 =

96

-10 19 120

-15 110

-20

1995 1996 1997 1998 1999 2000 2001 2002 100

Financing through shares Financing through other sources 90 Net acquisition of financial assets

Net acquisition of non-financial assets 96Q1 97Q1 98Q1 99Q1 00Q1 01Q1 02Q1 03Q1 04Q1

Source: ECB, Commission services. Source: Commission services.

With hindsight, all three economic disturbances that had production in ICT hardware showed a typical boom-bust been identified as having triggered the economic pattern. Since production in ICT goods powerfully slowdown in the euro area can be considered a outpaced that in manufacturing in particular in 1999 and correction of past imbalances. Technical progress in the 2000, partly to modernise ICT in view of the year 2000 ICT sector, globalisation of markets and low commodity problem, it is reasonable to assume that overly high

prices in times of buoyant growth, might have led to capacity had been built up in this sector. distorted market signals and an undue expansion of

production structures in specific sectors. Adverse economic events in 2004 such as the appreciation of the euro exchange rate, the hike in

The case is probably most obvious for investment. commodity and especially oil prices and the terrorist Declining capital costs, in particular through interest rate attack in Madrid do not necessarily unveiled economic

convergence in some Member States in the run-up to imbalances in the euro area. EMU, and buoyant stock prices had eased financing

conditions for investment. Capital raised on stock This assertion is probably doubtful for the euro markets increased from just 1 per cent of GDP in 1997 appreciation, which brought the strong dependence of to almost 5 per cent in 2000. Non-financial corporations growth in the euro area to external demand to the fore. accumulated liabilities and holdings of financial assets In 2003/04, the effect of the appreciation on economic

from 1995 to 2000 (see Graph 13). activity in the euro area was cushioned by the simultaneous acceleration in global trade growth. 12

This trend of rising financial activity, however, stopped However, the continuous reliance of growth in the euro after 2000. A high level of corporate debt has been area on external demand suggests that this structural identified as a reason for slow investment growth in the weakness has not been worked out of the economy downturn. Financial institutions seem to have become during the growth slowdown. In the euro-area prudent in their lending policy, being on the one hand manufacturing sector, the share of sales on non-domestic eager to clean their balance sheets from bad loans while markets relative to those on domestic markets had trying to keep market shares in an environment of increased rapidly (see Graph 15). Cross-border sales of intense competition due to increasing financial capital goods rose even more strongly, suggesting that

11

integration. See Section 4.3 for more details. production structures had become increasingly focused

While it is difficult to identify an exceptional increase in overall physical investment in the late 1990s, which could indicate over-investment in the euro area, investment in some items had been buoyant. In 12

particular spending in ICT had been upbeat and roaring A further reason as to why the euro appreciation only mildly affected growth in the euro area is very likely that it

stock market prices have helped to expand activity in corrected the previous under-valuation of the euro. This

this sector. Although not all ICT goods produced and view finds support in the observation that the most

imported were also invested, Graph 14 shows that frequently mentined reasons for the preceding euro weakness were positive shocks to productivity and risk premiums in the USA. These factors can justify a temporary, but not a permanent, appreciation of the US dollar against the euro. Model analysis predicts that the

11 These factors were analysed in the chapter on the euro-area exchange rate should return to its previous level over time.

economy in last year’s R EVIEW . See, for instance, Tille et al. (2001), IMF (2004b).

Graph 15: Turnover in manufacturing, euro area (1996=100) Graph 16: Oil prices and world trade growth

170 10 20

Domestic market

160

Non-domestic market 15 15

150

Capital goods on non 20 10

.

 p .a 1 00 140 domestic markets

 = c

hg

96 U S D 25 5

 %

19 130 . 120 30 0 lu m

es

vo

110 35 -5

Oil price (lagged by 1 year, inverted lhs)

100 40 World trade (rhs) -10

1996 1997 1998 1999 2000 2001 2002 2003 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05

Note: Turnover on non-domestic markets includes cross-border intraarea turnover. Source: CPB Netherlands Bureau for Economic Policy Analysis, Source: Commission services. Commission services.

on global markets. If previous investment decisions The mechanisms through which an abrupt oil price were based on the expectation of sustained high growth increase impacts on economic activity and the in external demand, these expectations have been positioning of the euro area economy were described in disappointed since 2001. some detail in the EU E CONOMY 2000 R EVIEW and the

Oil price shocks are usually considered to have caused European Commission’s Q UARTERLY R EPORT ON THE the stagflations in the mid-1970s and early 1980s. E URO A REA 2/2004. The latter concluded that higher oil Although some researchers have meanwhile contested prices represent a small but significant downward risk to

this consensus view and stress the role of monetary economic activity.

policy in response to this shock as the actual trigger for • The still relatively weak cyclical position of the recession, 13 oil price hikes continue to raise a lot of euro area could make business and consumer

public concern. Simulations using macro-econometric confidence vulnerable to further increases in oil models consistently show oil price shocks to have a prices. significant impact on economic activity, with their

severity being largely dependent on the duration of the • The oil price hike could fuel inflation, particularly if

oil price hike (see Table 4). For example, simulations wage earners demand compensation for their loss in with the Commission’s Quest model show that the purchasing power. Whereas oil price increases led impact of a transitory increase of the oil price is quite to higher wage growth in the 1970s and early 1980s, short-lived, affecting GDP in the first year. The effect is second-round effects could not be observed after the

already unwinding by year 2 and essentially no longer year 2000 oil price hike.

noticeable in year 3. When the oil price increases • A significant risk is related to the response of permanently, Quest simulations show the negative effect emerging markets and world trade to the hike in oil to be marked in the second year (see Table 4). price because previous oil shocks have regularly

Table 4: The impact of a permanent increase in oil been followed by a slowdown in world trade.

prices (USD 44 per barrel) Growth in world trade may be more sensitive to higher oil prices than in the past due to the growing

2004 2005 2006 importance of emerging markets such as China.

GDP growth -0.33 -0.44 -0.16 Graph 16 relates the development of the oil price with

Consumer price inflation 0.21 0.36 0.05 growth in world trade, showing that both moved in

Source: Commission services. tandem since 1999 when oil prices are lagged by about 1 year. If temporarily lower oil prices were a major

driving factor of the acceleration in growth witnessed in emerging markets in the recent past, the oil price hike could uncover economic imbalances prevalent in these countries. Given that growth in the euro-area remains vulnerable to changes in external demand, oil price induced disruptions abroad could spill-over into the

economic outlook for the euro area. 13 See Barsky and Kilian (2001, 2004), Bernake, Gertler and

Watson (1997).

3.4 Conclusions would have been built up than output gap calculations

This section demonstrated that economic events that suggest.

make headlines do not necessarily amount to severe As identified in previous issues of the EU E CONOMY adverse shocks. R EVIEW , a number of factors were unusual in the growth

• The euro-area has frequently been hit by adverse slowdown from 2001-2003.

economic events, which could trigger a slowdown • Labour markets weathered the slowdown quite well,

in economic activity. On average, two of these with employment growth remaining positive and events occurred per year. unemployment rising only modestly. An important

• Most of the events identified had an impact on question is whether the resilience of employment to the slowdown translates into less supportive

industrial confidence and, with a delay, on employment growth during the upswing

industrial activity. The evidence presented suggests

the effect of confidence effects on industrial (Section 4.2).

production to be short-lived and not particularly • Considering the low level of interest rates,

deep. investment activity was shallow. Before the growth

• Those shocks that are held responsible for slowing slowdown, it was widely assumed that vivid

growth after 2000 heralded deeper underlying technical progress could spur an investment boom. Now, there are few indications that these

economic disturbances. The euro-area’s dependency expectations are being lived up to (Section 4.3).

on external demand has not been worked out during

the slowdown and represents an important risk for • Consumption was weaker than suggested by the

the cyclical rebound. trend in disposable income. At the current juncture, it is unclear whether the households’ saving rate should be expected to continue its increase or

• 4. 

  Factors shaping the return to potential whether a reversal to the lower level seen before the

growth growth slowdown will occur. This will crucially determine the contribution of private consumption

The hypothesis that structural factors might be an to economic growth (Section 4.4).

obstacle to accomplishing a faster return to potential • Consumer price inflation fell only marginally in

output is strongly linked to the observation that response to the shortfall in domestic demand. Price economic recovery set in earlier and more robustly in stickiness is considered to have prevented a larger the USA and some other Anglo-Saxon economies than decline in inflation, thereby depressing households’ in the euro area. These countries are perceived to suffer purchasing power and preventing a more aggressive

less from structural rigidities and their higher degree of easing of monetary policy (Section 4.5). economic flexibility has helped them to overcome the

global growth slowdown faster than the euro area. 4.1 The impact of rigidities on economic

Relative prices do not adjust quickly in the euro, recovery: results from model simulations

implying t the need for more costly quantity adjustments

in the form of scaling down production and Structural rigidities are theoretized to dampen cyclical employment. fluctuations. Shocks have a smaller but more durable

impact while recovery is slower. Unveiling the actual

This section reviews factors that can crucially inhibit the impact of structural rigidities on the timing and strength pace of the return to potential output. It starts with the of cyclical rebound is a thorny exercise because actual result of macro-econometric simulations that cyclical movements hardly follow the textbook model. demonstrate the role of structural rigidities in the Moreover, international comparisons like the one

adjustment to economic shocks (Section 4.1). 14 Then, it performed in Section 2.2 are sensitive to the control of

elaborates on special developments that were observed the influence of other variables such as the severity of during the 2001-03 growth slowdown and that may economic shocks or the stance of macroeconomic

indicate the effect of structural rigidities. A central policies.

question is whether they indicate a slowdown in

potential output growth. Such a consequence has This section draws on the insights from macro-economic become a possibility because a persistent period of simulations. The procedure is quite straightforward. A sluggish cyclical growth may lead to withdrawal from macro-economic model of the euro-area was exposed to the labour market, loss of skills of the unemployed and selected shocks and the impact of the shocks was low investment. This means that less economic slack compared to an alternative set-up of a more flexible

economy. Differences in adjustment patterns should be telling regarding the impact of rigidities on cyclical

behaviour.

14 The analysis presented in the EU E CONOMY 2002 R EVIEW

showed how economic structures can impact on the rate of growth in the medium to long term.

Graph 17: TFP shock in the Quest model, euro area Graph 18: Monetary policy shock in the Quest model: Reduction in interest rate by 1 percentage point over 1 year

1.6 0.6

1.4 0.5

1.2

1 0.4

Base case 0.8 0.3

0.6 Lower adjustment Base case

0.2 costs 0.4

Lower adjustment 0.1

0.2 costs

0 0

year 0 year 1 year 2 year 3 year 4 year 5 year 0 year 1 year 2 year 3 year 4 year 5

Source: Commission services. Source: Commission services.

The simulations draw advantage from the fact that The impact of a temporary decline in the interest rate as several mechanisms in the Commission’s Quest model a representative demand shock is considerably stronger allow for an explicit analysis of sluggish adjustment to in the flexible economy than in the base-line economy in shocks. For instance, the model’s investment function year 1, as well as throughout the return to the contains parameters for adjustment costs that reduce the equilibrium level. Even after five years, a small sensitivity of changes in capital and therewith of output difference is still noticeable. In Graph 18, it is assumed to determinants in capital costs. A parameter of price that the ECB would cut the interest rate relative to the adjustment reflects sticky adjustment of output prices to Taylor rule by 100 basis points for one year, after which input prices and changes in capacity utilisation. Taxes interest rates are set higher again to combat inflation in and bargaining power on labour markets reduce the accordance with the Taylor rule. Since the reversal of efficiency of the labour market and some households are the reduction in interest rates is already known in the liquidity-constrained, i.e. have no possibility to smooth model at the time when it occurs, the policy generates a

consumption over time. 15 temporary increase in GDP only, i.e. agents try to

For the following experiment, the original Quest model benefit from the lower costs of capital first of all in is exposed to a representative supply as well as to a year 1.

demand shock. The resulting path of GDP is then A key driver in the interest rate simulation appears to be compared with a model, which was exposed to the same the adjustment costs in the investment function, which shocks, but where the adjustment costs parameters in the make investment more responsive to changes in capital investment function are only half that of the original costs in the flexible model. It is, however, unclear at this model. A second departure is that the parameter stage whether the finding of structural rigidities capturing the wage bargaining of workers is assumed to impacting via adjustment costs of investment can be be equal to the one for the USA. directly translated into actual behaviour.

For the case of a supply shock, the most signficiant A drawback of the Quest model is that it is partly difference between a rigid and a flexible economy is calibrated and the coefficients reflecting nominal and visible in the long-term. Graph 17 shows the impact of a real rigidities are based on a collection of other permanent improvement in total factor productivity of 1 empirical studies. A new model, recently estimated by percentage point, which can be read as an autonomous Commission staff, contains both more explicit micro increase in the pace of technical progress. In the base foundations and a direct estimation of the adjustment line, GDP increases strongly in year 1 and converges costs through which the real and nominal frictions are afterwards slowly to the new long-term equilibrium. In the flexible variant of the model, the long-term GDP level is about 0.4 per cent higher, which also implies a

quicker adjustment in the first year. 16

15 For a more detailed explanation of the Quest II model, see

Röger and in’t Veld (1997).

16 The observation that most of the adjustment takes place in

the first year is due to the fact that the shock is assumed to investment behaviour accordingly and in a forward-looking be known to all agents, who adjust their spending and manner.

Graph 19: TFP shock in a New-Keynesian DSGE model of the euro area

GDP Investment Consumption 0.4

0.8 0.2

0.3 0.6 0.15

0.2 0.4 0.1

0.1 0.2 0.05

0 0 0 1 10 19 1 10 19 1 10 19

Quarters Quarters Quarters

Capital stock Employment Inflation

0.1 0.1 0

-0.0002 0.05

-0.0004

0.05 0 -0.0006

-0.0008 -0.05

-0.001

0 -0.1 -0.0012 1 10 19 1 10 19 1 10 19 Quarters Quarters Quarters

Note: DSGE model stands for Dynamic Stochastic General Equilibrium model. Black line standard euro area model, red dotted line the variant with

lower adjustment costs. For a model description see Ratto et al.. (2004). Percentage difference from base.

Source: Commission services.

modelled. 17 In this model, rigidities are derived from Both model simulations show that structural rigidities

adjustment costs for capital and investment, wages, have an impact on the adjustment to shocks. The largest prices and employment. The experiment consists in impact of rigidities is via investment and on long-term analysing the model’s impulse responses with the growth. The difference between the flexible and rigid responses of a model in which these adjustment costs are scenario was, however, not very pronounced. Thus, the reduced by 20 per cent across the board. observed difference in the adjustment to shocks arising

To see the effect of a more flexible economic from a quite strong variation in adjustment costs in the environment, both the baseline and the flexible scenario rigid and the flexible scenarios suggests that structural were exposed to all the structural shocks identified in the rigidities are unlikely to be the only explanation of why model. It turned out that the impact was stronger in all the latest rebound of the euro-area economy was less

cases for the flexibility scenario than for the base line. dynamic than in other industrial economies.

However, the difference was often only very small

between the standard case and the case for lower 4.2 The recovery’s employment-content

adjustment costs with the notable exception of A possible consequence of resilient employment growth investment, where differences turned out to be in the downturn is a more muted response of significant. These first results with the new model point employment growth to the upswing. This could imply to rigidities affecting cyclical activity first of all via their that households’ disposable income will increase effect on investment. In order to allow comparability relatively little and private consumption will expand at a with the Quest model, Graph 19 displays the impact of a modest pace, giving relatively little impetus to the

TFP shock on real output (GDP) and some other key upswing.

variables (investment, consumption, capital,

employment, inflation) over the first five years after the • The hypothesis of only modest future job creation

shock. finds support in the observation that the current rate

of unemployment is only slightly higher than the

estimate of structural unemployment. 18 This

suggests that there is little cyclical slack in the

17 The model is a New-Keynesian Dynamic Stochastic labour market, which would undo during recovery.

General Equilibrium model for the euro area. It is estimated using Bayesian estimation methods on quarterly data from 1980 to 2003. The model belongs to the class of new models building upon the New-Keynesian paradigm, which

combines elements from the RBC literature with more 18 The Commission estimates the NAWRU (Non-accelerating

traditional Keynesian ideas. This allows combining optimal wage rate of unemployment) is about 0.5 percentage point behaviour with rigidities in a way which avoids the Lucas lower than actually unemployment in 2004 and 2005. The critique. For a description of the model, see Ratto, Roeger NAWRU is a concept similar to the NAIRU but focuses on

and in’t Veld (2004). the impact on wage growth rather than on inflation.

Graph 20: Persons employed and unemployed, euro area Graph 21: Working time, euro area

135 10

Employment (lhs) 20 Average hours worked (rhs)

19 1640 ed

Unemployment 11 18 Part time workers (lhs) 1620

(inverted rhs)

130 t

m pl

oy

17 e

12 m

en 1600

rs on

ns ns 16 pl oy

m 1580 r

pe

M ill io pe M ill io 15

13 ta

l e d

to 14 1560 125 or ke

 o f % 13 1540 rs

 w

14 ou 12 l h

11 1520

ua

A nn

120 15 10 1500

91Q1 93Q1 95Q1 97Q1 99Q1 01Q1 03Q1 92Q1 94Q1 96Q1 98Q1 00Q1 02Q1

Source: Commission services. Source: Commission services, Groningen growth and development center.

• Further support for this hypothesis is based on the

observation that the muted labour market response • Graph 20 shows that the development of

in the early phase of the downswing was partially employment and unemployment decoupled in the due to firms’ labour hoarding. If firms are endowed recent slowdown after having moved in tandem with more labour than usual at the beginning of the throughout the 1990s. This indicates that the trend rebound, they would need to create fewer jobs if of rising labour force participation is unbroken in

demand strengthens. the euro area.

20

• The early phase of the US recovery has been • Working time has declined in the euro area. The

characterised by job-less growth. While this is share of part time work increased from 12 per cent partly related to the usual delay with which the in 1992 to 17 per cent at the end of 2003. Average labour market responds to changes in economic working time has declined by about 0.5 per cent on activity, it has also been argued that it could be due average since 1995. In the case of labour market to the structural change that is taking place. tightness, the trend of declining working time might

Unemployed workers cannot count on finding a job reverse.

in the industry in which they were previously Labour demand forces. Labour demand tends to follow employed, but must find a job in different firms or economic activity with a lag and the average lag is industries. This could be more difficult and takes longer in the euro area than in the USA or the United

more time. 19 The same reasoning could apply to the Kingdom. 21 Thus firms are quicker in more flexible

euro area. economies to adjust their work force to changes in Labour supply forces Reassuringly, aggregate labour demand. Assessing the relative importance of cyclical supply should not be a serious obstacle to a future and structural determinants of labour demand is upsurge in employment growth in the euro area. The complicated by the possibility of a structural break in current levels of unemployment, participation and employment-income relationships in the mid-1990s. working time suggest there is still sufficient margin for Since then, employment growth has been stronger than

an expansion of employment.

• About 1.5 million people have become unemployed

between 2001 and 2004. If they were re-employed within one year, employment growth would rise by 1 per cent.

20 Considering that only 62.4 per cent of the population aged

between 15 and 64 were employed in the euro area in 2003, whereas the employment rate was above 70 per cent in some EU Member States (DK, NL, SW, UK), there is further leeway for higher participation. See Chapter 3 for

more detailed analysis.

21 The coefficient of correlation between the growth rates of

industrial production and industrial employment is higher for the USA and the euro area in the periods 1996-2000 and 2001-04. When lags are taken into account, it reaches the

19 For an elaboration of the reasoning and empirical evidence, maximum after 2-3 months in the USA compared to 6-9

see Groshen and Potter (2003). months in the euro area.

Graph 22: Persons and hours worked in industry, euro area Graph 23: Employment growth in industry and services, euro area

106 4

104 3

2

102 1

0

 1 00 100

 = .p

.a .

-1 00

20 98 c

hg

Employment % -2 96 Hours worked per person -3

Industry

Services -4

94 Linear (Hours worked per person)

-5

92 -6

95Q01 97Q01 99Q01 01Q01 03Q01 92Q1 94Q1 96Q1 98Q1 00Q1 02Q1 04Q1

Note: Data for 2004 are Eurostat estimates. Source: Commission services. Source: Commission services.

expected given the observed growth in GDP. 22 The Employment growth continued during the growth

estimates in Chapter 3 suggest that part of the increase slowdown in a few service sectors, cushioning the job in the employment content of growth is of a structural losses in industry. Job creation took place in two

nature. Rising labour market participation of women, principal sectors, i.e. financial intermediation and the sowage

 moderation and possibly labour market reforms called “other services”. 23 Employment in industry was

enacted in some Member States are important drivers of shrinking in all quarters since 2002 and in trade, structural improvements. They imply that comparisons transport and communication services for almost a year. over cycles as shown in Graph 6 can be misleading. For In the service economy, more specifically, net job losses instance, the first half of the 1990s was characterised by occurred in water and air transport services and strong labour capital substitution, which caused a strong computer activities, where the latter had witnessed decline in employment and was itself caused by a pickstrong employment growth in the previous boom. These up in real wage growth. A reversal of this substitution observations somewhat qualify the notion of widespread has been observed in the second half of the 1990s, labour hoarding. Graph 23 suggests that employment spurred inter alia by moderate wage growth. growth in industry and services follows a similar

An important cyclical factor is the existence of labour cyclical pattern and therefore depends on similar hoarding in the slowdown, which suggests that hiring determinants over the business cycle. The main

could be smaller in the upswing. Estimating the level of difference is in the level of the rate of growth.

slack of labour in firms can be done by comparing the According to some economists, the re-allocation of number of persons employed with the number of hours labour across industries and sectors is a main benefit of worked because one would assume that if demand falls a recession because it shifts labour to more productive and labour hoarding exists, the persons employed work uses. A drawback is that structural change might cause a less hours. However, hours worked fell in the euro area delay in the rebound of employment growth because it and they did so most pronouncedly in the period of takes more time and efforts for unemployed workers to

strong demand 1999-2000. At that time, many part-time find a job in a different industry than become rejobs were created, supposedly in services, yielding a employed in the one where they had been employed. decline in average work hours per person. Graph 22 demonstrates that hours worked fell continuously in industry, where the trend towards part time work is less prominent. From 2001 to mid-2003, the decline in hours worked was particularly pronounced. From mid-2003 onwards, hours worked increased while employment in industry continued to fall. This is in line with the assertion that firms cut back the existing slack before

creating new jobs. 23 The statistical category of “financial intermediation”

includes real estate, rening and business activities. “Other services” includes public administration and defence, compulsory social security; education; health and social work; other community, social and personal service

22 This is the mirror-image of the decline in labour activities; private households with employed persons; extraproductivity

growth. territorial organizations and bodies.

Graph 24: Job adjustment by industries during boom and in world import demand than exports of other goods.

25

slowdown An explanation for this finding could be that the decline

of the share of consumer goods in production is due to 3 the weakness of households’ demand in the euro area

rather than driven by a re-allocation of global

4 2 production. -2 00

01 Industry 1 structural growth

Services

w n

20

do 0 4.3 Labour productivity growth, investment and

 s lo

w

potential output growth

th in -1 Hardly any development raised more concerns about the

gr ow

b euro area’s underlying growth potential than the

Jo -2

structural decline cyclical decline observed decline in labour productivity growth. Growth

-3 in average real GDP per person was less than 0.5 per -3 -2 -1 0 1 2 3 4 cent in the period 2001-03, which is about 1 percentage

Job growth in boom 1998-2001 point less than during the 1990s on average. It is well

Note: Due to data availability, job growth in services during the boom known that labour productivity growth is typically procovers

only the period 2000-2001. cyclical, i.e. increasing in a boom and declining in a bust

Source: Commission services. because of the lag with which employment tends to respond to changes in output. Nevertheless, the extent of

In order to shed some light on the question of to what the slowdown in labour productivity growth came as an extent job losses were cyclical or of a structural nature, unwelcome surprise. Some argued that it could indicate

Graph 24 relates employment growth in the previous a slowdown in the rate of potential output growth.

26

boom period with employment growth in the subsequent Since labour productivity growth is the key determinant slowdown in different industries. It neatly illustrates that of potential output growth in the medium to long term, job creation was prominent in the service economy this section addresses the arguments over whether the whereas jobs got lost, even during the previous boom most recent slowdown in productivity could be of a period, in some industrial sectors. Sectors where temporary or longer term nature. More detailed analysis employment growth or shrinkage was the same in both of this issue can be found in Chapters 3 and 4 of this

boom and slowdown seem to have witnessed structural Review and Chapter 2 of last year’s edition.

employment growth or deceleration, respectively The slowdown in actual labour productivity growth (Structural decline and structural growth quadrants in already had an impact on calculations of potential output Graph 24). Employment losses might be of a more growth. The Commission’s estimate, the method for cyclical nature in those industries where employment which is quite sensitive to recent developments, grew in the previous boom but declined during in the indicates that potential output growth in 2004 has been

24

slowdown (Cyclical decline quadrant in Graph 24). around 2 per cent, which is a good deal below the

Re-structuring within the manufacturing sector seems to widespread rule of thumb that sees potential growth take place from the production of consumer goods to the closer to 2.5 per cent. Graph 26 shows the Commission production of capital goods. The sector that produces estimates of potential output growth in 2000, when the consumer durables saw the most pronounced decline in method was first introduced, in 2002, which was done employment in the previous boom as well as in the with a refined approach, and in 2004. The sensitivity of slowdown. On the other side, employment increased in the method to actual developments, however, implies the boom and moderately declined afterwards in the that potential output growth estimates could return to capital goods-producing industry. A similar sectoral higher rates under the condition of a pick up in labour

change is visible in production figures. Industrial productivity growth.

27

production of consumer durables in the euro area declined by 10 per cent between 2001 and 2004, but was almost steady for capital goods production. Surprisingly, a similar change in the weight of sectors is not visible in export figures. Extra euro-area exports of consumer

goods were no more strongly affected by the slowdown 25 A breakdown into durables and non-durables is not

available for trade figures.

26 See, for instance, Gern et al. (2004). 27 The IMF revised downwards its estimate of potential output

24 Empirical evidence on the link between labour market growth in the euro area by 0.5 percentage point to 2 per cent

regulation and performance has been scant. However, in summer 2004, see IMF (2004a). CEPS (2003) reports Caballero et al. (2004) present new evidence that that long-term growth expectations in the Consensus employment protection regulation hampers the re-allocation forecast fell by about 0.5 per cent to 2.2 per cent from 2000

of jobs. to 2003.

Graph 25: Labour productivity growth Graph 26: Estimates of the rate of potential growth, (GDP per person employed, euro area) euro area

2.5 3

Autumn 2000 estimate 2.8

2

Actual GDP per 2.6 Autumn 2002 estimate

person employed

1.5 2.4

.

. p.

• a. 

  2.2

1 p

.a

 c hg hg 2 % % c

1.8 0.5 Autumn 2004 estimate

1.6

0 1.4

1.2

-0.5 1

96Q1 97Q1 98Q1 99Q1 00Q1 01Q1 02Q1 03Q1 04Q1 * * 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Source: Commission services. Source: Commission services.

In addition to actual productivity developments, which The first observation is regarding the impact of resilient can only give limited insights into whether the labour market performance on productivity growth. The slowdown is permanent or a cyclical phenomenon, a above discussed how labour hoarding by firms number of arguments can be brought forward as to why artificially reduces productivity growth on a temporary the slowdown in productivity growth could be of basis. Moreover, economic growth has become more structural nature. employment-intensive since the mid-1990s due to job

• Labour productivity growth has been on a creation in sectors with low or difficult-to-measure downward trajectory since the 1970s. It continued productivity. Analysis in the 2003 Review, updated in to head downwards in the euro area also in the the current issue’s Chapters 4 and 5, suggests that

1990s. 28 employment creation could explain about 20 per cent of the reduction in labour productivity growth measured

• Expectations that the reversal of the productivity during the 1990s.

trend in the USA could also occur in the euro area

have somewhat lost their justification after recent A second observation is related to the decline in investment witnessed over the past years. Graph 27

estimates on the basis of improved data have shown

that the gap in hourly productivity between the euro demonstrates the degree of co-movement between area and the USA is much smaller than previously investment and two measures of productivity growth

thought. Thus, the forces that would drive the catchover the 1990s. The graph suggests that the generally close relationship between investment and productivity

up process to the US level are weaker than believed growth did not hold in the late 1990s when investment a few years ago. activity boomed but productivity growth hardly

• There is now less reason than previously to assume accelerated. Possible reasons are the counter-veiling that the productivity growth acceleration due to ICT influence of strong employment growth in this period or witnessed in the USA would equally materialise in investment taking place in less productive parts of the

the euro area. A large part of these productivity economy.

gains accrued in the production of ICT, which has a larger share in production in the USA than in the euro area.

Some observations suggest that cyclical factors are responsible for part of the productivity slowdown witnessed since 2000. Their impact may fade or even revert once demand growth gains momentum.

28 Empirical estimates of productivity growth that control for

different input variables tend to find the time trend to be a significant determinant that enters the estimates with a negative sign.

Graph 27: Investment and productivity growth, euro area Graph 28: Corporate debt and investment, euro area, % of GDP

3 10 % %

70 22

6 2

65 21

• a. 

  2 a.

. p. . p.

hg 1 hg 60 20 % c -2 % c

0 55 Non-financial Labour productivity -6 corporate debt (lhs)

19

TFP growth Investment share (rhs) Investment (rhs)

-1 -10 50 18 1992 1994 1996 1998 2000 2002 2004 2006 97Q4 98Q3 99Q2 00Q1 00Q4 01Q3 02Q2 03Q1 03Q4

Source: Commission services. Source: Commission services.

In particular, the latter hypothesis fits with the Technical progress is a further factor impacting on assessment that investment since the late 1990s has investment and productivity. As stated, many firms followed a boom-bust pattern. Declining interest rates founded in the late 1990s that tried to exploit technical and buoyant stock market valuations, in combination advances in high-tech and ICT seem not to have with a wave of optimism stemming from potential gains withstood the growth slowdown. Moreover, benefits from the diffusion of ICT, advancing economic and from ICT use appear now to be smaller than assumed financial integration and not at least the emergence of earlier. But it should not be forgotten that incentives for macroeconomic stability are thought to have been research and innovation may vary with the cyclical

particularly conducive to investment. 29 The consequence situation. The incentive to innovate is largest when

of strong investment was a sizeable build up of economic activity is buoyant because this allows the corporate debt that did not prove sustainable when entrepreneur to accrue the maximum benefit from his economic growth waned. Graph 13 (Section 3.2) gives initial monopoly margins before imitation reduces his

an overview of main developments, which were profits. 30

discussed in more detail in the EU economy 2003

Review. It turned out that corporations trimmed down Investment goods continue to become cheaper than other investment in order to restore balance sheets. Issuance goods, suggesting that technical progress has not slowed activity to obtain financing from markets came to a down. Although the economic significance of technical virtual standstill for non-financial corporations. Banks progress is difficult to gauge, some indicators such as became particularly prudent in their lending policies, data on patenting activity, R&D expenditures and relying to some extent on a tightening of non-price venture capital financing are available. Graph 29 shows terms. This made it difficult for especially small, young an indicator derived from price developments directly enterprises and firms acting in a risky environment to linked to investment activity. It rests on the assumption obtain financing. Industry reports suggest that many of that any improvement in investment-specific technical the firms founded in the late 1990s and active in high change should – under the assumptions of a competitive tech or ICT did not survive the slowdown, implying that market – reduce the prices of the investment goods

the expectation of profits from exploiting technological concerned relative to other prices in the economy.

31 So,

advances had been exaggerated during the previous the relative price of investment goods can be a proxy for boom. The number of bankruptcies rose in a number of the pace of technical progress embedded in investment

euro-area countries. goods. Against this background, Graph 29 shows the price of three categories of investment good, expressed

in relation to consumer price developments. Investment in equipment has become cheaper over time, relative to

29 Total investment growth as recorded in the national accounts seems not to properly reflect the strength of 30 investment activity, which was most concentrated in ICT For a more elaborated view on how the cycle may interact and high-tech industries. Moreover, continuous diswith innovative activity, see Wälde (2004) or Francois and investment in the German construction sector as well as a Lloyd-Ellis (2003) and the references quoted therein.

considerable share of investment taking place in the form of 31 This variable usually turns out to be significant in empirical

both FDI and corporate re-structuring via the acquisition of estimates of investment. See the chapter on investment in financial assets lead national accounts data for the euro area the EU E CONOMY 2001 R EVIEW . For an appliation of the

to underestimate the strength of investment activity. concept, see also Fisher (2003).

Graph 29: Relative price of investment goods, euro area 4.4 Determinants of households’ savings

behaviour

3

2 4.4.1 Documenting the weakness of private

1 consumption and the rise in the saving rate

0 Since the beginning of the latest downturn, consumer expenditure in the euro area has been very subdued.

oy

 y -1

% Between the second quarter of 2001 and the fourth

-2 quarter of 2003, private consumption practically stalled,

growing by a mere 0.2 per cent (quarter-on-quarter) on -3 average. Although over the long run, consumption and

-4 real disposable income are closely correlated in the euro Other products Equipment Construction area, significant divergences have recently taken place.

-5 Between 1992 and 2000, for instance, the annual 96Q1 96Q4 97Q3 98Q2 99Q1 99Q4 00Q3 01Q2 02Q1 02Q4 03Q3 04Q2 average growth rate of household consumption was

Note: Difference in annual percentage growth over consumer goods. 1.9 per cent compared to a growth rate of 1.2 per cent

Source: Commission services. for disposable income. By contrast, over the last three years, the growth rate of consumer expenditure was

consumer prices and investment in construction or other constantly below that of real disposable income. On products and the price advantage even increased during average, between 2001 and 2003, consumption grew by

the slowdown. 3.3 per cent, while disposable income rose by 3.6 per cent. 32

The factors that affected investment growth during the

growth slowdown were either of a temporary nature or In contrast with the weakness of capital spending, which can be expected to reappear once economic activity was a common feature of many industrialised economies picks up momentum. Graph 30 summarises the cyclical during the downturn, the weakness of household development of the major determinants of investment as spending was more specific to the euro area. Subdued they were expressed by industrialists in the consumer spending over the last three years was also Commission’s bi-annual investment survey. Demand mirrored by an increase in household saving rates across and financial conditions were seen as very supportive to most euro-area Member states. This, again, contrasts investment in the boom period of the late 1990s. This with developments in some of the euro area’s main assessment weakened considerably afterwards and trading partners.

returned in 2004 to its average 1991-2004. Also the conduciveness of technical factors weakened after 2000

and subsequently re-approached its long-term average. Graph 31: Household saving rate in cyclical downturns, euro

area (in % of disposable income)

Graph 30: Factors driving industrial investment, survey data euro area 16 12

50 15 11

40

30 14 10

s

ce 20

an

B al 10 13 1992-94 9

0 2001-03 (rhs)

-10 12 8

Technical Demand Financial Q0 Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11

-20

1991 1993 1995 1997 1999 2001 2003 Note: Q0 corresponds to the peak preceding the beginning of the downturn (92 Q1 for the 1992-93 recession and 2000 Q4 for the 2001-

Note: Factors do not add up to 100 because multiple replies are possible. 03 downturn)

Other factors are not shown in the graph. Source: Fagan et al. (2001) and Commission services.

Source: Commission services.

32 Gross disposable income and consumption in current prices

in the euro area, excluding Ireland, Luxembourg and Portugal for which data was not available.

According to the standard (life-cycle) approach, 1990s and more resilient in the ensuing downturn in the

households save because of a desire to provide for a

smooth consumption profile over time in the presence of

cyclical income fluctuations and various types of Graph 32: Household actual and simulated consumption,

uncertainties concerning lifetime resources. Thus, the euro area (in billion of 1995 euro)

rise in the household saving rate, at a time of subdued economic activity, is difficult to reconcile with recent 890

business cycle facts. Consumption smoothing over the 870

life-cycle would imply a decline in household savings

during a slowdown. In the recession of the early 1990s, 850

for instance, households partly offset adverse 830 developments in disposable income by curbing their

savings rate (Graph 31). 810

790 Actual consumption

There is no completely satisfactory explanation for the

weakness of consumer spending in the euro area over 770 Simulated consumption (1) the last three years or why households saving rates 750 recorded a similar “adverse” pattern. This section briefly Jan-95 Jan-97 Jan-99 Jan-01 Jan-03 elaborates on a number of possible factors behind recent household consumption and saving behaviour, including Note: (1) As simulated with the consumption function over 1995-2003. the lingering effects of past large equity losses, house Source: Commission services. price developments, high perceived inflation, downward revision of expected future income streams and, finally,

uncertainty about policy and structural reforms. absence of the observed movement of stock market prices. Furthermore, given the long lags involved,

As regards potential implications of subdued negative stock market wealth effects were still weighing consumption growth in the past for the future pace of on consumption during the second half of 2003, despite consumption, it can be hypothetized that sluggish a pick-up in stock prices.

consumption could continue if the increase in the saving

ratio was caused by structural factors that can be However, while the estimated consumption function expected to persist. On the other hand, if the factors tracks actual developments in consumption over the causing subdued consumption are of a cyclical nature or 1990s relatively well, including the 1992-93 recession, it

may be influenced by policies, a reversal of past trends clearly performs less well after the turn of the decade. As shown in Graph 32, household spending has

could be hoped for. remained below that predicted by the equation since

2001. Overall, the estimated consumption function

4.4.2 The impact of wealth and financial factors suggests that the level of private consumption is

currently somewhat weaker – by about 0.7 per cent –

In standard consumption equations, household than what would be expected given prevailing consumption and saving decisions are affected by other macroeconomic conditions.

factors apart from disposable income, such as wealth,

interest rates and inflation (the latter is often used as a Due to the lack of appropriate quarterly data, non-stock proxy for the uncertainty concerning the real market wealth effects were not included in the estimated depreciation of non-indexed financial assets). Equations equation. As the contraction of total financial wealth using such variables generally provide a good fit for over the past few years has been less marked than the consumption and savings developments. A consumption contraction of stock market wealth and, at the same function of this type was estimated by Commission time, residential property prices have been picking up,

33

Services using euro-area quarterly time series. the equation may thus overestimate the negative impact According to the estimated equation, the weakening of of wealth on consumption. A major implication is that consumption over the 2001-2003 period is mainly due to recent consumption developments may have been even developments in disposable income and the bursting of more subdued relative to prevailing macroeconomic the equity bubble. In addition, the rise in inflation conditions than suggested by the consumption function.

weighed on private spending in 2001-2002, while the However, the interrelation between house price short-term interest rate provided moderate support for movements and consumption is not straightforward. On consumption in 2002 and 2003. Overall, the estimated the one hand, wealth effects may encourage

equation attributes an important role to stock market

wealth effects in the current cycle. Private spending consumption especially if the capital gains from the appreciation of house prices can be extracted via

would have been significantly less buoyant in the late mortgage capital withdrawal. This possibility, however,

is much less commonly used in the euro area than in the USA and the UK. On the other hand, higher house

33 See Q UARTERLY R EPORT ON THE E URO A REA I-2004 for prices may induce households to save more in order to

details.

Graph 33: Headline inflation and perceptions of past consumers’ marked overestimation of actual inflation. inflation, euro area (annual % rate of change, 3mma) The divergence between actual and perceived inflation may offer some explanation of the euro area

6 70 consumption/saving puzzle. However, it may also be the

case that the gap actually captures other non-observable

60

5 HICP factors such as a re-assessment of long-term growth

Perception of past inflation (rhs) 50

) perspectives or heightened uncertainty. vg 4 ) 40

 a vg (3 m

 a

3 30 ( 3m 4.4.3 Downward revision of long-term growth p .a

. ce

hg prospects

 c 20 % 2 B al an

10 A second line of reasoning attributes the weakening of

1 consumption growth in the euro area could be due to a

0 downward revision in households’ expectations of future 0 -10 income growth.

35 These expectations could stem from a

1991m01 1993m01 1995m01 1997m01 1999m01 2001m01 2003m01 number of factors, among which the impact of ageing on future labour input is the most obvious. The trend

Note: Consumers’ inflation perceptions are balanced responses on decline of labour productivity growth, as well as weaker

“price trends observed over the last 12 months” reported by the EC

survey on consumer confidence than expected actual GDP growth during the 2001-03

Source: Commission services. period, could also matter. In the same context,

Commission estimates of potential output growth were downscaled by about 0.5 percentage point between 2002

accumulate funds to cover the higher down payments and 2004 (see Graph 26 above), supporting the notion required for future house purchase or for capital that previously-more-upbeat growth prospects were repayment. See Box 2 for an overview of the impact of corrected.

housing prices on private consumption. The impact of diminishing expectations of future income

Perception of current and past inflation. Consumer’s

34 growth on current spending could be substantial. The

perceptions of the current level of inflation may also permanent income hypothesis predicts an immediate

have exerted a drag on household spending. In theory, adjustment of spending patterns. 36 If households expect

inflation perceptions should not affect consumer a decline in the discounted value of future labour

behaviour as, on average, they must equal actual income, savings should increase and consumption

inflation. In fact, the rate of inflation as measured by the should accordingly drop. A reduction of expected future

HICP Index and inflation perceptions as measured in the income growth from 2.5 per cent to 2.0 per cent, in line

European Consumer survey were broadly in line for with the downward revision of potential output growth,

most of the nineties up to the year 2001 when the euro would for instance yield a decline in the present value of

was physically introduced (Graph 33). Since then, there future income by a quarter of a percentage point,

has been a structural break between perceived and actual assuming a discount rate of 4 per cent, and even more if

inflation with the consumers systematically the discount rate is smaller. Consumption may be

overestimating actual inflation developments. The rise in assumed to shrink by about the same amount. 37

perceived inflation may have led households to

underestimate their purchasing power and, in turn, to There are, however, problems with the application of the

curtail their consumption. permanent income hypothesis. First, since permanent

income is the unobserved outcome of consumers’

The importance of inflation perceptions as a source of rational expecations, it cannot be tested directly. Despite consumption weakness in the euro area in recent years being a theoretically sound concept, i.e. being consistent should not be overestimated, however. Note for instance with the paradigm of the optimising individual, its

that, while in 2002 the rising gap between actual inflation and its perception by households may have acted as a drag on consumer spending, the gap narrowed significantly during 2003 without triggering a noticeable 35

release of pent-up demand. Also, it is noteworthy that See CEPS (2003), IMF (2004). 36

consumption weakness and the increase in households’ See Attanasio (1999) or Romer (2001) for overviews of the

savings rate took place in 2001, i.e. before the apparent theory and empirical evidence supporting the permanent income hypothesis.

37 Dreger and Reimers (2003) estimate that the elasticity of

consumption with respect to income is close to 1 in the euro

34 Note the difference between inflation perceptions and area. This is considered to be compatible with the

inflation expectations used in Graph 33 and 43. Both permanent income hypothesis. A similar elasticity was variables are drawn from the Commission’s consumer found in the Commission estimates published in the surveys, with the first variable relating to households’ Q UARTERLY R EPORT ON THE E URO A REA I/2004. In theory, assessment of past price trends and the second to future the magnitude of the elasticity depends on the properties of price trends. the utility function.

Table 5: Correlation between future labour force growth and indicators of private consumption across

euro-area Member States

Consumption growth Deceleration of Increase in gross

consumption growth saving ratio

Per capita, avg. 2001- Avg. 2001/04 over 2004 over 2000,

04 Avg. 1996/2000 9 Member States

Avg. growth in 2005-25 0.03 0.31 0.15

labour force 2005-50 0.03 0.14 0.36

Deceleration compared 2005-25 -0.32 0.68 -0.43

to 2000-05 2005-50 -0.25 0.65 -0.45

Source: Commission services.

predictions are difficult to reconcile with some empirical

observations. So, consumption growth should be Graph 34: Consumption growth and downward revision of potential output growth

counter-cyclical and not pro-cyclical. Specifically,

consumption should be invariant to foreseen changes in 4 2

38

income. However, the majority of the households have w th R = 0.3629

little wealth and need to finance consumption from ro EL

current income. Second, whereas a sudden downward pi ta g

  FI

r ca 00 ES revision in potential growth could explain the structural

pe 6- 2

  IE

  LU FR

break in the consumption function (Graph 32), it is 1 99 pt io n er

difficult to justify why expectations should have m

  AT

4 ov IT su BE adjusted abruptly. Most of the factors that could on -0 DE motivate a downward adjustment, namely ageing, labour in

 c 01

20 0 PT NL

productivity trend, actual growth, emerged gradually. io

n

The impact of ageing and the trend decline in labour el er at

productivity on future income was already known in the D

ec -2

late 1990s, when consumption growth was high. Finally, -1.5 -1 -0.5 0 0.5

it seems unlikely that the average consumer adjusted Downward revision in potential output growth consumption to a change in future income years before Source: Commission services.

economists started speculating that long-term growth prospects could have waned. If the hypothesis that the expected decline in income

Graph 34 shows that the correlation between the growth causes the recent weakness in private downward revision of potential output growth and consumption is true, there should be some relationship private consumption growth across euro-area Member between current consumption indicators and variables States is positive but weak. Those countries where the impacting on future growth. Countries where the labour potential output growth estimate was only mildly force is shrinking substantially and where growth will downscaled witnessed the highest consumption growth. therefore decelerate more than in other countries could, However, this result is partly due to the indirect impact for example, witness particularly subdued consumption of actual consumption growth on the potential output growth. This line of reasoning is sometimes heard when estimate, because consumption is the most important it comes to explaining the weakness of consumption in component in GDP and actual GDP has an impact on the Germany. Investigating this hypothesis by looking at calculation of potential GDP. As a cross-check of this cross-country variation in labour force growth and hypothesis, there is no evident relationship between the consumption growth yields mixed evidence. downward revision of GDP on the one hand and the Coefficients of correlations do not exhibit the expected deceleration of consumption growth or the increase in sign and their magnitude varies depending on the the saving ratio 2001-04 on the other hand. indicators used. Correlations are strongest between the

change in the future growth of the labour force and the change of consumption as well as with the change in the saving ratio. This would be consistent with the permanent income hypothesis if one assumed that agents had previously counted on a continuation of current labour force trends and this expectation was revised

38 Tests of whether consumption Granger-causes income, when more realistic population projections became

which is a prediction from the permanent income

hypothesis, yielded mixed results. While the results for the public during the debates on pension reforms.

euro-area aggregate and Germany supported the hypothesis, those for France did not.

Graph 35: Expected and actual unemployment, euro area Graph 36: Change in unemployment and saving ratio, euroarea Member States

0.8 50 6 -0 4

0.6 00 Change in rate of ) 20

0.4 40 unemployment (lhs) D P 4 NL 0.2 Unemployment G t of

0 expectations (rhs)

30 ar p

r 1

ye ce ( % 2

ve -0.2 20 tio

DE EUR-8

t o B

al an IT

2

% p -0.4 in

g ra AT BE R = 0.1271

FI

-0.6 10 0 ES

-0.8 e

in s

av

0 ng FR

-1 C ha -2

-1.2 -10 -1 -0.5 0 0.5 1 1.5 2 2.5

Jan 95 Jan 97 Jan 99 Jan 01 Jan 03 Change in expected unemployment 2000-04

Source: Commission services. Source: Commission services.

4.4.4 Precautionary savings and public policy various variables. However, the use of cross-country

Last but not at least, private consumption could have data means that only a limited number of observations is been discouraged by the consumers assessment of the available, 12 for consumption aggregates, 9 for the gross state of the economy. Households may have increased saving ratio and 5 for the net saving ratio. In principle, their precautionary savings because of heightened the ideal variable to be explained is the residual of a uncertainty during the growth slowdown. The consumption function, which depends on the availability

expectation of rising unemployment seems to have of such an estimate for each Member State.

dented consumer confidence as suggested by the strong As regards the impact of fears of unemployment on correlation between consumers’ assessment of saving behaviour, Graph 36 shows a very weak unemployment prospects, which is an important relationship between the increase in expected component of the Commission’s consumer confidence unemployment between 2000 and 2004, as expressed in indicator, and actual changes in the rate of the Commission consumer survey, and the change in the unemployment (see Graph 35). Elevated geo-political household saving ratio over the same period across uncertainty after the terrorist strikes of 11 September Member States. The implied trend line would be close to might have also contributed, although the structural horizontal, suggesting no relationship, if the Netherlands

break found in the empirical estimates presented above were taken away from the sample. 40 The suggested link

took place before autumn 2001. Households may also is also much weaker if based on the actual change in have increased savings in response to the uncertainty unemployment instead of the expected trend in

arising from the impact of structural reforms on future unemployment in the consumer surveys.

39

income. An important factor could also have been that

the deterioration of government balances led to so-called Ideally, a microeconometric approach would be needed Ricardian effects. Such effects rely on households’ to reveal any impact of the uncertainty related to expectations that higher public deficits today mean structural reforms and their impact on consumption via higher future taxes and should accordingly be met the uncertainty about future income. Households

through higher present savings. exposed to the effects of structural reforms would be expected to show different spending behaviour to those

Discriminating between the different hypotheses not or less affected. Experiments at the aggregate level empirically is again hampered by data considerations. similar to the cross-country comparison above using Finding adequate proxies for precautionary savings and various indicators of poverty and inequality do not factors capturing uncertainty is difficult. The following suggest any systematic relationship with the changes in graphs present an attempt to distill the reasons for the saving behaviour. Estimates did not produce any sluggish consumption from cross-country variations in evidence that households increased their saving ratio by more when they lived in Member States where income inequality or poverty rates were higher or where

40 This also holds if the estimate controls for the increase in 39 BIS (2004) conjectures that uncertainties related to households’ saving ratio over the 1990s. The variable

“proposals to urgently confront structural deficiencies in capturing the expected and actual change in unemployment, labour markets, pensions, health care and tax administration respectively, is significant if the Nethelands is included in

may have .. played a role here”. the sample but insignificant without the Netherlands.

governments spent more on public transfers, rate. However, the coefficient is not significantly respectively, than those where inequality, poverty and different from zero. Overall, these cross-country

public transfers are lower. 41 exercises did not provide clear results. This supports the

The most controversial debate in this area focuses on the point that more sophisticated methods that can both impact of public finances on private consumption appropriately control for the different intervening effects behaviour. Some Member States were hesitant in and are able to discriminate between the effects of the pursuing budgetary consolidation during the growth different variables in a common setting are required

slowdown with the same vigour as in the late 1990s. before policy-relevant insights can be expected.

Last year’s Review highlighted three pieces of evidence

supporting that fiscal stimuli seem to have resulted in Graph 37: Household saving and the fiscal stance

less private consumption. First, there is a close

correlation between the budget deficit GDP ratio and the NL )

households’ saving ratio over time. Second, there is a . in c. 3

puzzling negative correlation between quarterly private

and public consumption growth since summer 2001. t of

 d is

 p

not yielded the expected increase in private ra tio 00 DE IT 20

consumption. in g 1

FI AT

The correlation between the variation of households’ in

s av EUR-8 BE

e ES

saving ratio 2001-04 and the cyclically-adjusted primary ng balance (CAPB), which is the usual measure of the C

ha FR

R 2 = 0.0595

fiscal injection, across Member States, does not provide -1 -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0

clear evidence. Countries where the fiscal stimulus was Change in CAPB 2000-04

largest, i.e. a larger deterioration in the CAPB, experienced the largest increase in the gross saving Source: Commission services.

41 The same results hold if the social indicators are expressed

as changes over time. The information on the change over time suffers from the fact that the most recent observation on the structural indicators of income inequality, at-risk-ofpoverty and persistent risk of poverty are from 2001. To calculate the change over time, the 2001 information was compared with the average for 1995-2000. Another drawback is that the estimates could only be conducted for 9 Member States, for which information on the households’ saving ratio is available.

36

Box 2: The housing market and private consumption

Recent developments in house prices. Since the late 1990s, residential prices have increased significantly faster than consumer prices in the euro area as well as in the USA and the UK (see left panel of the chart below). Price rises were largely unaltered by the slowdown in economic activity after 2000, a development which stands in sharp contrast with the drop in real residential prices registered during the first half of the 1990s.

Recent trends within the euro area offer a contrasting picture (see right panel of the chart below). A majority of euro-area Member States have experienced a robust rise in real residential prices since 2000. However, several countries have registered either sluggish growth (the Netherlands and Finland) or an outright fall in real prices (Austria, Portugal and Germany).

Graph : Recent development in real residential property prices (1)

Euro area, USA and UK (annual growth in %) Euro area (average annual growth in % 2001-03)

20 14

15 Euro area USA UK

12

10

10 8

6

5 4

2 0

0

-5 -2

-4

-10 -6

1991 1993 1995 1997 1999 2001 2003 BE DE EL ES FR IE IT NL AT(2) PT FI EU12

Note: (1) consumption deflator. (2) 2001-02 for Austria.

Source: ECB, BIS and Commission services.

The possible impacts of housing markets on private consumption. There are several possible channels through which developments in housing markets may spill over into private consumption.

Firstly, changes in house prices affect household spending via changes in wealth and relative prices. Economic theory does not provide clear guidance as to the potential impact of a rise in house prices on private consumption. For landlords, a rise in house prices will have a positive impact on consumption via the wealth effect. Things are, however, somewhat different for owneroccupiers for whom the positive wealth effect will be partly offset by an increase in the imputed cost of the housing service associated with their home. Higher house prices will also have a negative incidence on the consumption spending of prospective new buyers who will need to save more to be able to afford their house purchase. Finally, to the extent that they permeate into rents, higher house prices will also weigh on tenants’ private spending.

A second channel is linked to the existence of credit market imperfections and asymmetric information. Insofar as houses are used as collateral, increases in house prices will raise the value of the collateral, thus allowing households to obtain additional credit and lift their consumption. The existence of such a channel is closely tied to the structural characteristics of the mortgage market and, in particular, the possibility of housing equity withdrawal.

Thirdly, housing markets also constitute a transmission channel of monetary policy that may have a direct incidence on private consumption. Changes in monetary policy affect mortgage rates and thereby the disposable income of indebted households if the mortgage rates of existing mortgage contracts are tied somehow to market mortgage rates.

Cross-country estimations of the size of the housing wealth effect. Although the issue of the size of the housing wealth effect has attracted significant attention in some countries such as the USA and the UK, empirical research on the euro area is still relatively sparse and cross-country comparisons remain relatively difficult. Nevertheless, several recent studies have endeavoured to quantify housing wealth effects in a multi-country setting. Methodologically, this recent research can be grouped along two types of approaches.

• Some studies rely on cross-sectional panel regressions to provide an average picture of the housing wealth effect in advanced

economies. For instance, Case et al. (2001) estimate a simple consumption function relating private spending to income, equity wealth and housing wealth on a panel of 14 OECD countries and find elasticities of consumption relative to housing wealth in the range of 0.11-0.17. Ludwig and Sløk (2004) estimate a more sophisticated consumption function on a sample of 15 OECD countries. Using an error-correction model (where consumption is determined by a few selected variables in the long-run but can temporarily deviate from its long-term path) the authors find a substantially smaller impact of housing prices with a range of long-term elasticities of 0.01-0.04. This suggests that, although the housing wealth effect is generally

positive in the OECD, the uncertainty regarding its size remains high.

• A number of studies have tried to assess the importance of housing wealth effects by estimating similar consumption

functions across several countries, thereby allowing more meaningful cross-country comparisons. The next table provides a snapshot of the importance of housing wealth effects according to five such studies.

Table: Estimated size of the housing wealth effect on private consumption

Selected recent cross-country studies

E & S (1) B & P (2) H & M (3) Catte et al (4) B & G (5)

Elast / house Elast / house Elast / house MPC out of MPC out of

prices prices prices housing wealth housing wealth

BE 0.0 0.13

DE 0.0 -0.12 0.0

ES 0.0 0.02

FR 0.05 0.0 0.05

IE 0.06 0.43

IT 0.0 0.0 0.01 -0.06

NL 0.07 0.06 0.13 0.08

FI 0.09 0.07 0.15

UK 0.07 0.16 0.21 0.07 0.04

USA 0.16 0.05 0.03

(1) Eschenbach and Schuknecht (2002) – Based on a simple growth equation. (2) Barata and Pacheco (2003) – The elasticities of consumption relative to house prices are derived from the long-run equation of an error correction model. (3) Henley and Morley (2001) – The elasticities of consumption relative to house prices are derived from a model of consumption growth with progressive adjustment to permanent income shocks. (4) Catte at al. (2004) – The marginal propensities to consume out of housing wealth are derived from the long-run equation of an error correction model. (5) Boone and Girouard (2002) – The marginal propensities to consume out of housing wealth are those of the long-run equation of an error correction model.

Overall, several features stand out from the recent empirical research:

• Whereas housing wealth effects are generally identified in Anglo-saxon countries, evidence for euro-area Member States is

mixed. Positive housing wealth effects can be found in some smaller Member States such as the Netherlands, Finland and Ireland. In contrast, it is likely that housing wealth has, at best, only a limited impact on consumption in the larger Member States, with France possibly being an exception. This explains the fact that a housing wealth effect is generally difficult to

identify in the euro area as a whole.

• There is still substantial uncertainty as to the size of the housing wealth effect as reflected by large ranges of elasticities in

most countries. Cross-country comparisons should therefore be made with caution. The estimation of the housing wealth effect is complicated by endogeneity issues (strong co-movements in house prices and consumption do not necessarily indicate a strong causal link from house prices to consumption) and serious data limitations. In particular, housing wealth data are available in only few countries and cross-country comparisons of wealth and house price data are still hampered by

insufficient statistical harmonisation.

• The analysis of the impact of housing on private spending may be complicated by the possible existence of complex shortterm

dynamic effects. All studies based on error-correction formulations of the consumption function establish a long-term relation between housing wealth and private spending. However, in some countries there is also some evidence that housing wealth may play a strong role in the short-term developments of consumption, with possible overshooting effects (see, for

instance, Catte et al. (2004)).

Finally, it is worth noting that little empirical research is yet available on the possible sources of country differences in the strength of the housing wealth effect. A major exception is Catte et al. (2004) who find a strong correlation between the importance of housing equity withdrawal and the estimated marginal propensity to consume out of housing wealth. This indicates that households’ ability to extract cash from increases in house prices is probably a key factor explaining the strength of housing wealth effects. Nevertheless, several other parameters may also be pivotal in this respect and would require further empirical research, including the rate of owner occupation and the size of housing transaction costs.

Explaining differences in the resilience of private consumption. The empirical research presented above suggests that developments in the housing market provide some help in explaining recent differences in the resilience of private consumption between the euro area and other OECD countries. In the past few years, household spending has probably been supported by a significant housing wealth effect in the USA and the UK, whereas gains in consumption attributable to housing wealth have, at best, had only a modest impact in the euro area as a whole.

Nevertheless, given the uncertainty still surrounding estimates of the housing wealth effect, the importance of this factor remains difficult to assess. For illustrative purpose, the left panel of the next chart displays the differences in per capita consumption growth between the euro area on the one hand and the UK and the USA over 2001-03. The chart also shows a range of estimates of the possible contribution of house prices to consumption in the UK and the USA for the same period. The contributions of house prices to consumption are calculated by simply applying elasticities of 0.05 and 0.15 to developments in real prices in the two countries (the range of elasticities broadly covers the estimates presented in the previous table). The exercise suggests that, even with conservative estimates of the involved elasticities, housing prices have made a significant contribution to consumption growth in the USA and the UK in the past few years. However, it is necessary to assume a very large difference in housing prices elasticities between the euro area (on the one hand) and the UK and the USA (on the other) for the housing wealth effect to become the main explanation of the comparatively poor performance of consumption in the euro area since 2000.

In contrast, housing prices provide only relatively limited help in explaining differences in consumption growth within the euro area in the latest downturn. As shown in the right panel of the chart below, the correlation between differences in the growth of residential prices and differences in the performance of consumption is relatively low for the 2001-03 period. Furthermore the observed correlation does not always reveal a causal link via the wealth effect. Given the small size of the wealth effect in that country, the association of fast growth in residential prices and a relatively good consumption performance in Spain probably reflects the existence of common factors that have boosted both residential prices and consumption rather than a positive impact of residential prices on household spending. A similar point can be made for the association of weak growth in consumption and residential prices in Germany. Nevertheless, empirical research would tend to support the idea that housing prices have contributed positively to the resilience of private consumption in countries such as Finland and Ireland.

Graph: House prices and private consumption resilience

Contribution of house prices to consuption growth Euro ares (average annual growth in % 2001-03) in the USA and UK * 2.4 FI

2.4 ES

Observed differ. with EU12 per capita consumption growth - 2001-03 ita 2.0

ap IE 2.0 Wealth effect - elast of 0.05 (1)

Wealth effect - elast. of 0.10 (2) -0

3 1.6

 p er

 c

01 FR 1.6

1.2

ro w

th in

1.2 l g pt

io n

20

0.8 AT(3)

ua um

ns BE

IT

0.8 a

nn

co 0.4

ra ge DE

R 2 = 0.2781

0.4 0.0 A ve PT NL

-0.4

0.0 -5 0 5 10 15

UK US Average annual growth in real residential prices 2001-03

Note: (1) Impact of rise in real house prices assuming an elasticity of 0.05. (2) Impact of rise in real house prices assuming an elasticity of 0.15. (3)

Growth in residential prices calculated over 2001-02. * Range of likely contribution annual average growth in per cent 2001-03.

Source: Commission services.

Overall, although recent empirical research on the link between housing markets and consumption has largely focused on the wealth effect in the broad sense (i.e. the first and second channels described above), the role of housing markets in the transmission of monetary policy (the third channel) should not be overlooked. Housing markets will amplify the impact of changes in monetary policy if the mortgage rates of existing debt are closely linked to base rates either because mortgage contracts are predominantly based on variable interest rates (as in the UK) or can be refinanced at a low cost (as in the USA). The share of variable rate contracts is low and refinancing is generally costly in the euro area. Cuts in base interest rates during the latest downturn have therefore probably had a bigger positive impact on consumption in the UK and the USA than in the euro area. It is also striking to note that mortgage rates are primarily variable in the three euro-area Member States that posted the stronger per capita consumption growth over 2001-03 (Finland, Spain and Ireland).

4.5 Factors behind prices stickiness Graph 38: Output gap and inflation, euro area

In the theoretical vision of a totally flexible economy, relative prices would adjust immediately to changing 2 3 economic circumstances and economic agents would

instantaneously modify economic plans so that )

economic quantities such as production or employment 1 ilt er

would adjust smoothly to the new equilibrium. A P

 f 2

 ( H a.

shortfall of demand would lead to lower prices. An D P 0 . p.

easing of price and cost pressure fosters external d G c hg %

competitiveness, increases real wealth and allows central f tr en 1 banks to cut interest rates. All this would ensure a o % -1

smooth rebound in economic activity. Output gap (HP filter, lhs)

Even abstracting from the vision of a totally flexible -2 Core HICP inflation (rhs) 0

economy, inflation in the euro area declined less than it Q1-99 Q1-00 Q1-01 Q1-02 Q1-03 Q1-04 was expected to do during the recent growth slowdown.

This inhibited the working of the factors just mentioned Source: Commission services.

and might therefore have induced the economic recovery

to start belatedly and sluggishly. This section provides The macroeconomic literature highlighted a number of some evidence on the reasons for stickiness of prices in factors, explaining the sluggish adjustment of nominal

the euro area. prices.

43 Price and wage rigidities imply that it usually

takes time for price and wage levels to fully adjust to a

4.5.1 Measuring price stickiness given macroeconomic shocks. They can arise from

Last year’s review already presented some evidence on • institutional factors such as wage and price

price inertia, showing that the euro-area rate of inflation contracts of fixed durations, existence of staggered has a high degree of correlation with its past values, contracts, wage indexation, the nature of wage which means that it tends to change slowly over time. bargaining; which can all cause persistence in

Moreover, the degree of persistence was measured to be nominal wage developments (Section 4.5.2), highest in service inflation. Since much of the • the presence of so-called menu costs, which signify persistence of core inflation stems from the service the costs of price adjustment, transaction and sector, the lower degree of competition in the service delivery costs. They are thought to be relatively economy, the relative importance of labour costs and more important the lower the rate of inflation and less exposure to external shocks were considered to be the lower the degree of competition (Section 4.5.3),

determinants of actual price stickiness in the euro area. • backward-looking elements in the inflation

Another informative measure of persistence is the time it

takes for inflation to respond to a turning point in GDP expectation formation used by economic agents growth. The picture of a high degree of cyclical (Section 4.5.3).

44

invariance of inflation in the euro area is reinforced by plotting directly the dynamics of the output gap against core inflation in Graph 38. It shows that from the first quarter of 2003 core inflation in the euro area did not change much despite a sharp and continuous widening of the negative output gap. Evidently, GDP needs to grow at a rate below potential for some time before a negative output gap opens. For the euro area this took place in the fourth quarter of 2002, after six quarters of below potential growth. Moreover, inflation started falling while the output gap, albeit narrowing, was still positive. This suggests that it is not only the level, but also the change in the gap, that matters for the cyclical responsiveness of inflation – i.e. speed limit effects

42

might be important.

43 The theoretical controversy does not contest the existence

of price stickiness but centres on the question whether it is compatible with models of microeconomic behaviour.

44 As regards the forward-looking elements of inflation 42 See Lown and Rich (1997), who provide evidence that this expectations, inflation persistence may arise due to a lack of

is the case in the USA. credibility of monetary policy.

Box 3: Inflation persistence in industrialised countries– Some recent empirical findings

Some of the key findings of several recent studies that investigate empirically the issue of inflation persistence in industrialised countries are summarised in this box.

Stock and Watson (1999), in a comprehensive study documenting the business cycle properties of 71 macroeconomic time series in the USA, find that price inflation is pro-cyclical and it lags the business cycle. The usual lag is of two to three quarters. Agresti and Mojon (2003), using data from 1970 onwards find that CPI inflation in the euro area lags GDP by three to four quarters- a slightly slower response than reported by Stock and Watson (1999) for the USA.

A somewhat counterintuitive result that some empirical studies have reported lately is that the level of inflation persistence in the euro area might not be significantly different from that in the USA. One of the most recent such studies is Levin and Piger (2003), who investigate the persistence of inflation in twelve countries, including the USA and the euro-area countries France, Germany, Italy, and the Netherlands. They analyse inflation persistence through autoregressions on headline consumer price inflation, core consumer price inflation, the GDP deflator and the personal consumption expenditure deflator. They find that for the USA all price measures exhibit a high degree of persistency, except headline consumer price inflation. For the four euro area countries covered they find roughly comparable levels of inflation persistence, although there appears to be somewhat more persistency over the four price measures in the Netherlands and Italy than that of France and Germany. The IMF (2003) finds only moderately higher, if at all, inflation persistency in the euro area compared with the US. Relying on univariate methods as well as estimation of Phillips-curves and VARs, this study finds little evidence of structural sources playing a more important role in persistency in the euro area. Differences vis-à-vis the USA are attributed to differences in the formation of expectations. Galí et.al. (2000) is an earlier study which, based on estimations of the so-called New Keynesian Phillips curve, also concludes that there is a substantial degree of stickiness in the euro area which is not significantly out of line with the estimates obtained for the USA using the same model.

Another study pointing in a similar direction is Coenen and Wieland (2002). Using a three country macroeconomic model of the euro area, Japan and the USA, containing Phillips curves which differ in nominal rigidities in wages across the countries, they find higher nominal persistence in the USA than in the euro area. However, this finding is attributed to credibility effects from the FED’s accommodative attitude towards the oil price shocks in the 1970s and the subsequent period of higher inflation, and thus is not thought to reflect higher nominal rigidity than in the euro area.

Some of the above studies have also found evidence of a decline in inflation persistence over time. For example, the IMF (2003) reports that the persistence of inflation is found to be significantly lower since the mid 1990s. The main cause for this change, it is argued, relates not only to a more stable inflation anchor but also closer alignment of actual inflation to the perceived target. Levin and Piger (2003) find empirical evidence of a significant change taking place in the late 1980s or early 1990s. For the four current euro area members (France, Germany, the Netherlands and Italy) and two additional EU-member states (Sweden and the UK), the finding is seen in relation to the Maastricht criteria and the countries prospecting to EMU-membership. However, it seems at this stage that the results are not yet robust to the different methods, specifications and time periods chosen to examine inflation persistence. For instance, Batini (2003) and O’Reilly and Whelan (2004) report results indicating stable inflation persistence over the last three decades.

The above discussion thus confirms the long-held tenet that the level of inflation persistence might be closely related to the behaviour of inflation expectations. On this issue, some authors have lately argued that differences in inflation persistence may also reflect how fast inflation expectations respond to changes in the output gap (e.g. Erceg and Levin (2002), Kozicki and Tinsley (2002), Paloviita (2002, 2003) and IMF (2003)). A main finding is that the expected inflation reacts somewhat faster to changes in the output gap in the USA. Thus, in the most recent down-turn the proposition would be that not only the output gap turned negative earlier and more sharply in the US, but in addition there was a more rapid decline in inflation expectations as the economy weakened, which likely played a role in the faster cyclical response of inflation in the USA.

4.5.2 Wage rigidities Graph 39: Price and labour cost persistence, first-order

In order to evaluate the degree of nominal inertia in autocorrelation of euro-area aggregates 1998Q1-2004Q1

wages, Chapter 4 of the EU E CONOMY 2003 R EVIEW

presented econometric estimates of Phillips-curve type 1

wage equations for euro area Member States. The model 0.9

Value added inflation

Hourly labour costs

used explained changes in wage inflation by the n 0.8

unemployment gap, which could be subject to various tio 0.7 re la

shocks, in particular shocks to labour productivity, terms 0.6

of trade and the wage share. Using the estimated to -c or

0.5

Phillips-curve coefficients, it was found that for the euro o f

au

area as a whole only about 65 per cent of the wage nt 0.4 adjustment to a disinflationary shock takes place within ff

ic ie

oe 0.3 C

the first year, suggesting the existence of substantial 0.2 inertia in nominal wages. 0.1

A striking feature of the empirical estimates is that the 0 degree of nominal wage rigidity found for the euro area Total economy Industry

as a whole does not appear to differ greatly from that of Note: Excluding construction for industry and excluding public

the USA. This finding underscores the difficulty to administration for total economy labour costs, calculated from y-o-y

empirically identify institutional labour market growth rates.

characteristics as major determinants of nominal wage Source: Commission services.

inertia.

Moreover, it is worth noting that establishing unlikely to be the only reason for aggregate price empirically a prominent role for wage indexation persistence. The graph also shows that price and labour regarding consumer price inflation dynamics is not cost persistence is lower in industry than for the straightforward. A number of structural changes, such as economy as a whole, meaning a potentially important a higher policy priority given to nominal stability and a role of structures in the service economy for explaining

heightened degree of competition in product markets, price and wage stickiness.

may have weakened the link between wage and price When the growth in hourly labour costs is plotted dynamics, implying that cost pressures due to temporary against the output gap as in Graph 40, which replicates price shocks (such as oil-price increases) would less Graph 38 but with the change in the hourly labour costs easily feed into consumer price inflation, despite the index instead of core inflation, labour costs

presence of wage indexation. Yet, despite a possibly developments lag the cycle by a considerable time. 47

weaker link between wage and price dynamics, the They continued accelerating until 2002 and started presence of wage indexation mechanisms may have falling only from spring 2002 onwards, about two years important macroeconomic consequences for economies after the peak in the output gap. The acceleration, participating in a monetary union. In particular, to the however, was modest as nominal labour costs growth extent that they contribute to wage rigidity, wage did not exceed 4 per cent. It was also to some extent indexation mechanism would form an obstacle to a driven by the simultaneous reduction in working hours. speedier adjustment of labour markets to asymmetric This is indicated by the more muted growth of nominal shocks. compensation per employee in the national accounts. It

In order to shed some further light on the importance of grew in a narrow range between 2¼ and 2¾ per cent

wage rigidities for price persistence, Graph 39 compares throughout the slowdown.

the degree of persistence, as measured by the coefficient of auto-correlation of the annual rate of change of value

added deflators and hourly labour costs. 45 It shows that

persistence is high for all indicators but higher for prices

than for labour costs. 46 Accordingly, labour costs are

45 The first lag was used for the calculation. Hourly labour

costs can be decomposed into wages and salaries on the one variation of labour productivity distorts the measure,

hand and other labour costs. These are employers’ social reducing further the degree of stickiness of labour costs.

security contributions and taxes net of subsidies related to 47 The relationship with nominal wage growth is more employment. Both components have similar properties as informative than that with unit labour costs because the regards stickiness, very likely because other labour costs are latter’s behaviour over the cyclical is strongly determined partly indexed on wages and salaries. by the pro-cyclicability of labour productivity. Therefore, it

46 A more theoretically consistent comparison would be the reflects more developments in GDP growth than

one of unit labour costs with prices. However, the cyclical developments in wages.

Graph 40: Nominal GDP growth, wages and labour costs Graph 41: Relative price variability, euro area and UK euro area, % annual growth

4.5 2 1 2

4 0.9 1.8 1.5 ) ilt er

3.5 0.8 1.6 P f

1

3 P (

H 0.7 1.4

• a. 

  0.5 G D 0.6 1.2

. p. 2.5 d ie

nt

hg en ff

ic 0.5 1

 c 2 0

% o f tr

C oe 0.4 0.8

1.5

-0.5 p,

 %

0.3 EUR (12, lhs) 0.6

1 Output gap (rhs) t

ga

Hourly labour costs (LCI, lhs) 0.2 UK (12, lhs) 0.4 0.5 -1 O

ut pu Compensation per employee (NA, rhs) 0.1 EUR (90, rhs) 0.2

0 -1.5 0 0

99Q1 00Q1 01Q1 02Q1 03Q1 04Q1 1996 1997 1998 1999 2000 2001 2002 2003

Source: Commission services. need to update output gap Source: Commission services.

4.5.3 Menu costs, competition and price variability weakening of demand in order to generate sufficient

A number of macroeconomic models justify the profits in the short term to remain able to meet their existence of price rigidities by the existence of menu obligations from their liabilities. As euro-area costs. The expression originally relates to the costs of corporations on average faced relatively higher debt in changing the price of menus in restaurants, but has the last years, their price-setting behaviour could have

become used to signify all costs embedded in changing followed this pattern.

50

the price. They can be physical costs but also those Since little is known about the magnitude of menu costs

48

stemming from strategic decisions of enterprises. For in the euro area, Graph 41 presents an indicator that can instance, when establishing long-term customer be regarded as a proxy at the aggregate level. It relationships, the impact of price changes on customers’ measures the variability of relative prices, derived from search costs or on competitors actions may reduce firms’ the monthly changes of the components of the HICP incentives to pass through higher input prices into output where the changes were weighted with the share of the

prices. Apparently, the degree of competition on product goods in the HICP goods basket. 51 The higher the

markets determines the importance of these number of the indicator the more flexible are prices and, considerations. The higher the degree of competition, accordingly, the lower should be the costs of adjusting

the less scope there is for strategic decision-making that prices. The indicator was calculated with 90 nonreduces

price flexibility. subcomponents of the HICP and a second time with 12

Economic research furthermore established some groups of components in order to ensure comparability theories on the cyclical behaviour of prices, conjecturing between the euro-area and the UK indicator. Since the why prices tend to increase less than costs in booms and resulting series is very volatile, the graph shows annual vice versa. Shifts in the composition of demand towards averages. It turns out that the indicator of relative price goods and services where the elasticity of demand is variability was higher until 2001 in the UK, which is lower, less competition as a result of less market entry or widely perceived of having a more flexible economy,

firms’ stronger incentives to collude rather than to fight than in the euro area.

for market shares by cutting prices when economic activity is low, are just three factors that might cause

49

prices to fall less than warranted in a downturn. An additional theory, which is possibly relevant for the experience in the euro area, builds on the pressure of weakening demand on financially constrained firms.

They might be reluctant to cut prices despite a 50 Chevalier and Scharfstein (1996) present evidence that

supports this theory based on the observation that financially restricted supermarket chains in the USA behaved in this way during regional and national recessions.

48 In their empirical analysis of menue costs, Zbaracki et al. 51

   Techncially, the indicator is (2004) distinguish between managerial costs (costs of

gathering information, decision-making) and customer costs RPV = ∑ s * ( p − i i p ) 2 where s i is the share of

(communication to customers, negotiations with i

customers). good i in the basket, p i is the monthly rate of change of

49 See Rotemberg and Woodford (1999). good i and p is the rate of change of the HICP.

Graph 42: Relative variability of hourly labour costs, intermediation).

53 A more detailed breakdown of sectors 4 sectors in 11 euro area Member States is likely to be more informative, but has not yet been

possible with the macroeconomic data available.

0.8

0.7 RWV 4.5.4 Inflation expectations

RWV(adj)

0.6 In theory, inflation expectations play a key role in the

relationship between the output gap and inflation. 54 All

nt 0.5 else equal, sticky expected inflation would intensify the

ff ic

ie

0.4 rigidity of wages and prices, reducing the

C oe 0.3 responsiveness of actual inflation to any given change of

0.2 the output gap.

0.1 The tentative evidence available for the euro area

0 indicates that this is not a likely candidate for explaining

1998 1999 2000 2001 2002 2003 price stickiness. Graph 43 shows that short-term price

expectations derived from the EC consumer surveys

Note: The adjusted series eliminates any bias due to missing observations tend to match actual inflation developments. 55 They

in some countries and sectors.

Source: Commission services. changed more forcefully than actual inflation in some episodes, for instance in 1993 after renewed currency

speculation in the ERM, in 1999/2000 when oil prices

The indicator of relative price variability (RVP) was increased and in 2002 when the full magnitude of initially developed to measure inflation uncertainty. The economic weakness materialised. Using the coefficient economic literature appears divided over whether it is of auto-correlation as a measure of stickiness, stickiness positively or negatively related with the rate of of price expectations in the consumer surveys is inflation. 52 In the sample 1995 to 2004, there is no marginally smaller than stickiness of actual consumer apparent correlation of the RVP with headline consumer price inflation. This finding, however, does not allow price inflation in either the UK or the euro area. This any conclusions to be drawn about the direction of also holds if energy goods are eliminated from both the causality. Inflation expectations might be sticky because panel and the consumer price index. Thus, the increase actual inflation is sticky but could also be a reason for

in price variability observed in Graph 41 is unlikely to price stickiness.

be driven by inflation. It can, however, be shown that the increase in the euro area aggregate in 2002 and 2003 is essentially driven by strong variability of relative prices in January. Items where prices changed strongly in January relative to the rest of the year are health, insurances, waste collection, passenger transport, restaurants, cultural services, but also water, gas and electricity. Overall, it is likely that the increase in the RVP reflects increasing changes in administrative prices rather than a general improvement in price flexibility in the euro area.

Applying the RPV concept to hourly labour costs does not generate a discernible trend in labour cost variability in Graph 42. If at all, relative wage variability declined

in 2002 and 2003. This does not necessarily point to an 53 Sectoral wage growth in each country was expressed

increase in the importance of wage rigidities, but could relative to total wage growth in the euro area and weighted

be due to some correlation of the wage variability with the share of the country’s sector in euro-area indicator with wage growth. Note that the indicator was employment. The data series for the labour cost index starts

constructed with quarterly growth of hourly labour costs in 1996 only, with observations in the first two years being

in 11 euro-area Member States and 4 sectors (industry, quite volatile. 54

construction, trade services, and financial This was forcefully illustrated by the work of Friedman and Phelps in the 1960s, Lucas and Sargent in the 1970s and

more recently by theories underpinning the so-called New Keynesian Phillips curve.

55 For the period since 1991, the coefficient of correlation

between both series is 0.76 for contemporary observations.

52 Banerjee et al. (2002) find a positive relationship for the It rises to 0.79 if one assumes expected inflation to lead

USA and the UK while Nautz and Schaffer (2004) find that actual price developments by 6 months. It is even slightly it is not linked with expected inflation in Germany but to a higher at 0.81 if actual HICP inflation is replaced by core

measure of unexpected inflation only. inflation and if a 13 month lead is assumed.

Graph 43: Actual and expected consumer price inflation, area of budgetary policy, structural reforms and

euro area pension reforms have adversely impacted on

consumer confidence and spending.

6 60 • The lower the costs of adjustment the more

responsive economic activity will be to changed

5 50 economic conditions: in a totally, flexible economy, 4 Actual (lhs) 40 relative prices would adjust instantaneously. When

Survey (rhs) it comes to understand sticky prices in the euro area, a. ce

. p. 3 30 wage rigidities and imperfect competition, in

hg al

an

 c B particular in the service economy, feature high.

% 2 20 They cushion the effects of adverse economic

shocks, but at the expense of slowing the recovery.

1 10

• 5. 

  Economic differences

0 0

Jan 91 Jan 93 Jan 95 Jan 97 Jan 99 Jan 01 Jan 03 The discussion above focuses primarily upon the euro

Source: Commission services. area as a whole. But within an economy as large of that over the euro zone it is to be expected that performance

will differ between different areas. This section takes a

4.6 Conclusions closer look at the performance of individual Member States in recent years and considers in particular whether

The recent slowdown featured a number of factors, the intra-area differences during the latest pick-up, in which were unexpected and made it unique in comparison with both the strength of the upswing in the comparison to previous periods of slow growth. These late 1990s and the weakness of growth in 2001-03, can factors can be expected to shape the dynamism of shed any further light on the causes of the overall slow economic activity over the next years. Both the direction response of the euro-zone to the global upturn. The and magnitude of their impact depends on whether they intention is to extract from cross-country variations to signify the working of structural or cyclical factors. what extent structural differences such as export•

 Macroeconomic simulations with two different reliance or structural rigidities on the one hand and the models showed that structural rigidities have an working of intra-area adjustment mechanisms on the impact on the adjustment to shocks. The difference other hand impact on their cyclical position.

between the flexible and rigid scenario was,

however, too small to suggest that structural 5.1 Measuring differences in timing and strength

rigidities can explain all of the sluggishness in the of the last upturn

rebound of the euro-area economy. They matter in A number of measures can be used to assess the timing particular for the rate of long-term growth. and strength of the pick-up of growth in the euro area

• Resilient employment performance was due to a economies. The most precise means of measuring the combination of cyclical and structural factors, i.e. timing is through examining quarter-on-quarter growth labour hoarding, sectoral change and the usual lag in GDP. Table 6 illustrates that the euro-area recovery with which employment responds to changes in was firmly centred on the third quarter of 2003, with the GDP. More favourable labour market structures majority of countries switching from negative to positive were based on conditions favourable to labour force quarterly growth during this period. A notable exception participation and sectoral change. The assertion that was Spain, where growth remained positive in each constant employment during the period of the quarter of the latest global downturn. Greece and Ireland slowdown would mean weak employment growth in experienced more volatility, albeit essentially positive

the next upswing is not inevitable. growth. The pick-up in Finland was slightly ahead of the euro-zone average, whilst in contrast, the upswing in the

• The prominence of cyclical factors such as a lack of Netherlands lagged by a quarter, without remaining

demand and financial constraints in explaining robust in 2004 and Portugal did not see a significant investment bodes well for future investment growth pick-up until the first quarter of 2004. once demand fully recovers and corporate overindebtedness has been fully worked out. Model simulations, however, suggest that structural rigidities mainly hold back investment activity.

• It is still uncertain whether the weakness of private

consumption encountered in the previous slowdown is of a cyclical or structural nature. It can also not be excluded that public policies, in particular in the

Table 6: GDP growth – quarter on previous quarter say Germany and Austria. The comparison of actual

2003 2004 GDP growth with the Commission’s estimate of potential GDP growth does not reveal any systematic

Q1 Q2 Q3 Q4 Q1 Q2 pattern across Member States, suggesting that both BE 0.1 0 0.6 0.7 0.7 0.7 cyclical and structural forces have been at play.

DE -0.4 -0.2 0.3 0.3 0.4 0.5 An obvious question is whether those countries expected

EL 3.5 -0.5 1.9 -0.3 2.9 -0.6 to have the strongest pick up in growth in 2004 and 2005 were also those which had the strongest growth in the

ES 0.9 0.6 0.6 0.7 0.7 0.5 late 1990s and the shallowest downturn between 2000 FR 0.2 -0.4 0.7 0.5 0.8 0.7 and 2003. Graph 45 shows percentage point changes in

the output gap in euro-area economies in those three

IE --0.1 2.2 -2.4 5.4 0.9 :0.3 periods, ranked according to performance during the

IT -0.2 -0.1 0.4 0 0.5 0.3 downturn.

NL -0.5 -0.7 0.1 0.5 0.8 -0.1 The graph points to three group of countries.

AT 0.5 -0.2 0.2 0.3 0.4 0.9 • Firstly Greece is the only country to have achieved PT 0.4 0 -0.4 -0.1 0.9 1.2 above potential growth in the first two periods

shown, with this trend expected to continue in 2004

FI -0.1 0.8 0.6 0.3 1.1 0.7 and 2005.

EU- • A second group including Germany, Spain, Italy,

12 0 -0.2 0.5 0.4 0.7 0.5 France, Austria and Belgium saw only a moderate

Source: Commission services. cyclical upswing prior to 2001 and a moderate

Graph 44 shows the acceleration of growth in euro-area downswing in the following period.

members during the current pick-up, i.e. between the • A third group includes those countries which

second quarter of 2003 and the second quarter of 2004. enjoyed the strongest growth on this measure in the All Member States shown recorded an upturn, and it is late 1990s upswing, namely Finland, Portugal, the noticeable that it were not the small countries that in Netherlands and Ireland (as well as Luxemburg, not general fared better than the euro-area average. shown), but which also suffered the strongest Whereas the smaller countries grew stronger than the cyclical slowdown from 2001 to 2003. larger ones in the upswing 1996-2000, the Netherlands,

Austria and Portugal performed worse than the euro area The correlation of the changes in output gaps between

average in the early phase of the current rebound. both periods across Member States is minus 0.67 and therewith quite strong. It confirms the assertion that

Quarterly GDP is particularly helpful in pinpointing the countries with strong growth in the late 1990s suffered timing of the upturn, a drawback is that stronger growth more acutely from the slowdown than those with weaker in, for example, Greece and Ireland, in part reflects growth. stronger potential growth in those countries rather than

Graph 44: The up-turn in quarterly GDP growth: average Graph 45: Changes in output gaps growth, 2003Q3-2004Q2

7 10

6 8 1996-2000

2000-2003

. 5 6

 p .a Actual GDP growth D

P 2005-2003

hg 4

 c 4 Potential GDP growth ia

l G

e % te

nt 2

ag 3 po 0

er av

2 p

ts o

f

-2 %

-4 1

-6

0 -8 EL IE ES FI BE FR EUR- AT DE IT PT NL EL ES DE FR EUR- IT BE AT FI PT NL IE 12 12

Source: Commission services. Source: Commission services.

Graph 46: The contribution of external and domestic Graph 47: Contribution from sectors to value added average demand to the cyclical rebound, average 2003Q3-2004Q2 2003Q3-2004Q2

1.2 Domestic demand Net exports 0.9 Construction, agriculture 1 0.8 Industry

0.7 Services

0.8 0.6

0.6

q q 0.5

qo 0.4 qo

 p t p

t 0.4

% 0.2 % 0.3 0 0.2

-0.2 0.1

-0.4 0.0

-0.6 -0.1 EL BE FR ES FI EUR- AT PT DE NL IT FI BE FR ES EUR- DE NL PT AT IT 12 12

Source: Commission services. Source: Commission services.

It is also apparent that the final period for 2004 and 2005 case in the 1995-00 upswing. If we consider GDP appears to some extent a continuation of the preceding growth since mid-2003, there is little correlation with three years, with notably Portugal, the Netherlands and the contribution to growth from net exports. Member Ireland all forecast to see a further negative impact on States with the highest growth had the highest their output gaps. In contrast, those countries expected contribution from domestic demand while external to see above potential output in 2004 and 2005 all demand contributed strongly to growth in those experienced milder cyclical downturns between 2001 countries with subdued growth. This contrasts with a and 2003. It therefore follows that those countries where strong positive correlation between external demand and output was strongest in relation to potential output in the growth for both the previous upswing and the strength of

second half of the 1990s are not the same countries in the aborted recovery in 2002. which we expect to observe above potential growth in

the latest upswing. 56 Graph 47 shows that those euro-area members with the strongest rebound in GDP growth also experienced the

5.2 Can country-specific factors explain growth largest increases in value added in services during the

differences? upswing. Countries where the main driver of growth was industrial activity recorded lower growth in value added.

Within the service economy, it has been in particular

5.2.1 Openness and production structure activity in financial intermediation, real estate, renting

and business activities that contributed strongly to

Small open economies in the euro area performed well differences in growth performances among Member

in the economic boom period 1996-2000. As world trade States.

growth faltered from 2001, those countries most

exposed to world trade saw the lowest growth in The analysis of differences among Member States absolute terms and relative to potential growth. That is, reveals a different perspective of the rebound than the open countries see a more sluggish cyclical rebound euro-area aggregate. Domestic demand and the service than more closed economies, which is somewhat at odds economy are the main factors driving growth

with the image that most analysts have in mind. differences. But domestic demand is weak in the euroarea

 aggregate where growth depends prominently on

Data looking at contributions to growth also supports the external demand. Apparently, the difference is due to the notion that trade performance can explain less of the structure and weight of Germany and Italy in the eurodifference in growth in the latest pick-up than was the area aggregate. Both countries recorded a slow rebound

driven by external demand.

56 Comparing cross-country differences in quarterly growth in 5.3.2 Structural rigidities

the current upswing and the aborted recovery in 2002

provides a further opportunity to consider if there is a Factors such as trade exposure and output structure may

correlation between the countries that pick up most rapidly affect the extent to which different economies are

in upswings. It is indeed the case that those countries which initially affected by a given shock, but factors such as

grew most strongly in this first abortive recovery were employment market regulations and competitive strongly correlated with those that grew strongly in the pressures in product markets will affect the ability of the

emerging recovery from the third quarter of 2003 to the second quarter 2004.

Table 7: Correlations between changes in output gaps

and indicators of structural rigidities (1) Graph 48: HICP inflation 2003, 2004 (forecast)

Government Employment Product 5 efficiency protection market

(IMD 2 ) legislation regulation

(OECD 3 ) (Nicoletti et 4

al. 4 ) 2003

• a. 

  3 2004 (forecast) 1999-2000 0.31 0.60 0.58 . p.

hg

2001-2003 -0.23 -0.58 -0.67 c % 2

2004-2005 -0.26 -0.48 -0.77

1

Notes: (1) A higher value on an indicator is considered more

favourable. 0

(2) IE EL PT ES IT NL FR EU12 BE FI AT DE

  Ranking of quality of business legislation (includes openness, competition regulation, labour regulation, capital market regulation). Based on country rankings (IMD has two separate

rankings for countries with populations greater and smaller than 20 Source: Commission services.

million).

 (3)

 Overall indicators of employment protection legislation (version

2). Late 1990s indicators used for first two periods, 2003 indicator 5.3 The working of adjustment mechanisms

for third period.

Source: Commission services, IMD World Competitiveness

Yearbook 2003, OECD (2004), Nicoletti et. al (1999). 5.3.1 Inflation differences

Within a monetary union with a single monetary policy,

economy to adjust effectively to the shock. Thus, inflation differences between Member States can act as differences in structural rigidities may cause differences an important adjustment channel to differences in

in the magnitude of cyclical variations. cyclical positions, to asymmetric impacts and adjustment to shocks. Graph 48 shows HICP inflation in

Table 7 shows the correlation between three widely used euro-area members for 2003 and the Commission measures of employment and product market flexibility forecast for 2004. HICP inflation in 2003 ranged from and performance of euro-area Member States in the 4.0 per cent in Ireland to 1 per cent in Germany, with three time periods considered. The compilation of such these countries expected to see the strongest acceleration indicators requires value judgements, particularly and deceleration in price rises respectively in 2004. Core regarding the weighting of different factors and one inflation (excluding energy and processed food) was should be cautious not to place too much weight on strongly correlated with HICP inflation and ranged from individual results. Nevertheless, it striking that the three 4.3 in Ireland to 0.9 in Germany in 2003.

indicators all tell a similar story. The ranking of Member States’ inflation performance is

The indicators suggest that those countries with more broadly in line with their growth performance with flexible product and labour markets were more likely to Germany and Austria on the lower end and Ireland and register above trend growth during the late 1990s Greece at the higher end of the scale. The position of upswing. But the correlation is strongly reversed in the relative growth and inflation performance, however, downturn with the more flexible economies more likely differs in Belgium, Italy, Portugal and – in 2003 - in the to register below potential growth. Indicators of Netherlands. There are two immediate consequences of structural rigidities continue to be inversely correlated inflation differentials for Member States, which have the with changes in the output gap in the emerging recovery, potential to alter internal and external balance at the i.e. those countries with more flexible structures, national level, namely heterogeneity in intra-euro area according to the indicators above, are not those that are price and cost competitiveness on the one hand and in projected to experience a quick rebound in 2004-05. real interest rates on the other hand. The effectiveness of Given the previous observation that the USA and other both channels is supposed to be related to country economies judged to suffer much less from structural characteristics: the competitiveness channel is more rigidities have seen more robust upturns, it is surprising important in small open economies and the real interest that there is as yet no sign of the more flexible euro-area rate channel can be more prominent in larger economies economies being better able to capitalise on the latest where prices are stronger influenced by domestic forces upturn. than in small open economies.

Since a Member State no longer has the option of using the nominal exchange rate to help in the adjustment to country specific shocks, adjustment must come through changes in relative prices and wages compared to those

of the rest of the monetary union, that is, through

changes in intra-area price and cost competitiveness. Graph 49: Real effective exchange rate and output gap

This effect is often referred to as the real exchange rate 14

channel. Specifically, persistent above average inflation NL

rates would contribute to a deterioration in intra-area ge 10 R2 = 0.30 PT

price and cost competitiveness, hence depressing an 3

demand and providing relief to inflationary pressures. ch 00 -2 6

e ex 98 ES

tiv 19 IT When looking at the evolution of intra-area price IE

ff ec e ng 2 EL

competitiveness on the basis of developments in relative ha BE

unit labour costs since 1998, four groups of countries re al e

), c -2 FI FR

can be distinguished. 57 A first group is made up of a re

a

ro (

U LC AT

Germany and Austria, who registered considerable gains ra te -6

in cost competitiveness over the first five years of EMU. In tr

aeu

  DE

A second group is composed of Belgium/Luxemburg, -10

France and Finland, where relative cost competitiveness -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 Change in output gap relative to euro area 2001-03

developed close to the euro-area average. A third group contains Greece, Ireland, Italy and Spain, where nominal Source: Commission services. unit labour costs grew significantly faster than in the

euro area as a whole. Finally, a fourth group includes The first-round effects of this channel would depend on Portugal and the Netherlands, two countries where the interest sensitivity of demand and the demand

nominal unit labour costs have grown at a markedly

higher pace than in the euro area as a whole. sensitivity of inflation. As stressed by Ford and Gerson (2001), it would also depend on the perception of

The effect of the real exchange rate may only become economic agents regarding changes in inflation. For significant with a lag of three-four years, Graph 49 example, if changes in inflation were perceived to be shows those Member States that registered larger temporary, then there might be little effect of higher competitiveness losses are also those where the cyclical inflation on ex-ante real interest rates. In addition, with a positions deteriorated the most, indicating that this pro-cyclical behaviour of real interest rates, there is also channel, working in a lagged countercyclical fashion, is a potential for asset prices to exacerbate inflationary helping to resolve inflation differentials due to cyclical pressures. Higher demand growth than potential divergences. In contrast, some of the Member States that combined with lower real interest rates could lead to recorded close-to-average changes or gains in cost higher asset prices, particularly housing prices. In turn, competitiveness also had more modest deteriorations in higher asset prices could induce positive wealth effects,

their (relative) output gaps, suggesting that the real resulting in more prolonged periods of overheating.

exchange rate channel, again working in a stabilising

counter-cyclical fashion, is also helping the recovery in Looking at the period of high growth (1998-2000), Graph 50 shows that those countries that in the upturn

some of the growth-lagging economies. The large had below average real interest rates, also experienced

positive contribution from external demand to growth in

Germany, Austria can be seen a consequence of the the largest relative output gaps.

58 As regards the

working of intra-area adjustment through the real situation at the beginning of the upswing, highest real

exchange rate channel. interest rates were found in Germany, Finland, Austria and Belgium. In two of them, namely Germany and

Since nominal interest rates are set for the euro area as a Austria, domestic demand contributed negatively to whole, higher than average inflation rates in this growth over the year of the rebound. The lowest rates economy would imply lower than average real interest were found in Ireland, which alone had negative longrates.

This would feed excess demand and create further inflationary pressures. That is, at a time when domestic demand would need to be brought in line with potential, national real interest rates would be providing an 58

additional boost in the opposite direction. Real interest Conceptually, real interest rates are derived by adjusting

rates would thus tend to exhibit a potentially nominal interest rates for expected inflation. As the expected inflation is not observable it has to be estimated.

destabilising pro-cyclical behaviour. Several approaches to this estimation are possible. In the

following, for simplicity and due to data availability real interest rates are obtained by adjusting the appropriate nominal interest rates by core HICP inflation (HICP excluding energy and unprocessed food). Other than its simplicity a further rationale for using such a proxy of

57 The preference for REER deflated by unit labour costs of inflation expectations is that, as noted in Section 4.5,

the total economy results from the assumption that wage inflation is typically a persistent process. It can then be growth relative to productivity is likely to be a key channel argued that domestic inflation expectations would tend to of adjustment. Different cost and price competitiveness follow developments in measured inflation relatively

indicators tend to move broadly together. closely.

Graph 50: Real interest rates and output gaps Graph 51: Changes in government cyclically adjusted net lending (1) and output gaps

3.0 2

2.5 EL EL e 2.0 1

la tiv e re 00 1.5 20 R2 = 0.48 4 FI 0 8- 1.0 2 00 FR re st r at

 1 99 DE te 0.5 FR ap IE BE g DE -1 AT

m in AT o ar

ea

0.0 tp

ut

2

ou R = 0.5038 LU IT

rt -t

er

NL -2

ho to e

ur

IT BE -0.5 FI

l s NL ES -3 PT

R ea -1.0

PT ES

-1.5 IE -4

-2.0 -1.0 0.0 1.0 2.0 3.0 4.0 -6 -4 -2 0 2 4

Output gap relative to euro area 1998-2000 Fiscal stance (Change in CAPB 2000-04)

Notes: (1) Based on potential output Source: Commission services. Source: Commission services.

term and short-term real interest rates. Portugal, Italy, pronounced is the difference in the real interest rate. Spain and Greece had also negative short-term interest That is, the smaller is also the probability that countryrates. With the exception of Italy, domestic demand has specific price developments become entrenched in been buoyant in these Member States since summer inflation expectations and the less powerful is the real 2003. interest rate channel that causes deviating growth

An additional concern sometimes mentioned in this performances. Policy measure that increase market context is the possibility of a “hard-landing”. For flexibility therefore not only improve the Member example, a country that has gone through a period of States’ capability to adjust to economic shocks, they overheating and is entering the downward phase of the contribute also to the smooth functioning of EMU.

cycle might have to face a tightening of area-wide A further means of adjustment is through changes in monetary conditions in response to contemporaneous fiscal policy, with Member States making discretionary stronger growth in the less cyclically advanced changes to government revenue and spending plans in countries. The implication is that for some countries the response to changing fiscal conditions. Budget deficits downturn would be more pronounced than if monetary to GDP ratio deteriorated in all except two euro-area policy had been less accommodating earlier on. This Member States between 2000 and 2004 in absolute mechanism might have contributed to subdued growth in terms and in all except four countries in cyclicallythe Netherlands and Portugal, which were cyclically adjusted terms. The exceptions are Spain and Austria in advanced countries in the last upswing and witnessed a absolute terms and these two countries plus Belgium and belated and weak rebound in 2003/04. Portugal in cyclically-adjusted terms. Since changes in

interest rates also affect budgetary figures, the most

widely measure of the fiscal stance is the cyclically

5.3.2 The role of economic policy adjusted deficit excluding interest payments. This Rigid economic structures may have inhibited the measure is also called cyclically-adjusted primary effectiveness of the competitiveness rate channel in balance and indicates the discretionary policy effort.

balancing growth and inflation differences within the

euro area. This channel is more powerful the more Graph 51 shows a negative relationship between the change in the cyclically-adjusted primary balance in the

responsive both costs and prices are to changes in

demand and the more responsive economic activity is to period 2000-2004 and the output gap in 2004, meaning

changes in prices. By simulating an empirical model of that those countries with a more expansionary fiscal stance in 2000-2004 had a smaller output gap in 2004. It

the euro area, Angeloni and Ehrmann (2004) show that is highly significant also for the deterioration in the

stickiness in inflation and output propagates economic

differences across Member States. output gap between 2000 and 2004 and even if the size of the output gap in 2000 is controlled for. Apparently,

Competition and a high share of imports and exports are Greece as the only country with a positive output gap conducive to the effectiveness of variations in price and with the most expansionary fiscal stance in the euro competitiveness as adjustment mechanism. Furthermore, area has a strong impact on the result. However, the more open and integrated an economy is, the more eliminating Greece and Luxembourg from the panel still important are international and euro-area price confirms the insight that more expansionary fiscal policy developments for domestic inflation and thus the less reduces the output gap. In dependence on the panel used,

the coefficient is between 0.5 and 0.6, indicating that

expansionary fiscal policy yields a sizeable impact • Countries with the strongest growth in the euro area

without, however, avoiding a non-negligible benefited from strong domestic demand and revived deterioration of public finances. activity in the service economy whereas growth in

the laggards was mainly driven by external demand

5.4 Conclusions and industrial activity.

It appears difficult to draw together a simple story • Inflation differences are broadly in line with growth

regarding the reasons behind different performance in differences. Adjustment via the real exchange rate euro area economies in the latest upswing. This is in part channel seems to have had only limited impact. due to the relative infancy of the latest upswing – there Low inflation countries recorded a stimulus from is just a full year’s data since growth picked up in the external demand but remained at the lower end of euro area and many of the calculations presented above the growth spectrum. The real interest rate channel are reliant upon forecasts. Nevertheless, a number of might have boosted domestic demand in high tentative conclusions emerge. growth/inflation countries, amplyfing the

• Those countries which performed well in previous differences.

upturns have not grown strongly in the current • The large role for the real interest rate channel

upswing. In the cases of Ireland and Finland such a relative to the exchange rate for growth and conclusion is dependent upon assuming a inflation differences during the last year reflects significant increase in potential output in recent structural rigidities that reduce the information years; both countries having seen actual GDP grow content of relative prices as well as the adjustment

more strongly since the latest upswing began. of both demand and output to price signals.

• Performance in the latest upswing appears to be

more dependent upon performance during the previous trough, with some countries – notably the Netherlands and Portugal still experiencing restricted growth as a consequence of previous overheating.

References

A GRESTI , A.M. AND B. M OJON (2003) “Some stylised facts on the Euro Area business cycle”, ECB Working Paper No 95.

A NGELONI , I. AND M. E HRMANN (2004), “Euro area inflation differentials”, ECB Working Paper No 388.

A TTANASIO , O.P. (1999), “Consumption” in: Taylor, J.B. and M. Woodford (eds.), Handbook of Macroeconomics, Elsevier, Vol. 1b, pp. 741-812.

B ANERJEE , A. P. M IZEN AND B. R USSEL (2002), “The long-run relationship among relative price variability, inflation and the mark-up” European University Institute Discussion Paper No ECO/2002/01.

B ANK FOR I NTERNATIONAL S ETTLEMENT BIS (2004), “74 th Annual Report”, Basle.

B ARSKY , R. AND L. K ILIAN (2001), “Do We Really Know that Oil Caused the Great Stagflation? A Monetary Alternative”, NBER Working Paper, No 8389.

B ARSKY , R. AND L. K ILIAN (2004), ”Oil and the Macroeconomy Since the 1970s”, CEPR Discussion Paper, No 4496.

B ATANINI , N. (2003) “Euro area inflation persistence”, ECB Working Paper No 201.

B AYOUMI , T., L AXTON , D. AND P. P ESENTI (2004), “Benefits and Spillovers of Greater Competition in Europe: A Macroeconomic Assessment”, US Federal Reserve Board International Finance Discussion Paper No. 803.

B ERGOEING , R., N. L OAYZA AND A. R EPETTO (2004), “Slow recoveries”, NBER Working Paper No 10584.

B ERNANKE , B., M. G ERTLER AND M. W ATSON (1997), ” Systematic monetary policy and the effects of oil price shocks”, Brooking Papers on Economic Activity Vol. 1:1997, pp. 91-157.

B OONE , L. AND N. G IROURARD (2002), “The stock market, housing market and consumer behaviour”, OECD Economic Studies No. 35, 2002/2, pp. 175-200.

C ABALLERO , R.C. ET AL . (2004), “Effective labour regulation and microeconomic flexibility”, Cowles Foundation Yales University Discussion Paper No 1480.

C ASE , K., S HILLER , R. AND J. Q UIGLEY (2001), ”Comparing wealth effects: the stock market versus the housing market”, NBER Working Paper No. 8086.

C ATTE , P., G IROUARD , N., P RICE , R. AND C. A NDRÉ (2004), “Housing markets, wealth and the business cycle”, OECD Economic Department Working Paper No. 394, June.

CEPS (2003), „Adjusting to leaner times“, 5th Annual report of the CEPS macroeconomic policy group, Center for European Policy Studies, Brussels.

C HEVALIER , J.A. AND D.S. S CHARFSTEIN (1996), “Capital-market imperfections and countercyclical markups: Theory and evidence”, American Economic Review, Vol. 86, pp. 703-725.

C OENEN , G. AND V. W IELAND (2002), “Inflation dynamics and international linkages: A model of the United States, the Euro Area and Japan”, ECB Working Paper No 18.

D REGER , C. AND H.G. R EIMERS (2003), “Consumption and income in the euro area – Empirical evidence based on panel integration models” and Paper presented at the 14th EC2 Conference, Endogeneity, Instruments and Identification in Econometrics, London, 12 - 13 December 2003, http://cemmap.ifs.org.uk/ec2/dreger.pdf

D REW , A., M. K ENNEDY AND T. S LØK (2004), “Differences in resilience betwen the euro-area and US economies”, OECD Economics Department Working Paper No 382.

E RCEG , C.J. AND A.T. L EVIN (2001), “Imperfect credibility and inflation persistence”, US Board of Governors of the Federal Reserve System Finance and Discussion Paper No 2001-45.

E SCHENBACH F. AND L. S CHUKNECHT (2002), “Asset prices and fiscal balances”, ECB Working Paper No 141.

E UROPEAN C OMMISSION (2004), “Public finances in EMU 2004”, European Economy No 3/2004.

F ISHER , J.D. (2003), “Technology shocks and the business cycle”, US Federal Reserve Bank of Chicago FED letter, No 187, March 2003.

F ORD , R. AND P. G ERSON (2001) “Cyclically Advanced Euro-Area Economies: Consequences and Policy Options”, Selected Issues Paper on the United Kingdom of the Netherlands, IMF Country Report No 01/95, Washington, D.C.

F RANCOIS , P. AND H. L LOYD -E LLIS (2003) "Animal Spirits Through Creative Destruction", American Economic Review, Vol. 93 , pp. 530-550.

G ALÍ , J., M. G ERTLER AND J.D. L ÓEZ -S ALIDO (2003), "The euro area inefficiency gap", Banco de España Documento de Trabajo No. 302.

G ERN , K.-J. (2004), “Moderate upswing in Euroland”, Intitut füer Weltwirtschaft Kiel Discussion Paper No 410.

G OLDMAN S ACHS (2004), “Global supply shocks: Is luck running out?”, Global Economics Weekly, No 4/25. 23 June 2004.

G ROSCHEN , E.L. AND S. P OTTER (2003), “Has structural change contributed to a jobless recovery?” Federal Reserve Bank of New York Current Issues in Economics and Finance, Vol. 9, No 8, pp. 1-7.

H AUSMANN ,R., L. P RITCHETT AND D. R ODRICK (2004), “Growth accelerations”, NBER Working Paper No 10566

H ENLEY , A. AND B. M ORLEY (2001), “European house price volatility and the macroeconomy: the implications for European Monetary Union”, Aberystwyth: University of South Wales, School of Management and Business.

IMF (2003) “Monetary and exchange rate policies of the euro area – Selected issues”, IMF Country Report No 03/298.

IMF (2004a), “Euro area policies”, IMF Country Report No 04/234.

IMF (2004b), “Euro area policies: selected issues”, IMF Country Report No 04/235.

JP M ORGAN C HASE B ANK (2004), “Euro area inflation and supply-side performance”, Global Data Watch, 30 April 2004, pp. 11-12

K OZICKI , S. AND P.A.T INSLEY (2002) “Alternative sources of the lag dynamics of inflation”, US Federal Reserve Bank of Kansas City, Working Paper 02-12.

L EVIN , A.T. AND J.M. P IGER (2003) “Is inflation persistence intrinsic in industrial countries?” revised version of Working Paper 2002-023E, Federal Reserve Bank of Saint Louis.

L OWN , C.S. AND R. R ICH (1997) “Is there an Inflation Puzzle?” Federal Reserve Bank of New York Economic Policy Review, December, p 51-69.

Ludwig, A. and T.M. Sløk (2004), “The relationship between stock prices, house prices and consumption in OECD countries”, Topics in macroeconomics, Volume 4, Issue 1, Article 4.

N AUTZ , D. AND J. S CHAFFER (2004), “Inflation and relative price variability in a low inflation country: Empirical evidence for Germany”, forthcoming in German Economic Review. http://much-magic.wiwi.unifrankfurt.de/professoren/nautz/downloads/papers/nautz_GER.pdf.

N ICOLETTI , G., S CARPETTA , S., B OYLAUD , O. (1999), ‘Summary indicators of product market regulation with an extension to employment protection legislation’, OECD Economic Department Working Paper No 226.

O’R EILLY , G. AND K. W HELAN (2004), “Has Euro-area Inflation Persistence Changed Over Time?”, ECB Working Paper No 335.

OECD (2004) “Euro area”, OECD Economic Surveys 2004.

P ALOVIITA , M. (2002), “Inflation dynamics in the euro area and the role of expectations”, Bank of Finland Discussion Paper No 20.

P EERSMAN , G. (2003) "What Caused the Early Millennium Slowdown? Evidence Based on Vector Autoregressions", CEPR Discussion Paper No. 4087.

P RESCOTT , E.C. (2002), “Prosperity and depression – 2002 Richard T. Ely lecture”, American Economic Review, Vol. 92, pp. XX.

R ATTO , M., W. R ÖGER , J. IN ’ T V ELD AND R. G IRARDI (2004), “An estimated new-Keynesian dynamic stochastic general equilibrium model of the euro area”; European Commission DG ECFIN, Economic Paper, forthcoming.

R ÖGER , W. AND J. I N ’ T V ELD (1997), “Quest II – A mujlti country business cycle growth model”, European Commission DG ECFIN, Economic Paper No 123.

R OMER , D. (2001), “Advanced macroeconomics”, 2 nd edition, McGraw-Hill Irwin.

R OTEMBERG , J.J. AND M. W OODFORD (1999), “The cyclical behaviour of prices and costs”, in: Taylor, J.B. and M. Woodford (eds.), Handbook of Macroeconomics, Elsevier, Vol. 1b, pp. 1051-1135.

S TOCK , J.H. AND M.W. W ATSON (1999), “Business Cycle Fluctuations in US Macroeconomic Time Series”, in J.B Taylor and M. Woodford (eds.) Handbook of Macroeconomics, 1A, Chapter 1, pp.3-64.

T ILLE , C, N. S TOFFELS AND O. G ORBACHEV (2001), “To what extent does productivity drive the dollar?” Federal Reserve Bank of New York Current Issues in Economics and Finance, Vol. 7, No. 8.

W ÄLDE , K. (2004), "Endogenous growth cycles", forthcoming in International Economic Review, http://www.tudresden.de/wwvwliwb/mitarbeiter/waelde/pdf/W%E4ldeIERforthcoming.pdf

W ORLD E CONOMIC F ORUM (2003), “Global competitiveness Report 2003-04”, Oxford University Press.

Z BARACKI , M.J. ET AL . (2004), “Managerial and customer costs of price adjustment: direct evidence from industrial markets”, Review of Economics and Statistics, Vol. 86, pp. 514-533.

• 2. 

  CATCHING-UP, GROWTH AND

CONVERGENCE OF THE NEW MEMBER

STATES

Summary

This chapter attempts to identify the main policy levers – including EU cohesion policy - which could sustain a process of catching-up in the new Member States in order to achieve the EU Treaty objective of economic and social cohesion. The ten economies that acceded to the EU on 1 May 2004 all have income levels below the EU average - some significantly lower - and there are even greater income disparities at the sub-national level, with purchasing power less than half the EU-25 average in many regions of the new Member States.

Policy development must be set in the context of the EU’s past experience, which has shown, firstly, that income convergence is not necessarily a rapid, continuous or automatic process. Secondly, convergence has been faster at the regional than at the country level – partly reflecting the fact that disparities have been higher within than between Member States. Thirdly, regional specialisation and concentration have not changed significantly. Lastly, in the early stages of catching-up, growth tends to strengthen first in agglomerations: thus regional income inequalities within countries may initially increase as the national growth rate accelerates.

Looking at the situation to date in the new Member States, data on sources of growth between 1996 and 2005 show that economic convergence has been driven by investment and total factor productivity (TFP), while underutilisation of labour has acted as a brake. A scenario for 2006 to 2010, based on a broad continuation of recent experience, shows that the contributions of capital and TFP may be expected to moderate somewhat in the future, while labour is likely to make a positive, though limited, contribution to growth. However, these projected growth rates are below 5 per cent, except for the Baltic countries, representing only limited progress in catching up to the EU average.

Existing trends reveal a number of major policy challenges. One important concern is that employment rates are fairly low in most of the new Member States - particularly among older cohorts of the population. It will therefore be especially important to review tax-benefit systems in order to provide incentives to create and take up jobs, and to extend working lives. Labour markets remain relatively inflexible owing to insufficient wage differentiation, the impact of tax-benefit systems, and low regional labour mobility.

Investment has been an important source of growth in the new Member States. Investment-to-GDP ratios are higher than in the EU-15, although production is still less capital-intensive. Given the early liberalisation of capital movements, foreign direct investment has been a major source of current account financing, closing the gap between domestic savings and investment. The heavily foreign-owned banking sector has been the main channel of financial intermediation. An important challenge for the future is to progressively mobilise higher domestic savings through channels such as pension funds and stock markets in order to promote faster, more broad-based growth.

Innovation and knowledge being important triggers for technical progress, it is worth noting that educational attainment levels in the new Member States do not differ much from those in the EU-15. Trade and foreign direct investment have been important for the cross-border transfer of knowledge in management and technology, but innovation has not yet been a central determinant of productivity growth in the new Member States. Activity and employment in R&D and innovation tend to be much lower than in the EU-15, which can best be explained by a different pattern of specialisation. The case for higher expenditure on R&D activities needs to be evaluated critically, given this specialisation, to ensure that it does not divert resources from other uses with higher economic returns.

The new Member States have made great advances in trade liberalisation since the early 1990s, and they have impressively increased trade with the EU, in particular under the Europe Agreements. This expansion of trade no doubt contributed significantly to their growth performance over the past decade. Membership brings some further trade liberalisation in sensitive sectors (agriculture, services) and reduction of non-tariff barriers – as well as a possible further reduction in transport costs as a result of lower waiting times at borders and improvements in infrastructure. Less exchange rate volatility in the case of ERM II participation and the adoption of the euro could reduce costs even further and trigger additional trade and growth.

The new Member States have also made good progress in establishing a stable macroeconomic framework, though those aiming for rapid progress towards euro-area membership will need to entrench this further, as inflation remains somewhat high and variable in some cases. ERM II can provide a framework within which to enhance policy credibility, though the alternative of keeping greater exchange rate flexibility offers more latitude for variations in inflation associated with the challenges of transformation and catching-up – thus helping to avoid a loss of external competitiveness. The majority of the new Member States still have budgetary deficits that are much higher than the 3 per cent benchmark for euro-area membership, although public debt levels are mostly below 60 per cent of GDP; however, fiscal consolidation remains a considerable challenge in the light of the need to build up and modernise infrastructure, reorient public spending, and cushion the costs of ongoing restructuring. To safeguard external and financial stability, attention needs to be paid to the interaction of monetary, prudential and fiscal policy regimes and the ways in which these may influence risk behaviour in the private sector. In particular, as the private sector enters a phase of strong expansion, the design of fiscal policy can play an important supporting role in ensuring that imbalances are limited and that financial market confidence is maintained.

Studies increasingly stress the quality of institutions as an important factor in convergence. Here, despite impressive progress in recent years, the new Member States still have considerable gaps to make up - particularly with regard to efficiency in public administration and the judiciary. Preparation for EU accession provided an external anchor for progress in this area, helping to catalyse political support for change. With the “carrot” of EU membership no longer available, there is a need for reflection on how mechanisms at the EU level might play a stronger role in providing further support for this process.

EU cohesion policy is the final subject considered in this chapter. Despite limited financial resources, this policy could have a substantial impact on catching-up - but only if a number of conditions are met: stronger spatial concentration, improved thematic concentration, and implementation approaches that better safeguard cohesion goals. Spatial concentration means focusing Structural Funds on those regions and Member States most in need - while ensuring that this selection process works with, rather than against, market forces. Thematic concentration means choosing, in each case, an effective investment mix - based on a sound analysis of existing infrastructure endowment, human resource requirements, and limits on aid to the productive sector. Effective implementation requires that the management of Structural Funds be further simplified, and that the new Member States complete the building of necessary administrative capacity. In short, the contribution of EU cohesion policy to real convergence will depend above all on the commitment of policy-makers in Member States to coherent national and regional policies – ensuring that the environment in which Structural Funds are utilised is characterised by macroeconomic stability, continuing structural reforms, and good governance.

In view of the still limited knowledge of economists about the relative importance and detailed interaction of each of the main policy levers, policy can best foster stronger and more broad-based growth through a comprehensive approach addressing all the strongest drivers of economic growth – trade, macroeconomic stability and institutional quality - as well as making efficient use of EU cohesion policy.

TABLE OF CONTENTS

• 1. 

  I NTRODUCTION ..............................................................................................................................................59

• 2. 

  C ATCHING - UP IN THE EU: W HERE DO WE STAND AND WHAT DO WE KNOW ? .................................................60

2.1 The lessons from the past ..................................................................................................................................... 60

2.2 Recent trends in convergence of the new Member States ..................................................................................... 63

2.3 Spatial dimensions of convergence ....................................................................................................................... 65

2.4 Summary .............................................................................................................................................................. 69

• 3. 

  H OW TO ACCELERATE CATCH - UP GROWTH IN THE NEW M EMBER S TATES ? ...................................................70

3.1 The accumulation and diffusion of production factors and knowledge ................................................................ 71

3.2 Other determinants of economic growth ............................................................................................................... 78

• 4. 

  W HAT CAN BE THE CONTRIBUTION FROM EU COHESION POLICY ? .................................................................89

4.1 Evidence of structural funds impact ..................................................................................................................... 89

4.2 Conditions for maximising the impact .................................................................................................................. 92

4.3 Policy challenges ................................................................................................................................................ 100

R EFERENCES ..........................................................................................................................................................102

ANNEX I: M ETHODOLOGICAL CONCEPTS OF CONVERGENCE ...........................................................................109

ANNEX II: S EMI - PARAMETRIC TECHNIQUES .....................................................................................................111

CATCHING UP, GROWTH AND

CONVERGENCE OF THE NEW MEMBER

STATES

• 1. 

  Introduction the policy challenges for the ten new Member States. Apart from the standard determinants of growth –

Income levels in a majority of the ten new Member labour, capital and technical progress – other driving States, which acceded to the EU on 1 May 2004, are forces of growth such as trade and geography, significantly below the average of the former EU-15. macroeconomic stability and institutional quality are Average GDP per capita in the enlarged EU is almost reviewed. Section 4 discusses the potential contribution

10 per cent lower than previously, and inequalities are substantially wider. This makes the objective of

achieving greater economic cohesion and convergence Graph 1: GDP per head in EU Member States, 2003

even more pressing than before. Graph 1, displaying the level of GDP per capita in euro and in Purchasing Power LU Standards (PPS) in the 25 Member States in 2004, IE

shows the considerable disparities between old and new DK AT

Member States, but also among the new Member NL

States. 59 The ranking of Cyprus, Slovenia and Malta is UK

close to that of the “old” cohesion countries (Spain, BE SE

Greece and Portugal). The Czech Republic and Hungary FR have a notably higher GDP per head than Slovakia, FI

Poland and the three Baltic countries. Disparities at sub EUR-15 DE

national, regional level are even larger. GDP per head in IT PPS in many regions of the new Member States is less EU-25

than half of the EU-25 average and the poorest ones ES EUR EL

have even less than a third of the EU-25 average. PPS SI

PT

Given that economic and social cohesion is one of the MT objectives specified in the EU Treaty, this chapter CZ

attempts to identify the main policy levers for a HU SK

sustained process of catching-up in the new Member CY States, based on past experience of real convergence in EE

the EU as well as on evidence from the broader LT PL

economic literature. Relevant developments in both the LV EU-15 and the EU-25 are described in Section 2.

Section 3 reviews potential determinants of catching-up, 0 10 20 30 40 50 60

and analyses the empirical evidence in the EU as well as 1000

Source: Commission services.

59 Due to higher costs of living, income expressed in euro is

higher than that expressed in PPS in most Member States above EU-25 average; the opposite holds for those below average.

of EU cohesion policy, the goal of which is to enhance This performance in Ireland went hand-in-hand with the growth and employment in lagging Member States and implementation of stability-oriented macroeconomic regions. policies, and a new approach to industrial relations - which was also initiated in the mid-1980s. However, Ireland’s success cannot be attributed to these factors

• 2. 

  Catching-up in the EU: Where do we alone, but was also the result of a variety of mutually

stand and what do we know? reinforcing policies, some of which had been pursued for more than 40 years under a pro-active strategy to

This section provides an overview of recent trends in foster economic development. Worth noting are the catching-up and convergence among countries in the continuity and predictability over this long period of the EU, at both national and regional levels. It focuses policy approaches to attracting FDI and promoting particularly on the EU cohesion countries - which clusters of export-led manufacturing and services include Spain, Portugal and Greece, as well as the 10 activities. Highly important, too, were the investments new Member States. 60 In addition, some relevant lessons made in education from the mid-1960s, which translated are drawn from wider experience in the EU. The into labour productivity gains in the late 1980s and analysis is based on a qualitative assessment of key 1990s. The evolution of Ireland illustrates that trends, as well as on econometric evidence; and the convergence is a process having deep roots in a range of

experience of the new Member States during the past policy areas which may take time to bear fruit.

decade is also specifically reviewed.

Graph 2: Evolution of GDP per head in PPS to the EU-15

2.1 The lessons from the past average (EU-15 = 100)

Experience suggests that convergence and catching-up are not automatic outcomes of accession to the EU. 140 Graph 2 provides evidence for the former four cohesion countries. It displays their level of GDP per capita, 120 Greece

measured in terms of Purchasing Power Standards Spain

(PPS), during the period 1960-2003. 61 Ireland 100

Portugal

Ireland, now often cited as a success story, is a particularly interesting case. In 1960, it had a level of 80 GDP per head of about 67 per cent of EU-15 average. Whereas notably during the 1960s and early 1970s the 60 other three economies experienced rapid expansion, the

Irish relative position in terms of per capita GDP per 40

headmore or less stagnated until the mid-1980s when the 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000

Irish economy truly took off. Since then the country went on to become, by 2003, one of the richest Member Source: Commission services. States with a GDP per capita nearly twice as high as Portugal. Furthermore, the experience of these countries suggests that catching-up does not necessarily occur at a steady pace. Table 1 below provides additional evidence by reporting the 10-year average annual rate of catch-up for these countries, between 1960 and 2003. This indicator measures the average percentage change in the gap between each country’s GDP per capita and the EU-15 average.

60 Since 1 st of January 2004 Ireland is no longer eligible to the

Cohesion Fund given the level of its Gross National Income (GNI) per head compared to the EU average and therefore no longer included in the group of so-called “cohesion countries”.

61 Given that convergence refers to a long-term process, a

sufficiently long period (1960-2003) is considered here while acknowledging the fact that this does not necessarily correspond to the accession dates of the cohesion countries, i.e. 1973 for Ireland, 1981 in the case of Greece and Portugal and Spain joined in 1986. Also, it should be noted that intertemporal comparison of PPS figures is limited for methodological reasons. These inconsistencies have been partly corrected in the data used here; see Eurostat (2002).

Table 1: Average catch-up rate for Spain, Greece, Table 2: Test of β convergence for the EU

Portugal and Ireland, 1960-2003 (%) 1 Country-level results (15 countries, period: 1960-2003)

Period Spain Greece Portugal Ireland OLS Fixed-effects

1960- -4.40 -5.94 -2.23 0.53

1970 5-year 10-year 5-year 10-year

1971- 0.55 -2.34 -1.05 -1.20 -0.021 -0.023 -0.024 -0.027

1980 (0.002) (0.003) (0.003) (0.003)

1981- -1.41 6.13 -1.49 -2.85 1990 Region-level results (187 regions, period: 1980-1996)

1990- -2.84 -1.14 -0.40 -19.75 2 2003 1-year 5-year 1-year 5-year

1 A negative catch-up rate indicates that the gap between a country -0.04 -0.043 -0.062 -0.046

and the EU average is falling while a positive rate means that this (0.002) (0.002) (0.002) (0.002)

gap is widening. Catch-up rate = 100 * Δ ( y − it y t * ) ( where y it

y it − − 1 y t − 1 * ) Note: Standard errors reported in parentheses.

is the level of index of GDP per head in PPS terms for country i at time t and y t * is the average value of y t for the EU-15 and Δ

denotes absolute variation between t and t-1 with y t * being the among and within EU Member States – an experience

weighted average for the EU-15.

2 which emerges as highly relevant, but potentially Average annual catch-up rate up to 1996, given that, after this

date, Irish GDP per head became higher than the EU average, see worrying, for the new Member States. also footnote 62. First, a common tool used in the literature is the

estimation of so-called β-convergence: this provides

The first observation that emerges is that catching-up indications how long, on average, convergence may take has been rather uneven across different periods. Overall, (see Annex I for a description of this methodology). the 1960s were years of rapid catch-up for all these Table 2 presents results of an estimation (of equation (i) countries except Ireland. For other periods, however, the described in Annex I) on β-convergence for the EU-15 evidence is more mixed across these countries. In countries for the period 1960 to 2003, and also for particular, if one looks more closely at the decade during regions within those countries for the period 1980 to

which these three countries acceded to the EU, i.e. the 1996, using the NUTS2 desegregation level.

63 The

1980s, it appears that catch-up was rather slow for average annual convergence rate is estimated using 5- Portugal and Spain - with the gap between these year and 10-year intervals, respectively. countries’ GDP per head and the EU average level The evidence in Table 2, based on least square falling at an average rate of only some 1 per cent per estimators (OLS), is that convergence has been taking year. The evolution was even less favourable for Greece, place rather steadily across the EU. The rate of where there was a rather sizable in terms of GDP per convergence varies between 2.1 per cent and 2.3 per head gap in the 1980s. Over the most recent period 1990 cent in absolute terms, when using country-level data. to 2003, Spain, Greece and Ireland experienced a This is rather similar to the β-convergence found by a

narrowing of GDP per head gaps but at markedly

62 number of authors in the economic literature.

64 Results

differing speeds. at the regional level show a β convergence rate that is While these results provide a first impression of past EU markedly higher: between 4.0 per cent and 4.3 per

experience, a more rigorous analysis is needed to cent.

65 The implied time to halve per capita GDP gaps determine whether, over the long run, convergence has vis-à-vis the EU average varies between 30 and 33 years indeed been taking place and whether it has been at the country level and between 16 and 17 years at the

significant. Several different approaches are available to region level.

assess this formally, and they are followed in turn below. The findings are quite complex and might even seem inconsistent, but on careful inspection they shed

rather valuable light on the experience of convergence 63 The country-level data is taken from Ameco (ECFIN)

database while the region-level data is from the Regio database (Eurostat). Note also that the regional data is available under two different classifications (ESA79 and ESA95) and cover different periods (1979-1996 under

62 Note that, strictly, the catch-up rate and the convergence ESA79 and 1995-2001 under ESA95). Data concerning the rate are not identical concepts. Both processes are year 1979 are rather incomplete, so the period 1980-1996 is

characterized by a negative sign. But their evolution need considered instead when using the ESA79 data. 64

not be the same. Catch-up is concerned with the distance See Magrini (2004) for a review. left to travel, and convergence addresses the pace of 65 Note that the higher convergence rate found when using

advance. Thus, for any given rate of growth that shrinks the region-level data may be partly due to the fact that the time gap, the rate of catch-up will be higher for narrow residual period is different from the one used at country-level and gaps, while the convergence rate will be correspondingly also to the fact that income disparities at regional level in

lower. the EU are wider than the ones at country-level.

These results provide a first indication that convergence Table 3: Test of σ- convergence in the EU, is indeed taking place, and how long it may take to run 1982-1996

its course. Although it appears faster among regions than

among countries, these results must be treated with Year 1982 1988 1996 % %

caution for at least two reasons. First, as discussed annual annual

above, experience shows that the pace of convergence change change 82-88 88-96-

may vary greatly across countries and time periods.

Second, as the literature on β-convergence points out, Country-level results

least square estimators are likely to be biased since they Gini 0.1337 0.1284 0.0977 -0.66 -2.99

do not control for time-invariant features that are Theil 0.0320 0.0291 0.0174 -1.51 -5.03

country- or region-specific. In its simple OLS form, one coef. 0.0303 0.0276 0.0167 -1.49 -4.94

implicitly assumes that all countries converge to the var* same steady state. In order to relax this hypothesis, a

“fixed-effect panel estimator” can be used instead to Region-level results

take account of unidentified country-specific or region Gini 0.2127 0.2115 0.2037 -0.09 -0.46

specific features. 66 Columns 3 and 4 of Table 2 provide Theil 0.0720 0.0704 0.0652 -0.37 -0.92

such estimators for the EU countries and regions. The coef. 0.0703 0.0677 0.0656 -0.62 -0.39 fixed-effect estimators obtained are only slightly larger var* than the OLS ones when considering country-level

results, but when using region-level data the difference Note: Concerns regions NUTS2 of Belgium, Germany, Spain, France, Italy, Netherlands, Greece, Portugal. * Half of the square of the

appears to be more substantial. On this approach, the coefficient of variation. estimated convergence rate oscillates between 2.4 per

cent and 2.7 per cent at the country-level and 4.6 per in the variation around the mean GDP per head. Table 3

cent and 6.2 per cent at the regional level. 67 Again, provides evidence using three indicators generally used

convergence is present, and appears to be generally in the convergence literature: the Gini index, the Theil stronger among regions. index and the square of the coefficient of variation. 68

As mentioned earlier, evidence of β-convergence among The results depicted in Table 3 show, rather countries, and across regions EU-wide, does not unsurprisingly, that inequalities are larger between EU necessarily mean that disparities in GDP per head within regions than between countries. More importantly, these the EU are falling, see Annex I. In order to get a more results show that inequalities have tended to decrease complete picture of the convergence process it is over the period considered, i.e. from 1982 to 1996, with necessary to analyse the evolution of GDP per head an accentuated fall from 1988 onward. Interestingly, disparities as σ-convergence which measures the change while the same result holds for both country-level and

region-level data, the average annual fall in inequalities seems to be higher for countries than for regions, as shown by the last two columns of Table 3.

This suggests that, while some convergence took place, it was more pronounced at the country level than at the regional level. Although such evidence seems to be at odds with the above β-convergence analysis, this needs

66 See Islam (1995). Other authors have criticized the not to be the case. The estimated β-convergence results

regression approach to convergence on the ground that this at the regional level show that the average convergence method provides no information on the dynamics of the rate was well above 2 per cent: individual regions thus entire cross-sectional distribution of regional income and had very different experiences, explaining in turn the

have proposed alternative methods based on non-parametric results obtained for the σ-convergence. 69 A number of

statistical techniques which allow considering the existence economists have also suggested that region-level and

of “convergence clubs” where countries and regions

converge to different steady states; see for instance, Quah country-level convergence have not followed the same

(1996) and 1997) and Durlauf/Quah (2002) for a review. rhythm in the EU over the past decades. In particular,

67 Note that the fact that fixed-effects estimators of β- Esteban (1999) and Duro (2001) show that, while GDP convergence display larger estimates in absolute terms is a per head dispersion between EU countries has decreased

well-known fact in the literature suggesting that the bias of OLS estimators is downward. However, these estimators are

more sensitive to the sample of countries or regions 68 considered as well as to the time-length of each time-series; Not all EU-15 countries are considered in this table since see Tondl (2001). For instance, the estimates found here are regional data were not available for all years and all rather lower than the ones generally found in convergence countries. The results thus only concern Belgium, Germany, studies. Islam (1995) finds a rate of 9 per cent for a sample Spain, France, Italy, Netherlands, Greece and Portugal. of OECD countries, Canova/Marcet (1995) find a rate of 23 Also, for the same reason, only the years 1982, 1988 and per cent for EU regions and Tondl (1997) a rate of 20 per 1996 are considered. cent for EU regions. 69 See Chatterji (1992).

the former EU-15. It may be that – as convergence Table 4: Decomposition of σ- convergence: within proceeds at the country level – these internal disparities

vs. between countries components, 1982-96 could become yet wider, at least on a temporary basis.

1982 1988 1996 % %

annual annual 2.2 Recent trends in convergence of the new

change change

82-88 88-96 Member States

Theil index Analysis of convergence developments in the new Member States is constrained by the fact that the time

Between 0.0494 0.0464 0.0396 -1.01 -1.86

country series for GDP per capita are available only for a short time span - in general, since the beginning of the

Within 0.0225 0.0240 0.0257 1.09 0.89 1990s. 73 This poses a major problem for estimating β-

country convergence, for example, since this requires time series

Coefficient of variation over a much longer period. The consequence is that no

Between 0.0450 0.0410 0.0372 -1.45 -1.18 proper econometric tests can be carried out. country Nonetheless, apparent patterns in the available data do

Within 0.0253 0.0266 0.0284 0.86 0.82 suggest some interesting insights. Graph 3 displays the

country relative level of per capita GDP for Greece, Portugal,

Spain and the 10 new Member States, individually, Note: Concerns regions NUTS2 of Belgium, Germany, Spain, France, compared to the EU-25 average for the years 1991, 1997

Italy, Netherlands, Greece and Portugal. and 2003. The figure also shows how the weighted

during the 1980s and the 1990s, inequalities between average of GDP per capita for these respective country regions within the same country have tended to groups – the three existing cohesion countries and the

increase. 70 In order to see this, the Theil index as well as new Member States evolved.

74 On average, the relative

the coefficient of variation for EU regions can be level of GDP per head of both groups rises over the decomposed into within and between countries’ period. In 1991 the level of GDP per head of the three

71

variations. The results of such a decomposition are cohesion countries amounted to 84.2 per cent of the EU-

reported in Table 4. 25 average, while by 2003 it had risen to 90.3 per cent. For the group of new Member States, the increase is

According to these results, the slight decrease in even more pronounced in relative terms - advancing regional inequalities observed in the EU between 1982 from 42.3 per cent to 53.3 per cent of the EU-25

and 1996 masks in fact two opposite shifts: inequalities average.

between countries have tended to decrease, while

inequalities within countries have tended to increase. Graph 3: Evolution of GDP per capita in Greece, Portugal,

The overall picture for the EU noted above – one in Spain and the new Member States, GDP per capita in PPS

which there is a general fall in regional inequalities – 1991-2003

thus reflects the dominance of favourable changes across

countries over adverse changes within countries. ES + PT + EL 100

A number of authors have offered potential explanations ES 00 ES ES for this phenomenon. The main one put forward in the 1 CY = literature is that economic integration, which advanced ita 80 CY

CY EL

EL PT SL ap PT EL PT

quite strongly during the period considered here, may c MT CZ SL MT CZ Ne w M e m be r CZ benefit mainly a limited number of regions, at least p

er SL States

MT

D P 60

initially. These would include, notably, the most LT

HU LV

dynamic and innovative regions in each country – those -2

5 G HU HU PL SK

SK LT PL EE

that are also best placed to benefit from potential E

U R 40 LV

SK EE PL LT

EE

externalities within the EU economy as a whole. 72 The LV resulting pattern would be that convergence increases at 20 the country level, but that it is in practice driven mainly 1991 1993 1995 1997 1999 2001 2003

by a few regions. Within countries, by contrast, levels of GDP per head could well tend to diverge. (Section 2.3 Source: Commission services. will consider these issues in more detail.) Such a conclusion would be of clear relevance to the new Member States, where GDP per head disparities within countries typically are at present more marked than in

73

70 Even for that period, data are only fully comparable Duro’s (2001) result is reported by Puga (2002).

between 1995 and 2003 because a revision of purchasing

71 For the description of such decomposition, see Annex I. power standards (PPS) before 1995 has not yet been made. 72 See Giannetti (2002). 74 Total population is used as weight.

Table 5: Average annual % catch-up rate for the Table 6: Test of σ- convergence in the EU, new Member States, 1991-2003 1995-2001

1991- 1995- 1999-2003 1991-2003 % chg. % chg. 94 98 Year 1995 1998 2001 1995- 1998-

new MS 1.84 -1.74 -2.07 -1.01 1998* 2001*

Cyprus -6.34 0.57 -2.87 -2.59 Country-level results

Czech Rep. 1.04 0.71 -1.29 -0.04 Gini 0.177 0.165 0.160 -1.54 -0.95

Estonia 0.62 -2.44 -2.48 -1.90

Hungary 0.88 -0.86 -2.73 -1.21 Theil 0.055 0.050 0.047 -3.08 -1.96

Lithuania 16.00 -2.56 -2.51 2.10 coef. 0.050 0.045 0.043 -2.80 -1.92

Latvia 14.84 -1.21 -2.88 2.11 var

Malta -5.18 -3.36 0.76 -2.10 Region-level results Poland -1.53 -2.55 -1.05 -1.67 Gini 0.284 0.259 0.248 -2.95 -1.45

Slovakia -2.33 -2.08 -1.29 -1.81 Theil 0.143 0.124 0.112 -4.49 -3.11

Slovenia 0.36 -3.64 -4.38 -2.95 coef. 0.129 0.112 0.105 -4.58 -1.96

ES+EL+PT 3.37 -2.82 -2.98 -1.34 var

Spain 3.33 -6.12 -6.20 -3.79 Note: Including regions NUTS2 of France, Italy, Germany,

Greece 3.74 1.38 -5.34 -0.83 Netherlands, Portugal, Spain, Greece, Austria, Italy, United Kingdom,

Portugal 3.04 -3.73 2.59 0.59 Belgium, Sweden, Slovakia, Hungary, the Czech Republic and Poland.

• * 

  percentage annual change.

Source: Commission services. Source: Commission services.

The overall evolution seems rather favourable, however, Cyprus, Malta, Poland and Slovakia experienced with some differences both across time and countries. catching-up during the years 1991-1994, while the rest The years between 1991 and 1994 represent a period of of the countries experienced a less favourable evolution

relatively slow catching-up which can be explained by

the economic downturn of the early 1990s and by the over that period due to transition crises. In particular,

75 Lithuania and Latvia saw their GDP per capita drop on

transition process in new Member States. Some average by 16 and 15 percentage points, respectively, differences also appear between countries which do not compared to the EU-25 level, reflecting the deep impact

necessarily correspond to the distinction between

cohesion countries and new Member States. For of transition. Following this mixed picture, the years

instance, countries such as Spain, Cyprus, Slovakia and after 1994 are marked by a general tendency for most countries to catch-up toward average EU GDP per capita

Slovenia experienced steady catching-up, while other levels.

countries such as Lithuania and Latvia - and also, to

some extent, the Czech Republic and Portugal - While a β-convergence analysis cannot be undertaken experienced uneven developments. because of a too short data time series, some results can

still be obtained for σ-convergence although the results

In order to shed more light on how fast countries must be considered with caution for the same reason.

actually caught up towards average EU income during

the past decade, Table 5 reports the average annual Table 6 shows the results for all EU-15 members except Ireland, Denmark and Luxembourg (for which regional

catch-up rate of the new Member States together with data were not available at the NUTS2 level) but, in Greece, Portugal and Spain, using the EU-25 average as addition, Poland, the Czech Republic, Slovakia and

benchmark. 76 Overall, Spain has experienced the fastest

catching-up, with an average annual rate of convergence Hungary.

77

of -4 per cent. Other countries such as Cyprus, Estonia, Malta, Slovenia, Poland and Slovakia have displayed average catch-up rates of around -2 per cent. Again, the timing differs across these economies. Countries such as

75 During the period 1992-1994 the average growth rate of the

new Member states was equal to -0.08 per cent, 0.76 per cent for Spain, 0.01 per cent for Portugal, 0.37 per cent for Greece and 1.58 per cent for the rest of the EU.

76 Note that the differences in the catch-up rates between the first column of Table 5 and the last row of Table 3 are due 77 Other new Member States did not have regional data on an

to the different reference group considered which is the EU- annual basis for the period considered while others, such as 25 average in the first case and the EU-15 average in the Estonia, Cyprus, Latvia, Lithuania and Malta have no

second case. NUTS2 breakdown.

Table 7: Decomposition of σ- convergence: within tendency that regions within countries have, initially at vs. between countries components, 1995-2001 least, diverged rather than converged which reflects the

% chg. % chg. strong performance of the more dynamic regions in a

1995 1998 2001 1995- 1998- country.

1998 2001

Theil index 2.3 Spatial dimensions of convergence

Between The economic literature suggests two potential tradecountries

0.117 0.095 0.082 -6.12 -4.75 offs that may explain why convergence is not even Within across countries and regions. The first is that countries countries 0.027 0.029 0.031 2.65 2.36 and regions differ in their initial potential to benefit from

Coefficient of variation any given increase in integration as some may be more attractive for the location of economic activities than

Between others. The second is that, over long periods of

countries 0.095 0.075 0.066 -6.98 -4.13 integration, regions within countries may develop along

Within different paths. In particular, for countries starting from

countries 0.034 0.036 0.039 2.12 2.54 relatively low levels of income, fast national growth

Note: Includes NUTS2 regions of France, Italy, Germany, may entail rising regional inequalities given that

Netherlands, Portugal, spain, Greece, Austria, Italy, United Kingdom, economic development is rather localised around a Belgium, Sweden, Slovakia, Hungary, the Czech Republic and Poland. limited number of growth poles. In practice, both of

these effects interact and determine the way the benefits As expected, inequalities are significantly larger when of economic integration spread across regions. These including the new Member States. The results also tend issues are considered in more detail below. to confirm the developments noted in the earlier discussion relating to the 1982-96 period. In particular,

the average annual variation of the three measures of 2.3.1 The location of economic activities in the EU

convergence shows that in all cases GDP per head The question of the potential impact of economic disparities in the EU have narrowed. This result holds at integration on the location of economic activities has both country-level and region-level, although it is less generated a sizeable amount of literature over the past pronounced when considering country-level results for decade. In particular, researchers have largely used the the period 1998-2001. Furthermore, the pace of framework of New Economic Geography (NEG) to catching-up seems to have increased compared to the draw possible conclusions about the impact of EU earlier period, especially at regional level, although the integration on the location of economic activities and, starting level of regional inequalities is also much ultimately, the relative wealth of the countries and higher. regions concerned. 78 A frequent general interpretation is

Table 7 indicates that the decrease in regional that economic integration may, at least initially, improve inequalities is essentially due to a fall in betweenthe competitiveness of core EU regions more rapidly country inequalities, as was found in the earlier analysis. than peripheral areas - thus deepening income In turn, within-country inequalities have increased at inequalities throughout the EU.

79 Accordingly, the

rates varying between 2.4 per cent and 2.6 per cent a relationship between economic integration and the year depending on the indicator used. This result thus spatial distribution of activity would be non-monotonic: tends to reinforce the findings observed for the EU-15: as trade costs decline, agglomeration initially increases - while some convergence can be observed at the country but subsequently it begins to decline, provided trade level and regional level for the EU-25 as a whole, there costs fall to a sufficient degree.

80

has been a rise in regional inequalities within countries. Using this theoretical background, empirical studies on

In sum, experience suggests that the road to convergence the EU have considered how the spatial distribution of

is far from an easy one. First, over the long run, some convergence has been taking place in the EU, but this

process was rather slow. Econometric results show that 78 This literature has provided extensive discussion of the the rate of convergence was just under 2 per cent over importance of elements such as market size, economic the past decade - meaning that it may take around 30 linkages, imperfect competition and returns to scale in

years, on average, to halve any GDP per capita gap visdetermining the geographic location of economic activities.

à-vis the EU average. Second, the pace of catching-up See Krugman (1991), Krugman and Venables (1996) and Duranton and Puga (2004).

has varied a good deal across countries and time periods. 79

Third, the experience of former cohesion countries See Combes and Overman (2004). 80

underscores that accession does not automatically Martin and Ottaviano (1999) and Baldwin, Martin and

trigger rapid catching-up. Fourth, evidence at the Ottaviano (2001) have built economic geography models with endogenous growth to show that the interactions

regional level is complex. Convergence periods appear, between agglomeration and growth are also likely to be

at first glance, shorter for regions than for countries, influenced by the decrease in transport costs and act as an based on EU-wide developments. But this masks a additional force in favour of agglomeration.

economic activities evolved during the 1980s and the A number of recent studies have also analysed the case

81

1990s. The evidence in these studies presents a mixed of the new Member States and the candidate countries picture. Studies using value added and employment data during the 1990s, although available evidence is still show that specialisation increased, but that this scarce. Landesmann (2003) analyses the trade structure

82

development was very slow. In turn, studies using of manufacturing sectors in these countries and shows trade data tend to show that export specialisation has that specialisation in some of them changed significantly

83

slightly increased in the EU over similar time spans. during the last decade, and was characterised by a rise in

By contrast, studies using regional data tend to find technology-intensive branches. This was particularly stable or slightly decreasing specialisation during recent true for countries such as Hungary, the Czech Republic, decades. 84 Molle’s (1997) study is noteworthy in this Slovakia, Estonia and Poland. By contrast, Bulgaria, respect as it provides the longest time analysis – based Romania and Lithuania remained strongly specialised in on industry/region-level data for every 10 years between traditional, low-technology sectors. Traistaru et al. 1950 and 1990 – and thus includes years of strong and (2002) instead use employment data for a number of rapid economic integration. In addition, Molle includes countries at NUTS3 regional level and present rather service sectors, for which the determinant of mixed results. They find an overall increase in regional geographical location may arguably be different. specialisation for Bulgaria and Romania, but Overall, Molle’s results show no strong changes in the specialisation seems to have decreased in Estonia, and EU, although a minority of regions experienced a no significant changes occurred in Hungary and decline in specialization, rather than the rise predicted Slovenia. Finally, von Schütz/Stierle (2003) use gross by the core-periphery hypothesis. In addition, Molle value added data at the regional/sector level to study the shows that the service sector tends to be relatively more evolution of specialisation patterns in most old and new dispersed than manufacturing. Further evidence, also Member States, as well as candidate countries, during using region/sector level gross value added data, the period 1995-2000. They show that, while these similarly shows that the service sector is likely to favour countries appear to differ widely in terms of the dispersion rather than concentration, given that firms in structure of their productive activity, no strong changes this sector need to be geographically close to their can be observed – a finding that probably reflects the

respective market. 85 short time span considered. Empirical studies using sector/spatial concentration Summing up, most studies come to the conclusion that measures across EU countries and regions also provide the impact of European integration on regional mixed evidence. Studies at the country level show again specialisation and sectoral or spatial concentration has that results depend on the sectors being considered. been rather insignificant during the past decades. The Labour–intensive sectors display a tendency to locate lack of strong shifts in the location of economic preferably in southern EU countries, while sectors with activities during economic integration in the EU high technology intensity and economies of scale, and probably reflects specific features of the European which depend on strong backward and forward linkages, economy - especially low labour mobility. If workers do

86

remain highly concentrated. However, these studies not move according to wage differentials, then wage find that changes in location patterns during the 1980s inequalities will persist and act as a dispersion force by

and the 1990s have been, at most, very slow. increasing production costs for firms active in relatively dense areas. 87 Another possible explanation is that, over

the past decades, the service sector has become increasingly important, and is also known to be less

81 Note that a number of studies, in particular studies based on footloose than manufacturing. Because of the absence of

micro-level data have considered more closely the spatial

distribution of economic activities by considering only one labour mobility, the service sector is also less EU country as, for instance, Maurel and Sédillot (1999) and concentrated geographically which exerts another strong

Devereux et al. (2003). As these studies do not consider the dispersion force.

88

potential impact of EU economic integration, they are not The evidence reviewed so far thus provides little support

reviewed here. for a “spatial trade-off” in which deeper economic 82 See Amiti (1999), Midelfart-Knarvik et al. (2002), integration is associated with greater agglomeration. Aiginger/Davies (2000) and WIFO (1999). A different However, the methodological and conceptual limitations

picture arises for trade specialisation measured by import or

export data (Midelfart-Knavrik et al. 2002) or by export noted above call for caution when interpreting these surplus (WIFO 1999). Here, overall national specialisation results, especially when considering possible scenarios

decreased between 1970 and 1988. This result may be due for the future.

to increased intra-industry-trade leading to similar trade structures.

83 See Midelfart-Knarvik et al. (2002), WIFO (1999) and

Sapir (1996).

84 See OECD (1999), Hallet (2000) and Molle (1997).

85 See Combes and Overman (2004). 87 See Puga (1999) for a theoretical analysis.

86 See Brülhart (1998) and Midelfart-Knarvik et al. (2002). 88 See Barrios and Strobl (2004a).

2.3.2 National economic development and regional • Technological and knowledge-related spillovers,

inequalities in the EU which are essential for economic growth, are likely

Before considering evidence for the second trade-off, to be geographically bounded.

namely the “national growth/regional disparities” effect, • Despite the fact that increased economic integration

this section discusses how far location influences tends to lower the barriers to technological technological diffusion as a vehicle for growth. This spillovers, the diffusion of knowledge and

issue is rather important for the national growth/regional innovation in the EU still have strong countrydisparities trade-off, given the potential role played by specific components.

knowledge-related spillovers in transmitting growth and

innovation across countries and regions. For these reasons, both country-level catching-up as well as knowledge spillovers (within and between

A central starting hypothesis concerning the link countries) appear to be fundamental in order to promote between growth and location is that innovation involves regional convergence. More generally, these results may interactions that are easier when agents are located close help explain why economic growth in the EU appears to

89

to each other. These arguments suggest that growth is be spatially uneven. 92

necessarily unequal across space because of its very

nature. Spatial inequalities must then arise, at least The existing theoretical literature on country-level initially; and their potential reduction essentially relies growth and convergence offers a wide array of on various forms of transmission mechanisms that arguments pointing either to the long-term reduction or, include technological externalities, but also trade and on the contrary, to the persistence and self-reinforcing

factor mobility (including labour and capital). 90 Since nature, of economic inequalities across countries.

93

 Such

knowledge and innovation are crucial for growth, arguments can be combined when analysing economic integration may trigger regional income simultaneously developments at the country level, and at inequalities by favouring the emergence of growth and the regional level within the same country. In particular, innovation poles within EU countries. This is the growth and development may raise regional inequalities, hypothesis supported by Giannetti (2002), who argues especially for countries lagging behind in development that greater economic integration intensifies where barriers to regional spillovers are potentially international knowledge spillovers (compared to withingreater. Initial investigations of these issues date back to country spillovers). This would favour convergence at the 1950s and the 1960s. Kuznets (1955) explicitly the country, rather than regional, level in the EU over refers to the existence of a “long swing” in income the period 1986-1992, which corresponds to the settinginequalities across regions - where there is first a rise, up of the Single Market Program. Recently, Keller and then a decline in income differentials, caused by the (2002) has also shown that global integration tends to urbanisation and industrialisation process accompanying lower country-specific barriers to knowledge spillovers. the decline of agriculture. Williamson’s (1965) seminal Nonetheless, innovation and technological diffusion in paper in turn provides coherence to these arguments by the EU seems to remain dominated by country-specific identifying the key elements driving the evolution of features. Bottazzi/Peri (2003) show this by studying the regional inequalities according to the stages of spatial distribution of research and development (R&D) development of a nation – which are essentially related and innovation spillovers, and by linking R&D and to structural changes, factor movement and public patenting activities across EU regions over the period policy. This implies that regional inequalities are likely

1977-95. They find that R&D spillovers are subject to to rise while countries are engaged in a rapid catchingstrong

distance-decay effects, with a significant up process. Any attempts at reducing them may

influence exerted by national borders. 91 eventually run counter to this process - lowering national growth and, consequently, the potential for

Two important results emerge from this literature. future regional spillovers. 94

The Kuznets-Williamson hypothesis is especially helpful in understanding the EU experience where catching-up of cohesion countries (as illustrated in Section 2.2) has translated into rising inequalities within

89 See Lucas (1988). This seminal paper builds on this idea to these countries. Quah (1996, 1999) shows that while

point out that the externalities central to endogenous growth Spain and Portugal experienced high growth rates and are mostly local in nature, and that they provide cities with an important role in promoting growth. A similar argument holds in Romer-type models where the location of

innovative activities is crucial for growth and technological 92 Within this context, growth and development may drive

progress. See Baldwin and Martin (2004) for a review of the rising regional inequalities, especially for countries lagging theoretical literature, and Audretsch and Feldman (1996) behind in development, where barriers to spillovers are

and Feldman and Audretsch (1998) for evidence. potentially greater.

90 See Lucas (2000). 93 See, for instance, Solow (2000) and Lucas (2000).

91 In a recent paper Bode (2004) provides similar evidence 94 These arguments are also well known in the urban

concerning German regions. economics literature. See for instance Alonso (1969).

Graph 4: Evolution of regional GDP per capita inequalities Graph 5: Evolution of regional GDP per capita inequalities by cohesion country, 1988-1996 for some new Member States, 1995-2001

0.3

0.12 ES EL PT other EU15 CZ HU PL

SK other EU15

0.11 0.25

0.1

ex 0.09 ex

0.2 nd

i i nd G in 0.08 G in i i

0.15

0.07

0.06 0.1

0.05

1988 1989 1990 1991 1992 1993 1994 1995 1996 0.05

1995 1996 1997 1998 1999 2000 2001

Source: Commission services. Source: Commission services.

rising regional imbalances during the 1980-89 period, The evolution of income inequalities in some of the new Greece experienced only modest growth rates, Member States provides even clearer evidence in favour accompanied by decreasing income inequalities across of the Kuznets-Williamson hypothesis. Graph 5 displays its regions. Petrakos and Brada (1989) and Petrakos and the evolution of the Gini index for the Czech Republic,

Saratis (2000) find similar evidence for Greece, while de Hungary, Poland and Slovakia. 97 All countries but one

la Fuente and Vives (1995) provide arguments along the (Poland) experienced regional inequalities that are larger same lines for the EU as a whole. Davies and Hallet than in the rest of the EU (excluding Portugal, Spain and

(2002), in a qualitative assessment of data, support the Greece). 98 More importantly, however, while for the rest

view that regional income imbalances tend to rise in of the EU regional inequalities remain fairly stable, in fast-growing cohesion countries. the four new Member States considered here we observe

Further evidence pointing in this direction is presented a clear rise in regional income inequalities, which is in Graph 4 which displays the evolution of the Gini especially pronounced for the Czech Republic, Hungary

index computed at the NUTS2 geographical level for the and Poland.

99

cohesion countries compared to the EU average, except Econometric tests of the Kuznets-Williamson hypothesis

95

Ireland for the 1988-96 period. The graph shows that have been rather limited so far. Petrakos et al. (2003) inequalities in Spanish regions are always greater than use standard econometric regression but find no clear for the rest of the EU, although variations are rather evidence. In fact, running a simple regression of these limited; while for Greece the level of regional two variables appears rather inappropriate, given the inequalities is always below the EU average. assumed non-linear nature of the relationship. In a recent

As mentioned earlier, this result possibly relates to the study, Barrios and Strobl (2004b) makes use of semifact that Greece is also the country which has parametric techniques in order to tackle this issue. This experienced the slowest growth of GDP per head on allows, in particular, a graphical representation. Their average over the same period. For Portugal the evolution approach is to regress the level of each country’s Gini is more contrasted, with a marked rise in regional GDP inequality index on the level of national GDP per capita,

per head inequalities just after EU accession in 1986,

which extends until the slowdown of 1993/94. 96

95 97 The regional data come from Eurostat following the ESA79 The data is taken from Eurostat’s REGIO database for the

definition of GDP which provides data up to 1996. Regional 1995-2001 period using ESA95 classification. No data for Ireland was not available at NUTS2 disaggregation comparable regional data at NUTS2 level was available for

level. Other data were also available for other countries but the other new Member States. they did not have enough regions in order to get a 98 Note that differences in Gini index values for the rest of the

representative EU-15 average or there were data problems EU between Graph 4 and Graph 5 are due to the fact that for some countries, in particular Portugal in the earlier datasets are taken from different accounting systems, the

period. first being the ESA79 system and the second the ESA95.

96 The average growth rate of Portuguese GDP was close to - 99 Similar evidence is found in Barrios and Strobl (2004b)

0.5 per cent against 1.15 per cent for the rest of the EU. who consider also other new Member States.

Graph 6: Semi-parametric estimations for EU-15, 1980-1996 Graph 7: Semi-parametric estimations for EU-15 and Hungary, Czech Rep., Poland and Slovakia, 1995-2001

4 3.5

in i

ge 3.5 G

ra 3 ve 3 ra

ge

in i a 2.5

2.5 U

 a ve

U G / E

2 x

2 ex

in i /

 E

nd in de

l G 1.5 1.5 in i i

l G

tr yle

ve 1 ve 1

un -le co 0.5 nt ry 0.5

0 C ou 0

0.75 0.80 0.85 0.90 0.95 0.99 1.04 1.09 0.70 0.75 0.80 0.86 0.91 0.96 1.01 1.06

Country-level GDP per head / EU GDP level average Country-level GDP per head / EU average GDP per head

Note: Excludes Denmark, Ireland and Luxembourg. Based on ESA79. Note: Dotted lines= confidence interval. Dotted lines= confidence interval. Source: Commission services. Source: Commission services.

corresponding fall occurs when a country reaches

both variables being measured relative to the EU approximately 70 per cent of EU GDP per head average. average (see Annex II for details on the econometric These results suggest that the rise in regional methodology). Graph 6 reports results of this regression, inequalities experienced by the countries with the lowest

together with the confidence intervals. 100 levels of economic development is likely to be only According to these results the relationship between temporary which is in line with the descriptive statistics national GDP per capita and regional inequalities is nonprovided above. In addition, Graph 7 shows that the monotonic, following an inverted u-shaped curve in line initial rise in regional inequalities is likely to be less with the Kuznets-Williamson hypothesis. This shows in pronounced in absolute terms than the subsequent fall as

particular that, for the cohesion countries which are all national development proceeds.

located at the left of the curve, i.e. with a GDP per head These results have also important policy implications as

inferior to the EU average, regional inequalities tend to

101 they point to the possibility of an equity/efficiency decrease as their national development proceeds. trade-off through which GDP per capita inequalities Graph 7 provides supplementary evidence adding to the would necessarily rise at the earlier stages of a country’s former EU-15 Member States the Czech Republic, development process. Indeed this idea fits well with the

Hungary, Poland and Slovakia during the period 1995-

102 current experience of the new Member States as national 2001. growth in these countries seems to be largely localised

Results shown in Graph 7 again depict an inverted uin the most dynamic areas around the capital cities shaped curve although several differences arise where investment, including public investment, is likely

compared to the previous result. First, the left hand-side to be more productive.

 103

of the curve is less accurately estimated as shown by the

wider confidence bands. Second, the rise in regional 2.4 Summary

income inequalities appear to be potentially much lower

in absolute terms than the subsequent fall experienced Income convergence in the EU has not proved to be a

for higher levels of GDP per head. Here the rapid, continuous or automatic process. The example of Ireland illustrates this best, with a first set of growthoriented

 policies initiated in the 1960s, yet catching-up

gaining momentum only in the mid-1980s.

Convergence in the EU has been faster at the regional

100 Data are taken from the Regio database for the period 1980- than the country level - due to wider initial disparities at

1996, using the ESA79 nomenclature for EU-15 countries. regional level and the strong catching-up of the most Note that the y-axis values are not reported given that they dynamic regions in some cohesion countries. New are estimated values with no direct interpretation. Member States started catching up at a moderate pace

101 Note that the end of the tail of the curve plotted in Graph 6

is slightly increasing. It is important to note that estimations

become less accurate at the beginning and the end of the 103 From a regional policy viewpoint, these results also support distribution, see Annex II. the findings of a paper by de la Fuente (2003) and Castells

102 Graph 7 uses instead data available under the ESA95 and Solé-Ollé (2004) who estimate that, in the case of

classification which provides regional data for the period Spain, the allocation of Structural Funds was under-optimal

1995-2001. from a national growth point of view.

after the transition crisis of the early 1990s. This process endogeneity (or circular causality), and the large number was accompanied by increasing within-country regional of potentially relevant variables influencing growth. disparities. Nonetheless, a fair degree of consensus has emerged in

Looking at the spatial dimensions of convergence in the this literature on the key policies likely to enhance – or, EU, regional specialisation and concentration has not respectively, damage – the prospects for growth.

changed significantly during the period for which data Second, given the heterogeneity of the new Member are available, and which includes episodes of rapid States, this section does not aim to put forward a economic integration. Hence, the existing core-periphery standard recipe for rapid catching-up. These economies pattern has remained broadly stable. Regional inherited different industrial structures, with for example inequalities appear to be influenced by national a large share of agricultural activity in Poland and a development paths with cohesion countries and the new strong reliance on tourism and the financial sector in Member States experiencing rising regional inequalities Cyprus and Malta. Eight of them are transition during periods of fast catching-up. Empirical evidence economies whereas two are not. Five are very small suggests that in the early stages of catching-up there is economies. Due to their openness, effective growth potentially a trade-off between national growth and strategies will rely much more on external regional income inequalities. Policy actions aimed at competitiveness than in larger Member States, for which maximising national growth may come at the price of trends on the domestic market will be more important. (initially) increasing regional imbalances. Against this This has also implications for the role of exchange rate background, economic policy in the EU aimed both at movements or domestic capital costs. Therefore, any favouring national growth and at fostering more rapid attempt to copy successful policies from other countries technological diffusion across regions (within and - such as Ireland, for instance - is likely to fail unless between countries) could help boost convergence at country-specific conditions are taken into account. country level and smooth the catching-up process of

lagging regions. Economic theory presents growth as ultimately driven by individual behaviour in households, enterprises, or

Given this past experience in the EU, and the education and research institutions: it thus assigns to considerable income gap of the new Member States, it is policy an indirect role only. This role is, however, very pertinent to ask how, if at all, policies can stimulate critically important. While most economic activity takes the process of catching-up. The remainder of this place on markets, the relevance of the policy framework chapter addresses this question from two angles. In for private decision-making can hardly be Section 3, policy-relevant insights are distilled both overestimated. For instance, the security of property from the economic literature and from empirical rights and returns from investment in capital, research or evidence for the new Member States, with the purpose education, are decisive inputs for individual decisions – of identifying priorities for policy-making in these and equally important is the availability of countries. On this basis, Section 4 discusses the role of infrastructure. Such factors are thus crucial determinants EU Structural and Cohesion Funds. of the growth process. Moreover, it is well-recognized

that an entirely market-driven allocation of resources may not lead to an optimal provision of goods.

• 3. 

  How to accelerate catch-up growth in the (Formally, the market may not reward goods that have

new Member States? features of non-excludability and non-rivalry in consumption, or produce certain externalities.) This

A primary goal of policy-makers is to improve standards applies notably to investment in knowledge: policy of living by stimulating economic growth – including needs to design incentives appropriately so that society notably where incomes are below those in neighbouring benefits to the maximum from individual decisions.

countries or trading partners. And many intuitively Despite numerous advances in the theoretical analysis of appealing proposals float in policy debates concerning economic growth in recent years, the traditional what policy can and should do to accelerate this production function approach remains the standard catching-up process. Two words of caution are thus analytical tool. This approach assigns little importance warranted up-front. to demand, which is generally considered to be more

First, the abundance of recommendations stands in sharp relevant for cyclical behaviour: rather, it focuses on the contrast to the difficulty of finding clear conclusions that supply-side of the economy - i.e. the accumulation of are supported by rigorous empirical tests, and are policylabour and capital, as well as technical progress - as the relevant. Indeed, the scope for unchallengeable results is drivers of any increase in output over time. Section 3.1 inherently limited by three features of the growth will take this perspective.

literature: the lack of sufficient data, the problem of

Table 8: Decomposition of the GDP growth rate in the new Member States

1996-2005 2006-2010

Growth Labour Capital TFP Growth Labour Capital TFP

Cyprus 3.41 0.73 1.53 1.24 3.63 0.46 1.82 1.29 Czech Rep. 2.20 -0.93 2.64 0.57 3.45 -0.62 2.48 1.59 Estonia 5.85 -0.61 2.85 3.48 5.76 0.30 2.85 2.42 Hungary 3.80 0.67 2.02 1.06 3.55 0.23 2.08 1.21 Latvia 6.32 -0.07 2.77 3.49 6.34 0.10 3.26 2.75 Lithuania 5.64 -0.37 2.80 3.11 5.73 0.29 2.69 2.56 Malta 2.48 0.23 2.07 0.18 1.99 0.03 1.60 0.42 Poland 4.25 -0.09 2.11 2.17 4.38 0.46 1.86 1.94 Slovakia 4.00 -0.53 2.49 2.00 3.94 0.48 1.20 2.14

Slovenia 3.76 -0.09 2.57 1.27 3.13 -0.20 2.15 1.15

Source: Commission services.

However, there are important elements of the growth percentage points or more in all cases except Cyprus. process that are not captured in the production function The contribution from TFP was highest in the Baltic approach: notably, the determinants of factor countries, and only clearly below 1 percentage point in

accumulation and innovation. Largely, these “deeper” the Czech Republic and Malta. sources of growth are attributed in the literature to trade,

104

geography and institutions. While the influence of Based on a number of assumptions, in particular the each of these factors on growth remains controversial, Commission’s autumn 2004 forecast and trend estimates there is some consensus that they all matter. Moreover, for the years after 2006, a medium-term scenario for many accept that policies should be considered potential GDP growth in the period 2006 to 2010 was separately from institutions. Section 3.2 will analyse the calculated. The technical extension to the years 2006 to potential contributions from these driving forces of 2010 is in no way a forecast for these years. It is simply

catching-up. an attempt to illustrate what would happen if the underlying trends of the most recent years were to 3.1 The accumulation and diffusion of continue. Average GDP growth would be similar or higher than previously in most of the new Member

production factors and knowledge States. In contrast to the previous period, labour should

make a slightly positive contribution in most countries -

3.1.1 Growth decomposition and a medium-term with the exception of the Czech Republic and Slovenia

scenario (and a broadly neutral effect in Malta). Capital and TFP are projected to remain important, but somewhat less so

To identify the respective contributions of labour, than in the previous period. Again, the three Baltic capital and total factor productivity (TFP, the “Solow countries achieve the highest contributions from capital

residual”) a decomposition of actual GDP growth in the and TFP among the ten countries.

new Member States between 1996 and 2005 was

calculated on the basis of the Commission’s production A further exercise was to transpose the projected function method. 105 The period was chosen not only for potential growth rates for the period 2006 to 2010 into reasons of data availability, but also to avoid the values of GDP per capita in PPS relative to the EU-15 influence of the early-1990s transition recession in eight average (see Graph 8). According to this medium-term of the countries. 106 scenario, all countries - with the exception of Malta -

would converge to the EU-15 average. The reason is that

Table 8 shows that average GDP growth was higher than in all new Member States except for Malta potential 3½ per cent in all transition economies except for the GDP growth is projected to be significantly higher that Czech Republic and even above 5 per cent in the Baltic of EU-15 (which is between 1½ per cent and 2 per cent). countries. Employment made a negative contribution to At the same time, the population is projected to decrease growth in most of the new Member States - the main in most of the countries while there is a small increase in exceptions being Cyprus, Hungary and Malta. the EU-15 between 2001 and 2010, except for Malta and Investment made an important contribution of 2 Cyprus where population is projected to increase much

stronger. As a result, by 2010, Slovenia and Cyprus

104 See Rodrik et al. (2002). would be around 80 per cent of the EU-15 average

105 For methodological explanations see Denis, Mc Morrow, income. The Czech Republic, Estonia Hungary and

Roeger (2002). Malta would be in a range of 60 per cent to 70 per cent.

106 However, it should be noted that financial crises took place

in some cases in the period under consideration.

Graph 8: GDP per capita in PPS in the new Member States Graph 9: Employment rates, 2003

90 e 80

80 e

ag 70

70 s

am

he 60

e 60 n in t

50

er ag

av 50 la

tio

5 pu e 40 ng

U -1 40 o f po ra 30

 o f

E %

% 30 in

ed 20 20

pl oy 10

10

s em

0 on 0 rs

CY CZ EE HU LV LT MT PL SK SI pe EU-15 CZ EE CY LV LT HU MT PL SI SK

2001 2005 2010 15 to 64 years 55 to 64 years

Source: Commission services Source: Commission services.

The remaining countries would converge to a range of While the unfavourable ratio of employed persons 50 per cent to 60 per cent. Thus the scenario shows that relative to those who are not employed and have to live if the currently favourable growth trends are assumed to on income distribution or savings (i.e. the dependency continue and potential growth rates were actually ratio) is mostly seen as a problem of public finance, it is achieved, the income gap vis-à-vis the EU-15 would still also reducing the prospects of economic growth. This is remain considerable in many of the countries at the end in particular so in countries where a strong demographic of this decade. decline in the next decades will lead to an ageing population. Between 1990 and 2003 all new Member

3.1.2 Labour utilisation States except Malta, Poland and Slovakia have already lost in population, the most severe losses being in the

With a socially unacceptable high rate of unemployment Baltic countries due to the out-migration of people of and an employment rate far below the Lisbon target of Russian origin. According to the medium scenario of the 70 per cent, raising employment is a policy priority not UN population projection, the median age in most of the only for the old but also for the new EU Member States. new Member States is expected to increase by more than It does not only serve to stimulate growth but is also 10 years until 2050 from below 40 in 2005. The Czech important for the distribution of income and the Republic, Estonia, Latvia and Slovenia would then have reduction of social exclusion. Chapter 3 in this volume a median age of above 50. Hence, there is an increasing gives a more detailed account of incentives on labour need to redesign the tax-benefit systems in a way which markets and means to raise employment and labour gives incentives to older people to stay longer in force participation. 107 employment and have them participate in the generation

Graph 9 illustrates the differences among the new of income.

Member States regarding employment rates in the In theory, these demographic developments could

working age population of 15 to 64 years. While Cyprus deteriorate further if there were further substantial outalmost

achieved the 70 per cent target in 2003, Hungary, migration of younger people. Transition periods of up to

Malta and Poland were even below 60 per cent. The 7 years after accession to restrict the free movement of

overall employment rate is to some extent influenced by labour from the new Member States (except for Cyprus

the rate of the older age group of 55 to 64 years which is and Malta) are applied by all old Member States except

also given in Graph 9. In Hungary, Malta, Poland, Ireland, Sweden and the United Kingdom. While some

Slovenia and Slovakia less one third of the persons in migration has already taken place before accession, most

that age group are employed. In many transition empirical studies suggest that no substantial migration

economies generous schemes of early retirement were flows are to be expected and estimate the long-run

used to cushion the adverse social effects of labourmigration potential from the 10 central and eastern

shedding enterprises in restructuring. European countries (CEEC-10: 8 new Member States,

Bulgaria and Romania) into the EU-15 at between 2 per

cent and 4 per cent of the population. A study carried

out for the European Commission projects, after full

liberalisation, an initial net increase of residents from the

CEEC-10 of 290,000 persons with the net increase

107 See also European Commission (2002a), Chapter 5, and

European Commission (2004a), Chapter 1.

Graph 10: Employment share in the total economy, industry The high unemployment in some of the countries and

and services the need for further adjustments in the future give rise to

the question whether labour markets are sufficiently % flexible to support a fast process of catching-up. The 100 OECD index of the strictness of employment protection

legislation for 2003 is available for the Czech Republic,

80 Estonia, Hungary, Poland, Slovenia and the Slovak

60 Republic. It suggests that employment protection is less

strict than in many of the old Member States with the 40 exception of Slovenia where the index turns out to be

rather high and thus indicates some excessive rigidity.

20 However, in some of the countries the wage bargaining

0 system and the tax-benefit system lack flexibility and AC 1996 AC 2003 Euro area 2003 reduce the incentives to create jobs or take up a job. For

example, the tax wedge on labour costs for low-wage

Construction Agriculture et al. earners is higher than the EU-15 average in all new Manufacturing + energy Services Member States except for Estonia, Cyprus and Malta.

Source: Commission services. Although most quantitative indicators do not show this

to be a problem in the new Member States, there is some evidence of quality problems to provide a well-educated

peaking at around 370,000 persons and a long-run stock and trained labour force which is key to high labour

of 3.8 million persons (about 3.7 per cent of their

108 market flexibility.

population in 2003). Nevertheless, even if quantities

are not large, there could be constraints to growth by High disparities in regional unemployment also point to out-migration of the most qualified (i.e. “brain drain”). problems of labour market flexibility. At the level of

NUTS3 statistical regions in 2002, about one third of the

The sectoral structure of employment can also give an labour force were unemployed in some Polish regions indication on growth prospects with a view to either and about one quarter of the labour force were future adjustment needs (e.g. reduction of agriculture) or unemployed in several Polish and Slovakian regions the potential for employment in activities of higher whereas their capitals had single-digit rates of productivity. The economic literature, following the unemployment. Analyses of similar cases of high Kuznets hypothesis, identifies several regularities as disparities in regional unemployment in Germany, Spain employment structures change in the course of economic and Italy give a number of explanations which also tend development – patterns that also seem evident in the to hold for the new Member States: 110 low level of new Member States. Among these regularities is a regional development, insufficient labour force decline of employment in agriculture, and an increasing qualification, a wage bargaining system which does not share in services. The proportion of employment in take into account regional differences in labour industry follows a non-linear pattern. It first increases productivity, and insufficient geographic labour

and later on declines. 109 mobility. The latter is particularly the case in many of Graph 10 presents the change in employment structure the new Member States due to the frequently applied over time, showing a decline in the share of agriculture privatisation approach of giving housing to the tenants and a build-up in services. The share in manufacturing and an inadequate regulation of the housing market - in the new Member States as a whole has somewhat which leads to a high share of owner-occupation and an

declined over the last several years but is still higher almost negligible rental market.

than in the euro area. Overall, while some convergence

to the present euro-area employment structure is evident, 3.1.3 Capital deepening

the difference is still apparent. Employment in

manufacturing is particularly high in the Czech Investment is considered a key driver of economic Republic, Slovenia and Slovakia. The breakdown of growth in general and in the new Member States in employment in services shows that the share of particular. When capital is scarce in an economy, the employment in trade and transport is much higher, in working of market forces should result in high returns finance lower and in public sector activities about the on capital, which provides incentives to further same as in the euro area. accumulation of capital either financed through

domestic savings or from abroad.

108 See Alvarez-Plata et al. (2003).

109 See Raiser et al. (2003) for a model and an empirical

estimation of structural changes in employment in transition

economies. 110 See for example Davies and Hallet (2001).

Graph 11: Investment share in GDP in the new Member Graph 12: FDI inflows into the new Member States States

35 1995-99

30 2000-03 6

ic es 25 10 new Member States 5

t pr Excl. SK rr en 20 cu 4

P , P 15 G D

G D 3 o f

 o f %

% 10 2

5 1

0

  CZ EE CY LV LT HU MT PL SI SK EUR- 0

12 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

Note: No data available for Slovakia 2001 and 2002. Source: Commission services. Source: IMF; Commission services.

Data on aggregate investment are available in the A less positive assessment of the new Member States national accounts. Graph 11 shows that the investment could be brought forward in view of their current share is higher in most Member States than in the account deficits. According to Orlowski (2004), external euroarea. In those countries, where the decomposition imbalances caused restrictive stabilisation policies that into private and public investment is possible, the higher constrained growth in almost all transition economies. investment share is not explained by higher public He quotes the episodes of Hungary 1995-96, Czech investment and the share of private investment is higher Republic 1997-99, the Baltic States in 1999, Slovakia in than in the euro area. It is, however, not evident whether 1999-2000 and Poland in 2001-02. The only exception

this finding implies a higher capital intensity of was Slovenia which has a high domestic saving ratio. production inherited from past production structures or Since the new Member States opened their capital rapid capital accumulation required by a low capital accounts at a relatively early stage of economic stock. The change of the investment share over time – transition, a large share of investment was financed from here 1995-99 vis-à-vis 2000-2003 – can shed some light abroad. 112 Given a shortage of domestic savings, on this issue. In all countries except the Czech Republic, financing of investment relies to a strong degree on Malta and Slovakia, the investment share increased, foreign savings, particularly in the form of foreign direct suggesting that a too capital-intensive production investment (FDI). However, since at least a part of FDI

structure is unlikely to be the reason. 111 inflows were related to privatisation, which is initially

Employment structures provide some further indication only a change of ownership from the state to a foreign whether the new Member States are characterised by investor, FDI is not equal to capital formation. Over the rather capital-intensive production. Employment in past years, FDI inflows to these economies amounted, industry was typically far higher in 2003 than in the euro overall, to about 4 per cent of GDP, meaning that FDI area - and particularly in those that also have a high was the main way of financing their current account investment share (see Graph 11). Whether these deficits. Graph 12 demonstrates that FDI inflows economies have a comparative advantage in industrial picked-up only in the late 1990s, and seem to have production seems to depend very much on the weakened somewhat after 2000 when global capital availability of cheap skilled labour. They would face flows also softened in the wake of slower global increased pressure for structural change if this economic growth. Comparing FDI flows to other comparative advantage ended - i.e. if productivity regions of the world in the 1990s, Campos/Kinoshita growth does not keep pace with wage growth. (2003) conclude that the high expectations in transition

economies had not materialised. Disproportionately more capital was diverted into Asia and Latin America

than into the transition economies.

The composition of FDI flows to the new Member States can inform about the direction of structural

111 change. Lovino (2003) identified some patterns on the It is not evident that the decline in the investment share

indicates an already very capital-intensive production in the

mentioned three countries. On the other hand, the fact that 112 Relatively means in comparison to western economies,

the investment share was already high in 1990 in these which only gradually opened capital accounts during the

countries supports this interpretation. Bretton Woods period.

basis of the stock of FDI in 2000 which are consistent evidence in favour of microeconomic determinants, with the direction of sectoral change derived from which is in line with the findings of Campos and

employment shares: 113 Kinoshita(2004), who find institutions, openness and

• More than a third of the FDI stock (36 per cent) was agglomeration effects to matter most in Eastern invested in manufacturing in the new Member European countries, differs from the previous finding States in 2000, compared to 28 per cent in the EU. that macroeconomic factors had been key driving factors

In 1997, the share was still about 50 per cent. of FDI inflows in the earlier time of transition.

117

• FDI flows into services have become more Graph 13: Saving rates in the new Member States, 2003

important over time, in particular in the sub-sectors financial intermediation and trade. 30

• In 2000, 77 per cent of FDI stocks were 25 National

concentrated in the Czech Republic and Poland. Private

20

• The lion’s share of 73 per cent of FDI stocks in the P new Member States had its origin in the EU-15. The G

D 15

 o f

Netherlands, Germany, France and Austria were %

particularly important investors. 10

While the ultimate assessment of the new Member 5 States’ liberalisation strategy is still outstanding, the

large inflow of FDI is generally viewed as positive. For 0

the host country, it meant the import not only of capital CZ EE LV LT HU MT PL SI SK EUR- 12

but also of technology and managerial skills. 114

Empirical estimates of the impact of FDI on domestic Source: Commission services. economic activity have also shown that FDI interacts positively with domestic investment, i.e. higher FDI

spurs domestic investment and vice versa. 115 Mody et al. 3.1.4 Innovative activity and knowledge

(2003) find evidence that FDI has also a positive impact Although it is widely acknowledged that stimulating of the efficiency of the capital stock and its allocation innovative activity and generating technical progress are across firms in a sample of industrial countries, i.e. not crucial determinants of economic growth, the economic covering the new Member States. According to the literature still offers surprisingly little policy-relevant empirical estimates of Tondl and Vuksic (2003), FDI insights. The most promising venue in economic theory inflows were the key driver for economic growth in has been the modelling of knowledge creation whereas central and eastern European regions in the second half the empirical literature has identified a number of of the 1990s. In particular capitals and border regions potentially important determinants of total factor benefited from FDI. productivity growth. In general terms it can be claimed

As regards the motivation of foreign investors, the that knowledge creation through research, knowledge available evidence suggests that both access to domestic diffusion through education and training and its markets and lower production costs play a role in the application are important pre-conditions, but they have case of the new Member States. 116 For instance, to be combined with incentives to draw economic Carstensen and Tourbal (2003) found market potential, benefits from it. It is, however, difficult to translate the relative low real unit labour costs, skilled workforce and academic insights into more concrete policy advice. This relative endowment to be significant determinants in section aims at highlighting two aspects that are of their estimates. Among transition-specific factors they relevance for catch-up growth in the new Member find support in favour of the importance of the level and States, namely the role of knowledge transfer from other

method of privatisation and country-specific risk. This countries and a comparable evaluation of factors that are considered of relevance for stimulating domestic

innovative activity.

113 Due to problems of data availability, the new Member

States in this study exclude Cyprus, Hungary and Malta. The most apparent form of knowledge diffusion across

114 borders is education abroad and labour migration. Razin (2002) considers the import of management skills a

major advantage of FDI, improving the efficiency of the Ireland is currently seen as the prime example that brain

economy and therewith making FDI more than a simple drain, which is usually regarded with scepticism when it substitute for trade. occurs, can be of benefits in the long term. Key elements

115 According to Hecht et al. (2002), the effect is, however, of such a “development strategy” is the preparedness of

smaller than initial estimates had suggested. students and workers abroad to return to their home

116 Campos and Kinoshita (2003) find evidence that country and make use of newly acquired knowledge. In

endowment with resources and infrastructure matter more as determinants of FDI in the former Soviet Union than in

the Baltic States and other CEECs. 117 See Garibaldi et al. (2001).

this context, the main insights from the Irish experiment

are more of social than of economic nature. The number Graph 14: Educational attainment and high-tech employment in the new Member States, 2002

of researchers from the new Member States at

universities and research facilities has undeniably 16

increased over the last decade and a similar trend can EU-15

realistically also be present for workers. However, little s) 14 new Member States vi ce

is known in quantitative terms about this phenomenon. s er 12 Linear (EU-15)

nd

A second important mechanism for importing fg a 10 knowledge is through trade and FDI. The endogenous t (m 8 growth literature has provided some support for the view m en that imports, in particular of intermediate goods, have a pl oy 6 positive impact on productivity growth in the importing h em 4

country. Concerning FDI, Barba, Navareti and Tarr -t ec 2

(2000) paint a less upbeat picture. Although the hi

gh

productivity of the recipient economies increases 0 through the activity of multinational enterprises and 40 50 60 70 80 90 100 Youth education attainment level

industry-level studies suggest that spill-overs are

positive, studies on firm-level are less positive. Activity Note: Youth education attainment level - total - Percentage of the

of foreign-owned firms may have negative spill-over population aged 20 to 24 having completed at least upper secondary education.

effects if it reduces domestic firm’s ability to benefit Source: Commission services.

from scale economies. 118 A crucial intervening element

in this debate seems to be the time period and the degree

of competition prevalent on the market. Boeri and Higher education is an important determinant of human resources. For instance, Tondl and Vuksic (2003)

Bruecker (2000) found that FDI was often directed to

underdeveloped market segments to make use of firstreasons that higher growth in the new Member States’ mover advantage in markets with little competition. capitals is due to these regions’ endowment with a more They claim that the strategy of extracting rents might qualified labour force which makes them more attractive explain why the new Member States benefited relatively as a location for FDI. Graph 14 shows on the horizontal

little from FDI in terms of technological spillovers. 119 axis that upper secondary education among the currently young people in the new Member States’ labour force is

The new Member States’ ability to benefit from not systematically different from that in the EU-15 importing technology can be considered to depend Member States. With the exception of a low share of largely on the same factors that would allow them to secondary education in Malta, the range is about the develop domestically-driven innovative activity. In same in the old and the new Member States. A similar order to structure the discussion and highlight Member picture would emerge for tertiary education. One States’ potential, the subsequent part builds on the observes an extraordinarily high share of tertiary Commission’s 2003 European Innovation scoreboard education in Latvia, but the range is about the same in

indicators, which distinguish among four main factors of the old and the new Member States.

innovative activity. These are determinants governing

human resources, knowledge creation, transmission and One finding is the absence of a clear relationship application of knowledge, and innovation, finance, between higher education and employment in high-tech

output and markets. manufacturing and services. The existence of such a relationship for the old Member States is suggested by

While indicators are available for almost all EU-15 the trend line in Graph 14. It is similar when tertiary Member States, gaps are still common for the new education is used instead of upper secondary

Member States, in particular for indicators in the fourth education. 121 If the share of graduates in science and

category data. Since the innovation scoreboard is based engineering is related with employment in high-tech on the most recent data available, data entries can be sectors, there is a positive correlation for the old different across countries. In most cases, data is from Member States but an inverse one for the new Member

120

2001 or 2002. States. That is, employment in high-tech sectors in the

118 See Djankov and Hoekmann (2000) for an analysis of

Czech firms.

119 See also Barrel et al. (2001) and Holland and Pain (1998).

120 For more detailed information on the innovation scoreboard see European Commission (2003a). Key input to the 121 Since there is no breakdown in the tertiary education into

innovation scoreboard is the Community Innovation age classes and education of elderly people in the work Survey, which is conducted all four years. The latest data force under the previous regimes may not be worth much

was released in August 2003. today, this indicator is probably less telling.

Table 9: Innovation indicators for the new Member States

CY CZ EE HU LT LV MT PL SI SK EUR- 15

Public R&D expenditure, % of GDP 0.22 0.52 0.53 0.57 0.49 0.28 -- 0.43 0.69 0.22 0.69

Business R&D expenditure, % of GDP 0.05 0.78 0.26 0.38 0.2 0.16 -- 0.24 0.94 0.45 1.3

High tech patent applications (per mio

inhabitants, EPO+USPTO) 3.2 0 0 4.6 1 0 4.1 0.3 9.1 1.3 44

Patent applications (per Mio inhabitants,

EPO+USPTO) 17.1 13.7 13.2 26.3 3.8 8.4 15.3 3.6 53.8 6.8 241

SMEs innovating in-house (% of SME,

mfg.) -- 25.8 39.1 -- 26 19.1 15.4 4.1 22 14.1 37.4

SMEs involved in innovative cooperation

(% of SMEs, mfg.) -- 5.8 11.8 -- 12.1 4.1 4.9 -- 8.4 4.4 9.4

Innovation expenditure (% of turnover,

mfg. and services) 2.2 3.35 3.89 5.31 6.8 16.3 5.28

ICT expenditures (% of GDP) -- 9.5 9.6 8.9 5.9 7.9 4.1 5.9 4.7 7.5 7 Share of value added in high tech

sectors (mfg.) -- -- -- 14.9 22.3 -- 22.4 -- 15.9 -- 14.1

Note: Data not completely comparable since the methodology in some cases is different and the data processing has not been harmonised.

Source: WITSA/IDC, Commission services.

new Member States is rather invariant to their Member States. 123 But again the empirical patterns are endowment with human resources. 122 Although an different between new and old Member States. In the old

explanation for this finding is not straightforward, it Member States labour productivity growth across could suggest that most new Member States have not countries is positively correlated with all innovation (yet) specialised into high-tech sectors. indicators bar innovation in SMEs. For the new Member

Numerous factors related to knowledge creation and diffusion have been highlighted in the economic

literature. Some of them stand as proxies for input in Graph 15: Correlation of innovation indicators with labour productivity growth, coefficient of correlation of ranks across

research, such as R&D expenditures. Others measure the new and old Member States

number of patents as the intermediate output. Since SME’s are expected to play a central role in innovation,

indications of how they actually perform can be Public R&D 1.00

informative. While the theoretical link of these variables

with technical progress is apparent, complications occur high tech VA Private R&D

when it comes to presenting empirical evidence. 0.00 Relationships over time or across countries are often weak and seldom stable if other control variables are ICT exp. High tech patents included. -1.00

Table 9 gives an overview of selected indicators on innovation in the new Member States. Although data are

not fully comparable across countries, there are clear Innov. exp. Patents signs of a lag in innovative activity in the new Member new Member States EUR-15

States relative to the EU-15 Member States. Ignoring SMEs cooperation SMEs innovation problems of data comparability, one would expect that

Member States that fare high with the innovation Note: The dark dots show the coefficient of correlation between the rank

indicators should also have higher labour productivity of the innovation indicator among new Member States and the new Member States’ rank of labour productivity growth 1999-2003. The light

growth and vice versa. By plotting the coefficient of line shows the result of the same exercise for the old Member States.

correlation between labour productivity growth and For more explanations on the variables, see Table 10.

innovation indicators across countries, Graph 15 shows Source: Commission services.

that this is neither the case for the EU-15 nor for the new

122 This result is consistent with the empirical analysis of Tondl 123 More precisely, it shows the correlation between the rank

and Vuksic (2003) that finds own innovative activity not to among countries of labour productivity growth and the rank

be a significant growth factor in the new Member States. of innovation indicators.

States, it is negatively related to 5 of the indicators and improvements in the education systems in several of the

close to zero for a sixth indicator. Only in SME countries. 124 Although less than one would expect, trade

innovation and ICT expenditure is the relationship and FDI have been important for the cross-border among new Member States more consistent than across transfer of knowledge in management and technology. the old Member States. This seems to suggest that However, innovation has not yet been a central innovation has not yet been a central determinant of determinant of productivity growth in the new Member productivity growth in the new Member States. States. Activity and employment in R&D and innovation

tend to be much lower which can best be explained by a

3.1.5 Policy challenges different pattern of specialisation. One should however be cautious in urging the new Member States and their

The decomposition of growth in the new Member States business sector to spend substantially more on R&D indicates that, between 1996 and 2004, it was driven by activities at the current stage given that, due to their investment and total factor productivity, while labour specialisation, spending in other areas could have much utilisation had a negative effect. In a scenario from 2005 higher returns at this stage.

to 2010, which is a projection of what would happen if

the trends of the most recent years were to continue, 3.2 Other determinants of economic growth

contributions from capital and total factor productivity moderate somewhat and labour is expected to have a

positive, although limited contribution to growth. Except 3.2.1 Trade and geography

for the Baltic countries, growth rates are projected to be Trade and geography are among the factors that have below 5 per cent which allows only limited progress in long been considered as the most important driving catching-up vis-à-vis the EU average. forces of long-term growth and development.

Employment rates, in particular in the old age group, are It is a well-established fact in economic theory that trade fairly low in most of the new Member States. In view of liberalisation promotes economic efficiency and the expected demographic trend of an ageing population, consumer welfare, but proper modelling of the link to tax-benefit systems need to be reviewed to provide growth is more recent. In the context of studies on the incentives for creating and taking up jobs and to increase expected Single Market effects, Baldwin (1989) argued the actual age of retirement. Deficits in the adaptability in a Solow-type model that the first-round allocation of the labour market do not appear to arise from effects due to a larger market would induce a secondexcessively rigid employment protection legislation but round effect of higher income, savings and investment rather from insufficient wage differentiation, the taxas medium-term growth effects. The intuition would be benefit system and a lack of regional labour mobility for that export-oriented and import-competing firms would a number of reasons such as an inflexible housing invest to improve their competitiveness. New growth market. In view of the still high importance of theory focussed on knowledge spillovers that can go agricultural employment in some of the new Member along with the trade of goods. 125 Taking into account States, an unduly generous support to this sector should that trade often goes along with FDI, this argument of be avoided in order not to decelerate the required technology transfer seems rather plausible. However, the structural change to employment of higher productivity empirical evidence on the trade-growth links is rather in other sectors. weak and has been subject to scepticism. 126 Causality (or

Capital deepening has been an important source of endogeneity) is a major problem. For example, the growth in the new Member States which tend to have a standard result that more open economies tend to be much higher investment-to-GDP ratio than the EU-15, richer can either prove the growth effects of trade or although they still have a less capital-intensive prove that richer economies find it easier to liberalise production. Given the early liberalisation of capital their trade. Investment in export-oriented sectors can movements, FDI was a major source of financing the have positive effects on both growth and trade current account deficit as a reflection of the gap between simultaneously. Furthermore, trade liberalisation often domestic savings and investment. The mostly foreigngoes in parallel with other economic policies which owned banking sector, rather than the stock exchange, makes it difficult to isolate the effect of trade was the main channel of financial intermediation. Given liberalisation. Lee et al. (2004) are trying to deal with that the net inflow of foreign capital to finance these problems more explicitly by applying continuing high investment will not last forever, the sophisticated econometric methodology and find a main challenge will be to gradually mobilise more robust effect from growth to openness and a positive, domestic savings through other channels than banks although small effect of openness on growth.

such as pension funds and the stock markets.

Innovative activity and knowledge are important triggers

for technical progress. Educational attainment levels in 124 See European Commission (2002a), Chapter 5.

the new Member States do not differ much from those in 125 See Rivera-Batiz and Romer (1991).

the EU-15, but anecdotal evidence suggests that there 126 See, for example, Rodriguez and Rodrik (2000) and Wälde

could still be a quality problem which require and Wood (2004).

Other authors argue that geography is the most

important determinant of growth and point to the Graph 16: Change in exports and imports of goods and services in constant prices from 1995 to 2003 in %

influence through resource endowments, productivity

and access to markets. 127 Some models of imperfect 188

competition of New Economic Geography illustrate how

small initial differences in market size can lead to the 168 Exports Imports

formation of a high-wage centre and a low-wage 148 periphery. The EU itself provides some evidence of the 3 128 importance of geography when considering that, -2 00 108 although with several exceptions, the richest regions 95 tend to be located in the centre of the EU whereas the e

19 88

ng 68

poorest regions tend to be located at its periphery. c ha

Again, causality is a major methodological problem % 48

since high-income countries have the possibility to 28

mitigate the adverse effects of geography, for example 8 through investment in infrastructure and technology. CZ EE CY LV LT HU MT PL SI SK

Whatever the difficulties of providing empirical

evidence on the individual effects of trade or geography Note: Data are from national accounts.

on growth, the gravity model is a powerful empirical Source: Commission services. tool combining economic size and distance between

countries to predict bilateral trade flows without followed by Italy. While at the time these were only

implying strong causality among them. 128 Rose (2000)

used the gravity approach to find that currency unions rough estimates due to limitations of data, it clearly illustrated how trade barriers can impede potential trade

have a tripling effect on trade. This finding was on the basis of geographic and economic conditions.

subsequently critically discussed and revised somewhat

downwards but there tends to be agreement on the Although time periods are not fully comparable due to overall large effects of currency unions. Applying this problems of data availability, Graph 16 shows the approach to EMU, trade effects of up to 50 per cent enormous growth in total trade of the new Member

129

were found. States. Trade in constant prices has doubled in all of

them except for Cyprus, Slovenia and Malta between 1995 and 2003. Imports have been growing much faster

The situation in the new Member States than exports and several of the countries are running

large trade deficits. The relatively slow growth of trade

Trade between the EU and the 10 new Member States in Slovenia could be related to the low FDI inflows, due has been liberalised to a large extent already before to the special way of privatisation, which triggers fewer accession in the context of the Europe Agreements imports of intermediate goods for the production and

signed in the early 1990s. This was done in an export of manufactured goods.

asymmetric way, i.e. the EU opened up its markets

faster. In their transition process most of the eight new Member

States have substantially reoriented their trade from the

Baldwin (1994) used a gravity model to compare the ex-communist trade partners towards the EU-15. Except potential trade of central and eastern European countries for Lithuania, more than half of all their merchandise in the absence of trade barriers to the actual trade in trade in 2002 was with EU-15 countries and in all 1989. The ratio potential to actual trade was nearly central European Member States (except Slovakia) as always larger than unity and exceeded 2 in the case of well as in Estonia the share exceeded 60 per cent. The Poland and 4 in the case of Czechoslovakia. The largest EU-15’s share in the new Member States’ total exports potential EU exporter to CEECs would be Germany was considerably higher than the share in total imports

except for Malta, Cyprus and Slovenia. Within the new Member States, the regions closest to EU-15 usually had the strongest trade and growth effects. The most

127 For example, Gallup et al. (1998) demonstrate that location important trade partners in EU-15 are those closest to

and climate have large effects on income levels and income new Member States, i.e. above all Germany, but also

growth through their effects on transport costs, disease Italy, Austria, Greece and Finland. Over the last years

burdens, and agricultural productivity, among other the EU-15 had a trade surplus with the new Member channels. States which can be explained by strong demand for

128 In physics, gravity is a function of mass and distance. durable consumer goods and investment goods which

However, economic theory has problems to model the are still hardly produced locally. Trade specialisation is results of the gravity model. For such an attempt see for still in labour-/low-skill-intensive production but

example Evenett and Keller (2002). Hungary, the Czech and Slovak Republics, Slovenia and

129 For an overview on trade effects of EMU see European

Commission (2004b), Chapter IV.

Estonia are gradually moving into more technology 3.2.2 Macroeconomic policies

/high-skill intensive production. 130 Accomplishing and preserving macroeconomic stability

is consistently seen as an essential contribution of public Policy challenges policy to economic growth. The justification is apparent.

The new Member States have undergone considerable Volatility in macroeconomic conditions entails that trade liberalisation since the early 1990s and have long-term planning is exposed to a higher degree of impressively increased their trade, in particular with the uncertainty. Investment projects with long gestation EU in the context of the Europe Agreements. A part of periods and high sunk costs are likely to be most the growth performance in the past is likely to have responsive to changes in macroeconomic stability. This

benefited from this increase in trade. does not only hold for investment into physical capital but also for investment in research, skills and

Membership brings some further trade liberalisation education. 132 The more uncertain the macroeconomic

regarding sensitive sectors (agriculture, steel, services) environment, the less resources are used for long-tem and non-tariff barriers as well as a possible further investments and the lower is the potential rate of growth.

reduction of transport costs arising from reduced waiting

times at borders and from improvements in Whereas economic theory allows postulating that high infrastructure. The latter will also depend on more inflation must not impede economic growth, high efficient network industries as a result of their successful inflation tends to mean in practice also a high variation

liberalisation. Less exchange rate volatility in the case of in the rate of inflation, which is an obstacle to longer ERM II participation and later the adoption of the euro term planning and could result in distorted relative could even further reduce trade costs and could have prices. For instance, it is often argued that high inflation

substantial trade effects. leads to over-investment in assets such as real estate that are considered to offer a safeguard against inflation. The

Improving even further the already good market access interaction of inflation with the tax system implies should enhance their competitive position in the Single distorted incentives to invest and this might entail Market. The gravity model suggests that these significant economic costs even at moderate rates of

reductions in trade costs will result in further increases inflation. 133 As regards fiscal policy, a high level of

in trade. It also points to the importance of the growth taxation may lead to distorted incentives to invest and to performance of the core euro area economies for the bear risks. An unsustainable path of public debt implies trade and growth performance of the new Member austerity measures in the future. Current investment States. For those among the new Member States, whose decisions will be framed in anticipation of high future location puts them at a geographic disadvantage in taxes and less public spending, with practical experience developing closer economic ties with the EU (such as showing that investive public spending is often strongest

the Baltic countries, Cyprus and Malta), particular curtailed when public finances are consolidated. efforts on reducing trade costs will be key to further

growth from trade integration with the EU. Given that While the empirical evidence of an inverse relationship they are likely to specialise their intra EU-trade in those between very high inflation and economic growth is goods and services with lower transport costs, such as undisputed, the case is less clear for moderate rates of

the exchange of data and information, developing the inflation.

134

  Despite some reservations on their

relevant infrastructure and education - in particular in robustness, several empirical studies on the basis of data information and communication technologies - could be for the OECD countries, which have low to moderate

particularly important. rates of inflation, were able to establish a link between the two variables. 135 It might be that for low rates of

The expected trade-related growth effects will also inflation, it is less the efficiency of the price system in depend on the external competitiveness of local firms the allocation of resources that matters for growth but and their incentives to improve their efficiency through the distortion of incentives due to the interaction of investment. In order to be able to compete on EU and inflation with capital taxes in particular. Thus, empirical global markets, flexible and liberalised product markets results may depend on the kind of investment or fiscal

are of major importance. While aspects of capital and variables that are included in the regressions. labour markets have already been discussed above, there are indications that there is a considerable potential to increase the efficiency of product markets in most of the

countries. 131 132 For a literature review of the theory and evidence on the

link between macroeconomic stability and growth, see Ahn

and Hemmings (2000).

133

  See Feldstein (1996).

134

  For example, Bruno and Easterly (1998) argued that the inverse relationship between inflation and growth was only due to outliers with very high rates of inflation and the use

130 See Landesmann (2003). of high frequency data.

131 See European Commission (2004a), chapter 4. 135 See Ahn and Hemmings (2000) for an overview.

Studies that analyse the impact of public finances on expected negative relationship to growth. Public economic growth tend to find a negative correlation investment and consumption tend to have different signs between public deficits or debt and economic growth. A in most estimates. The positive impact of public caveat is, however, related to the direction of causality investment on growth has also been found in some but since it is difficult to establish whether either high not all other studies. European Commission (2003b) deficits are hampering growth or be themselves a concluded that results appear weak and fragile, pointing consequence of low growth. While moderate changes in to the consensus that public investment is less important

fiscal policy may have little or no negative effect on for growth than other factors. economic growth, especially if the government has

access to finance on the global capital market, there may The situation in the new Member States

be a country-specific threshold above which market The new Member States’ performance varied with participants perceive the fiscal stance as not sustainable respect to a number of macroeconomic variables, and demand a higher risk premium for holding the including inflation, public finance and the current

country’s assets. This may then give rise to crowdingaccount deficit.

out effects to private investment. 136 Graph 17 plots both average inflation (x-axis) and

Evidence is generally more robust on that the inflation variability (y-axis) in the new Member States composition of public expenditure and their financing (light squares) in two periods, namely 1997-99 (dark matters rather than global variables such as total public diamonds) and 2001-03 (light squares). For comparison, expenditure, revenues, debts or deficits. This is why the the observation for the euro area in 2001-03 (light

policy discussion on this issue has increasingly focused triangle) is also included. 139 It shows that inflation rates on the quality of public finances. 137 have come down markedly in the new Member States.

The OECD’s growth project 2000-2001 aimed at In the later period, the difference to the euro area is compiling all the available evidence on the factors small for some of them. The reasons for the success of driving economic growth. 138 As regards inflation, the monetary authorities in the new Member States in results provide support for the notion that evidence in engineering disinflation are still disputed in the favour of a negative impact of moderate inflation on economic literature. Disinflation is attributed to economic growth (or investment) is hard to obtain. favourable developments in import prices, institutional Evidence is more apparent for the impact on growth of developments driven by the prospect of EU accession the volatility of inflation. Whenever the variable was and conducive to a sounder policy-mix, and the included, it turned out with a significantly negative sign. diminishing need for adjustments and liberalisation of

As regards public finance variables, the estimates administered and regulated prices.

140

suggest that the share of tax and non-tax revenues in

GDP is inversely related to economic growth and Graph 17: Level and variability of consumer price inflation in the 10 new Member States

investment. This, however, leaves open whether the

level of taxes is negatively affecting growth or whether 4.0

high taxes are correlated with a high level of distortive

taxes. The ratio of direct taxes to indirect taxes, where 3.5

the latter are supposed to be less distortive to 3.0

investment/saving decisions, turned out to have the yr s) 2.5

r 3

ve 2.0 o

136 Identifying such a threshold would be a serious challenge (s td

y 1.5 1997-99

for theoretical as well as empirical work. Even if it were bi lit possible, it would be of no help for practical policy making ria

1.0 2001-03

va

as these estimates were subject to the Lucas critique, i.e. 0.5 Euro area 2001-03

change once they are known. 0.0

137 An obvious indication for this shift can be seen in the fact 0 2 4 6 8 10 12 14 16

that the World Competitiveness Report has replaced the HICP inflation (avg over 3 yrs) variable public spending relative to GDP with an index capturing the amount of distortive public activity. It consists

of three sub-indices measuring distortive subsidies, Source: Commission services.

diversion of public funds and public trust in politicians’

fiscal honesty. See Sala-i-Martin (2003). Graph 17 also shows that the variability of consumer

138 More detailed explanation of this research, which provides price inflation is considerably higher in the new Member some kind of benchmark estimates for industrialised States than in the euro area. It was more volatile despite

countries, is given in Bassanini et al. (2001). As regards the

impact of macroeconomic variables on growth, the 139 empirical approach was a pooled-mean group estimator that 1997 is the first observation of HICP inflation for some of exploits the information content of both differences across the new Member States.

21 OECD countries and variation over time (1971-98) while 140 See Brada and Kutan (2002) and Wachtel and Kurhonen

imposing some coefficients to be uniform in all countries. (2004).

the fact that the average level of the inflation rate was Republic (12.6 per cent), Hungary (6.2 per cent), Malta not markedly different from the euro-area rate of (9.7 per cent), Poland (3.9 per cent) and Slovakia (3.7 inflation in some of them. With the dark diamonds in the per cent). Policy-makers in all of these countries are graph representing the realisation of both variables in likely to implement measures of fiscal consolidation in

the new Member States in the period 1997-99, it the next years. 142 Regarding public debt, the situation

becomes evident that both level and variability of looks better in that all countries except Cyprus and inflation have declined over time. Malta were below 60 per cent of GDP in 2003, although

The new Member States are generally expected to Hungary only marginally so. experience average inflation higher than incumbent EU Public investment in 2003 was close to or above 3 per Members because of convergence effects. Catch-up cent of GDP in all countries with the exception of Latvia growth tends to be accompanied by higher inflation in which spent only 1.6 per cent of its GDP. Public the non-tradeable sector, causing overall inflation also to investment in infrastructure with the aim of bringing it be higher. Empirical estimates of this so-called Balassato EU-15 average standards entails large costs. Samuelson effect have come up so far with very Calculations by DIW, a German research institute,

different results, depending on method, data and time

141 suggest that they amount to about € 500 billion of which period used. This variability is very likely due to the about two thirds on environment, water and energy. This fact that the central assumption of the Balassawould be more than 5 per cent of annual GDP if

Samuelson effect, namely of productivity growth investment is spread over 15 years.

primarily taking place in the tradable sector, does not

necessarily hold. Recent productivity growth has been For public expenditure, which the empirical growth driven to a large extent by the take-up of ICT in the literature considers to be inversely related to economic services sector, which consists of many non-tradeables. growth, the upper-hand panel shows that the public Nevertheless, a positive relationship between price level consumption to GDP ratio is not very different in most and income level is well documented. Cross-country of the new Member States than in the euro area. Four of analysis presented in European Commission (2002b) the countries have a ratio around 20 per cent, 3 are suggested that a 1 percentage point increase in GDP per higher and two are considerably lower than the euro capita relative to the EU-15 average would raise the area. European Commission (2002) analysed CEEC-10 price level as measured by PPS by 0.86 per cent relative budget data in 2000, taking into account as explanatory to the EU-15 average. It also cautioned that alternative variables GDP per capita, trade openness, debt level and techniques would exhibit a considerably smaller – albeit demographic variables, and compared predicted and still significant – effect. actual expenditure-to GDP-ratios in CEEC-10. Except

for Poland, Bulgaria and Latvia where the actual ratio Graph 18: Net lending in % of GDP, was higher, most of the new Member States did not

new Member States 2003 deviate considerably from their predicted ratio.

-14 -12 -10 -8 -6 -4 -2 0 2 4 Somewhat surprisingly, national accounts data suggest an inverse relationship between the compensation of

CZ public employees and public consumption across the EE new Member States (coefficient of correlation of minus CY 0.5) in contrast to the strong positive link between both LV across the EU-15 Member States (coefficient of LT correlation of plus 0.7). This suggests that countries with HU a high share of public consumption do not have a MT particular high share of public employment. A tentative PL conclusion could be that these countries have more SI flexibility to adjust public finances than euro-area SK Member States. However, more detailed analysis on a country-by-country level suggests that up to 80 per cent

new MS of government expenditure is rigid. 143 A reason is that a

Source: Commission services. larger share of public consumption falls on the consumption of collective goods, i.e. security, defence,

infrastructure, legal and political administration. These

As regards public finance, general government deficits expenditures are likely to feature scale effects. That is,

in 2003 as notified to the Commission in autumn 2004 are illustrated in Graph 18. Taking the Maastricht

criteria as benchmarks, deficits were higher than 3 per 142

cent of GDP in Cyprus (6.4 per cent), the Czech These countries have already received recommendations under the excessive deficit procedure to bring down their

deficits in the coming years.

141 For an overview of different empirical studies, see 143 Rigidity means here determined outside the budget bill

Chapter 5 in the EU E CONOMY R EVIEW 2002 and Égert et al. process. See European Commission (2003b), Part V, for a

(2004). review of key budget issues for the new Member States.

Graph 19: Public consumption in the new Member States, Graph 20: Current account deficits and net foreign direct % of GDP investment in % of GDP, 2003

25 25 14

12 CA deficit Net FDI

20 20 10

15 15 8 f G

D P D P

 o

% 10 10 6 o f

G

% 4

5 5 2

0 0 0

 CZ EE CY LV LT HU MT * PL SI SK EUR-

Collective government consumption Other government consumption 12 -2 Compensation of public employees CZ EE CY LV LT HU MT PL SI SK

Note: Malta is 2001 observation for total government consumption, no Note: No FDI graphs for Slovakia. breakdown available. No comparable data on compensation of public Source: Commission services. employees for Slovenia.

Source: Commission services.

Policy-makers have made great strides in strengthening

they are higher than in the old Member States because frameworks for financial supervision. Banking systems the new Member States still have a lower level of GDP are on average well capitalised and sizable foreign per capita but these expenditure items should grow ownership stakes have typically helped to improve

under proportional to GDP. management. Already in the transition decade, hard budget constraints were imposed on former state-owned

Most of the new Member States have similar or smaller enterprises – removing a key source of quasi-fiscal revenue/GDP levels than euro area and the average tax pressures on banks and governments. And in general the burden is also consistently lower. This does not leverage of households and corporations is low.

necessarily imply that taxes have a less distortive effect

on individual incentives in the new Member States than In the external sector, current account deficits have been in the euro are because the variable of interest is the covered significantly by foreign direct investment (see marginal tax rate rather than the average tax burden Graph 20). Short-term debt typically is well covered by shown in the graph. Absent comparable information on reserves. Monetary and exchange regimes are mostly the marginal tax rates, at least at the aggregate level, the “corner solutions” of hard pegs or qualified inflation graph can nevertheless be expected to give a good proxy targeting, reducing vulnerability to capital flows.

for tax incentives. Moreover, adjustment mechanisms in the real economy display greater flexibility than in other Member States –

Finally, the tax system is not less supportive to growth notwithstanding some rigidities that keep structural in the new Member States compared to the euro area in unemployment high – and competitiveness has been the aggregate perspective. A higher share of tax revenue quite well preserved.

falls on indirect taxes, which are perceived to be less

distortive to the allocation of income to consumption In terms of possible vulnerabilities over the medium and investment/savings than direct taxes. A qualification term, however, the discussion above of current trends in to this finding, however, results from the size of social the public finances presents a decidedly more mixed security contributions, which - though smaller relative to picture. The larger economies in central Europe, as well GDP than in the euro area - are rather large in relation to as Cyprus and Malta, have experienced sizable fiscal the tax base, leading to high rates of taxation on labour. deficits; and in a number of cases debt ratios are quite This negatively affects incentives for job-intensive high relative to income levels. This argues for a steady

growth and for work in the official economy. reduction in deficits – so that policy is positioned to respond flexibly to possible shocks. In the Baltic region

Regarding the external balance, the economies of the and in Slovenia deficits are much smaller, and three of

new Member states, at the time of accession, display these economies have already entered ERM II. 144 It

important strengths that should help ward off risks of remains important, nonetheless, to ensure that policy is instability. Nonetheless, policy-makers need to ensure free to allow automatic stabilisers to operate – and in that the process of convergence is not punctuated by particular to avoid a fiscal stimulus to demand at times external or financial sector stresses. Among the key when, as at present in the Baltics, credit growth is favourable elements in this regard are reforms that have strong.

restructured the financial sector and buttressed external positions against possible shocks.

144

  These are Estonia, Lithuania and Slovenia.

While policy-makers can feel considerable confidence Fiscal discipline facilitates the task of monetary about the present robustness of their economies, the authorities in keeping inflation under control. Some strong real and financial convergence ahead carries even argue that it represents a pre-condition for inherent risks. Experience in emerging market accomplishing price stability on a sustainable basis. In economies points to the potential stresses that can this respect, the tensions to which public finances in new emerge as financial systems expand in an open capital Member States are exposed, with the objective of account setting. Experience in some central European consolidating deficits below the 3 per cent ceiling on the economies has already illustrated the scope for financial one hand and improving infrastructure and social market exuberance to drive risk premia on external debt cohesion on the other hand, are a crucial challenge. and domestic instruments to levels lower than warranted Moreover, still high public deficits and outstanding by fundamentals. Under such circumstances, strong spending necessities also warrant a tone of caution on inflows, rapid credit growth, and buoyant asset prices whether the above painted snapshot of the favourable can lead to a cycle of real appreciation – and potential structures of public revenues is lasting. Unsustainable stresses when expectations at some point reverse. The public finances can mean crowding-out effects and new Member States in the Baltics and central Europe distortive taxation in the future. need to guard against such a cycle – given the setting of

positive credit supply shocks, rising permanent income Regarding external and financial stability, the striking

expectations, and open capital accounts. success of policy-makers in navigating the uncharted waters of transition over the past decade has left these

Policy challenges economies well-braced against external or financial

Overall, abstracting from sizable variations across the sector stresses. Nonetheless, the period ahead will bring new Member States, policy-makers have made good new challenges. To safeguard external and financial progress in establishing stable macroeconomic stability, attention needs to be paid to the interaction of frameworks, conducive to growth. Looking ahead, monetary, prudential and fiscal policy regimes, and to macroeconomic policies will remain a key focus of the ways in which these may influence the risk attention in connection with potential euro-area behaviour in the private sector. And in particular, as the membership. At least in those new Member States private sector enters a phase of strong expansion, the contemplating a concrete schedule for introducing the design of fiscal policy can play an important supporting euro, policy-makers will need to set their priorities so as role in ensuring that imbalances are limited and that to achieve the nominal convergence criteria set by the private sector confidence is maintained.

Maastricht treaty.

Where inflation has come down markedly, it will be 3.2.3 Institutional quality

crucial to keep it at a low level. The still high variability There is a growing emphasis in the economic literature of inflation indicates that inflation expectations might on the role of institutions for long-term economic not yet have followed the downward trend in actual developments. The IMF and the World Bank are inflation. Therefore, inflation surprises will challenge increasingly focusing on the role of institutions in their monetary authorities. On the one hand, ERM II could strategies of macroeconomic stabilisation and poverty

provide an external anchor for the credibility of reduction. 145 In Europe, the most prominent example of

monetary policy. On the other hand, catch-up in price attention given to the quality of institutions are the solevels, wages and growth may cause temporarily higher called Copenhagen criteria (political, economic and rates of inflation in some of the countries. In these cases, legislative) which candidate countries have to fulfil in keeping the option of a more flexible adjustment of order to become members of the EU. exchange rates for some time could be conducive to

securing external competitiveness. In general, institutions are defined as the "rules of the game" which can be formal and informal rules,

Budgetary deficits in most new Member States are still enforcement mechanisms and organisations. 146 Policies

much higher than the 3 per cent benchmark enshrined in should aim at efficient institutions by ensuring the rule the Treaty. The experience of some of the current euroof law in order to avoid unclear property rights, area Member States taught that stringent budgetary providing a well-functioning administration and consolidation can bring deficits down quite quickly. The integrating markets by reducing trade costs. Hence, the new Member States, however, differ in two important public sector has a crucial role to play in providing the respects. First, they are faced with the need to build up conditions for a functioning market economy by and modernise their infrastructure. However, public guaranteeing the exclusivity of private property rights, investment does not necessarily need to be financed by in particular by fighting crime and corruption, and by budget deficits. Second, economic restructuring is an reducing the costs of trading property rights, in ongoing process in the new Member States and may

require the use of public spending to cushion adjustment 145 costs by compensating the losers of structural change For a more extensive overview of how institutions and

and economic reforms. political factors impact on economic growth, see IMF (2003a) and Borner et al. (2004).

146

  See North (1990).

particular through the provision of macroeconomic solution is that most of the empirical measures for stability, good infrastructure, clear legal procedures for institutional quality tend to be based on opinion polls the enforcement of contracts etc. Efficient institutions and expert surveys about the data. The advantage is that are essential for economic development since they it reflects economic actors’ perception of the actual provide incentives for private agents to fully benefit institutional quality rather than the legal or social norms

from the investment and production of goods and to that govern institutions. 150

trade them with those who value them most.

Furthermore, without well-defined private property In spite of these data problems, the empirical results are rights, financial intermediation of savings and very robust. The main outstanding question from this investment and hence the accumulation of capital do not line of research is not whether institutions are important

function smoothly due to a lack of collateral. 147 for growth but how important they are. While some authors conclude that “the quality of institutions trumps

A fundamental problem for both theoretical analysis and everything else” others consider geographical variables empirical research on institutional economics is as equally important and stress the interaction between

endogeneity or the direction of causality, i.e. whether “institutions, policies and geography”. 151 Applying the

income is high because of good institutions or whether IMF (2003a) methodology to the World Bank data for institutions are good because a high-income country can the EU Member States, candidate countries and other

148

better afford to have them. Empirical research in this industrial economies, shows the expected positive area is quite recent and has only become possible after relationship between the quality of institutions and GDP different researchers and institutions had compiled data per capita (Graph 21). Here, the quality of institutions is on institutions and governance across countries, which able to explain about two third of cross-country has allowed for the empirical backing of the importance variation in growth. By adding further control variables, of institutions for economic activity. the more sophisticated econometric approach used in

The most comprehensive database on institutions and IMF (2003a) is able to explain three fourth of the governance currently available has been established by variations in cross-country growth regressions. It also

the World Bank. It covers six different variables for governance for 199 countries (including the new

Member States) and observations for four points in time Graph 21: GDP level and the quality of institutions, industrial economies

(1996, 1998, 2000 and 2002). The raw data is from surveys and opinion polls carried out by various different organisations (international organisation, risk 52.5

rating agencies, think-tanks, NGOs). 149 The main 47.5

variables, of which some are apparently more of 42.5 importance for developing countries rather than for the 37.5

EU, are voice and accountability, political stability and ita ap 32.5

absence of violence, government effectiveness, 27.5

regulatory quality, rule of law, and control of corruption. p

er c

22.5

G D

P

Despite the progress made in data collection on the 17.5 quality of institutions, this data is not without problems 12.5

y = 14.147x + 4.7983

R 2 = 0.6634

and results should be regarded with a certain caution. 7.5 For instance, it is well known that institutions are 2.5

context-specific because they develop on a longer-term 1 1.2 1.4 1.6 1.8 2 Average of institution indicator

historical and cultural background. It is also

questionable whether the legal characteristics are Source: World Bank, IMF, Commission services.

sufficient to describe the actual impact of institutions. A

147 See Bassanini et al. (2001, Annex) or Romer (2001, Chapter 150

  A potential drawback of this approach is that survey results

3.11) for two alternative approaches how institutions could could be culturally biased, which reduces the extent to be integrated in traditional growth models. which policy recommendations can be drawn from cross

148 Taking corruption as an example it is both difficult to country analyses or case studies. Kaufmann et al. (2003),

imagine that a country with a high degree of corruption can who built up a data set for the World Bank, found little achieve a high level of income given the disincentives to evidence of ideological biases in the assessment of

invest and that a poor country could afford public service corruption in the surveys.

wages that are sufficiently high to reduce the incentive to 151 See Rodrick et al. (2002) for the first and Sachs (2003) for take bribes. Moreover, due to a less developed system of the latter quote. Easterly and Levine (2003) find that control and justice, the risk of being discovered or institutions matter most for the long-term level of income sanctioned for taking or giving bribes may be lower. whereas geography and policies do not if their effects on

149 A detailed explanation of data and methodology is given in institutions are controlled for. Dollar and Kraay (2003) give

Kaufman and Kray and Mastruzzi (2003). The data itself is evidence of a strong effect of trade on growth and a much

published on the World Bank website. smaller role for improvements in institutions.

finds that improving the institutional quality by 1 the rest of the world. The gap between East and West is standard deviation would raise GDP per capita by 1.4 lowest for the sub-indicator of political stability, which, percentage point. according to Table 11, has the smallest relevance for

Identifying the institutions that have the largest impact economic activity and quality of other institutions,

on economic activity is difficult. The first column in the respectively.

table below displays that coefficients of correlation with An advantage of the World Bank data set is that it the World Bank indices of institutions are all high, allows tracing developments over time, although only except for the one that captures political stability. for a small period. Whereas institutions are usually seen Moreover, each indicator is strongly correlated with the as rather invariant over time, this might not be true for other indicators, again with the exception of political the new Member States, which had undergone huge stability. This probably reflects the importance of political, economic and social transformation in the interactions among institutions, i.e. good institutions in 1990s. Since the indicators of institutional quality are one field are supportive to the quality of institutions in derived from experts’ or citizens’ perception of the another field. institutions and this perception can reasonably be expected to adjust with a lag to actual improvements in Table 10: Correlation of institutional variables quality, it could be telling to consider the improvements

with GDP level (2002, PPS) measured between 1996 and 2002. 153 Given the kind of 35 ind. measurement, all improvements in the indicators are not

countries 25 EU MS in absolute terms but relative to all the other countries in

Voice and accountability 0.77 0.59 the panel. Political stability 0.48 0.28

Government effectiveness 0.87 0.73 Graph 23 illustrates the enormous progress the new Regulatory control 0.79 0.66 Member States made. Between 1996 and 2002 they were Rule of law 0.88 0.72 able to improve the quality of their institutions - as Control of corruption 0.85 0.70 assessed by citizens and experts - by 0.3 standard

Source: IMF, World Bank, Commission services. deviations. Assuming, for both simplicity and

illustration, that the convergence process is linear and

The situation in the new Member States

Eight of the ten new Member States underwent the Graph 22: Quality of institutions in the old and new Member States, 2002

transition from a central planning to a market economy within a very short time. They experienced an immense

deterioration in living standards in the early phase of 1.8 transition which the economic literature often attributes 1.6 EU-15 avg

to the “institutional collapse”, in particular the lack of 1.4 AC-10 avg

152

market-oriented legal structures. It is now tempting to 1.2 1

relate the strong growth, which some of the new 0.8 Member States witnessed over the past years, to their 0.6 progress with institutional reforms. Although the 0.4 imprecision of measurement described above requires 0.2

some caution in cross-country comparisons of 0 institutional variables, it allows a broad snapshot of the y ilit ss ne la w

perception of how efficient institutions work in the new nt

ab ta

bil ity

ive qu

al ity of

or ru pt ion

ffe ct or y le Ru

r c

 a cc ou liti ca

l s

Po t e

Member States relative to the EU-15. gu

lat l fo a nd Re nt ro

ice

Vo ve

rn m

en Co

Go

While the average is lower for the new Member States

than for the EU-15 in all categories and particularly Note: The graph shows the average of EU-15 and AC-10. The scale is

lower for the three sub-indicators of effectiveness of expressed in standard deviations with zero being the average of 198

governments, rule of law and control of corruption, there countries. Source: World Bank.

is at least one new Member State in each category that performs better than the lowest ranked Member State in the EU-15. It is also apparent that the gap between the EU-15 average and the new Member States is particularly large in those categories where the EU-15 has a high rank, i.e. performs especially well relative to

152 For an overview of the economics of transition, see Campos 153 For the sample of all countries, there is no evidence that

and Coricelli (2002), Svenjar (2002) and, for a review of the there could be a negative relationship between the quality of determinants of enterprise restructuring during transition, institutions in 1996 and the improvement in the quality

Djankov and Murrell (2002). between 1996 and 2002.

Graph 23: Catch up of institutional quality to EU-15 Graph 24: GDP growth and the improvement in institutional Change 1996-2002 in % of the gap to EU-15 in 1996 quality in the new Member States

Standard deviations

0 0.1 0.2 0.3 0.4 0.5 0.6 8

Control for corruption .

 p .a 7 LV

Rule of law g % 6 EE

; av LT

Regulatory quality 02 5

20

6- 4 PL HU Government effectiveness SI

SK

 1 99 3

Political stability w th MT

2 CY

Voice and accountability CZ D P

 g ro

y = 4.5036x + 2.7345

 G 1 2 al R = 0.2625

0 10 20 30 40 50 60 70 80 90 re 0

% of gap in 1996 to EU-15 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Relative to gap to EU-15 in 1996 in % (lower scale) Absolute catch up in s.d. (upper scale) Improvement in the quality of institutions 1996-2002, sd

Source: Commission services. Source: Commission services.

that the trend continues in the years ahead at same visible for the other nine new Member States. Excluding speed, the gap to the EU-15 would be closed in 12 to 15 Malta from the sample would yield a significant

years. 154 relationship that suggests growth gains from the

When translated into terms of the gap to the EU-15, the improvements in institutions far higher than calculated

improvement in the perceived quality of institutions by the IMF researchers.

means that more than 70 per cent of the gap in 1996 was In a panel of the ten Member States and the six subclosed between 1996 and 2002 for the sub-indicators of categories for the quality of institutions, the political stability and regulatory quality. Somewhat less improvement of only two sub-categories displays a assuring is the observation that catch-up has been more strong positive correlation with GDP growth. These are limited in government effectiveness, rule of law and the effectiveness of government (0.85) and control of especially control for corruption. These are categories corruption (0.74). Cross-country variations in the other where the EU-15 fares very well compared to the rest of variables are not correlated with differences in the the world as evidenced by a high level of these growth performance across the new Member States or,

indicators in the EU-15 Member States. 155 This very surprisingly, even weakly negative for the case of

somewhat puts into perspective the observation of less improvements in regulatory control. This might be due progress made in the new Member States. to the time lags with which improvements in regulation

Taking the IMF estimates quoted above at face value, usually impact on economic activity.

156

this improvement of the institutional quality by 0.3 When analysing how the new Member States managed points has contributed to raising average annual GDP to improve the quality of their institutions, the literature

growth in the new Member States by 0.4 percentage unanimously points to the role of EU integration. 157 In

points. Applying a more simplistic view, Graph 24 this context, at least three factors were important. First, relates the variations in the improvement in institutional accession to the EU required the adoption of the acquis quality across the new Member States to their average communautaire, i.e. the direct import of legislation that growth performance 1996-2002. While the slope is has advanced integration between the old EU Member positive, it is borderline significant at the 10 per cent States. EU pre-accession funding from the Phare

level. The graph shows that the observation for Malta instrument spent considerable amounts on institutionapparently

interferes with the stronger relationship building to help achieving the accession criteria (see

Box 1). Secondly, the accession process provided an

external anchor for policy makers’ constraints and

incentives, which helped overcoming domestic obstacles

to reform. Thirdly and related, the path towards EU

accession brought to the fore the importance of

154 In another project on indicators of “Doing Business” including 145 countries, the World Bank (2004) notes the 156 According to the literature on the credibility of monetary

reform progress in 2003 of Poland, Lithuania and Slovakia policy, which can be applied to this case, the track record is which brought the latter two countries into the top 20 a more important determinant of reputation than economies on the ease of doing business. announcements. It is therefore reasonable to expect changes

155 It should be noted, however, that the US records a still in the perception of the quality of institutions to materialise slightly better assessment of their quality in the institutions with a lag only.

in these three categories, though the distance is small. 157 For a review, see IMF (2003a), Box 3.2.

stimulating openness, competition and an administrative in structural reforms (“Cardiff process”) in the context environment supportive to business activity. of the business environment. Sanctioning mechanisms

While the path towards EU accession seems to have here are relatively weak and mainly based on exchange favourably impacted on institutions and economic of best practice and peer pressure.

performance, it remains to be seen which factors could It could therefore be useful to reflect how to reinforce stimulate further progress in the quality of institutions. mechanisms at the EU level which could serve as In this context, economic surveillance within the EU external anchors, to help further improve the quality of could play an important role, substituting the external institutions in the new Member States. A first option anchor of EU accession by the one of peer pressure and could be to widen the scope of monitoring structural best practices. IMF (2003a) points to some fundamental reforms to aspects covering institutional quality. A factors that have proven conducive to institutional second option is to use existing instruments in the EU reform. These are openness to trade, stronger cohesion policies to reorient them to the performance of competition, information and higher transparency. institutions or the implementation of recommendations Ownership of and commitment to reforms are in that area. Finally, spending on institution-building, as considered overriding determinants of progress with the under the pre-accession instrument Phare, should also quality of institutions. have more importance in Structural Funds programmes.

Policy challenges It would seem preferable to implement the link between

The available literature points to a strong link between institutions and EU policies as an incentive to make the quality of institutions and catching-up or GDP per further progress, and as an attempt to improve the capita levels. For a number of indicators of institutional efficiency of EU funding - not as a sanction. The most quality and in spite of impressive progress between 1996 difficult part will be to agree on indicators for the and 2002, the new Member States still have considerable quality of institutions which are not contestable and gaps compared to most old Member States, in particular based on sound methodology. When implementing with a view to the efficiency of public administration recommendations to improve the quality of their and judiciary. The preparation for EU accession as an institutions, Member States should be able to make use external anchor is the most frequently used explanation of the important function of the EU as an external for the progress in institutional reforms in the new anchor which allows more courageous reforms than

Member States (see Box 1). mere within-country political forces would do. However, it should be taken care that the focus should

Further progress in reforming institutions will be of only be on the "function" and not on the "form" of major importance for the new Member States’ process institutions (Rodrik et al. 2002) since there is broad of catching-up. To the extent that it is difficult to carry agreement in the literature that universally good out institutional reforms on a purely domestic political institutions do not exist. Institutions are context-specific basis, the disappearance of the EU membership “carrot” and therefore depending on the historical, cultural and prompts a question whether comparable new external political background of a country or region. When anchors have become available after accession. building institutions, norms and culture as well as Mechanisms of Community law and of economic policy existing institutions need to be taken into account. coordination could be thought of as possible substitutes Therefore, the World Bank (2002) holds that "best after accession. However, the Treaty is relatively silent practice in institutional design is a flawed concept" and on what is considered here as institutions and mostly suggests four key approaches to institution-building: based on co-operation between Member States rather complement what exists, innovate to identify institutions than on Community procedures. Economic policy that work, connect communities through information coordination addresses issues of institutional quality flows and trade, and promote competition. only marginally when it comes to assessing the progress

Box 1: Institution-building in the pre-accession process

The Copenhagen European Council in June 1993 concluded that “membership requires that the candidate country has achieved stability of institutions guaranteeing democracy, the rule of law, human rights and respect for and protection of minorities, the existence of a functioning market economy as well as the capacity to cope with competitive pressure and market forces within the Union. Membership presupposes the candidate's ability to take on the obligations of membership including adherence to the aims of political, economic and monetary union.” Since 1998 the European Commission publishes annually a Regular Report on each candidate country’s progress towards accession to provide an assessment of progress in meeting these political, economic and acquis criteria for accession.

Regarding the economic criteria, the existence of a functioning market economy requires that prices, as well as trade, are liberalised and that an enforceable legal system, including property rights, is in place. Macroeconomic stability and consensus about economic policy enhance the performance of a market economy. A well-developed financial sector and the absence of any significant barriers to market entry and exit improve the efficiency of the economy. The capacity to cope with competitive pressure and market forces within the Union depends on the existence of a market economy and a stable macroeconomic framework, allowing economic agents to make decisions in a climate of predictability. It also requires a sufficient amount of human and physical capital, including infrastructure. State enterprises need to be restructured and all enterprises need to invest to improve their efficiency. Furthermore, the more access enterprises have to outside finance and the more successful they are at restructuring and innovating, the greater will be their capacity to adapt. Overall, an economy will be better able to take on the obligations of membership the higher the degree of economic integration it achieves with the Union before accession. Both the volume and the range of products traded with EU Member States provide evidence of such integration.

With the objective of supporting the achievement of the Copenhagen criteria, the Commission regularly adopts "Accession Partnerships" which provide an assessment of the priority areas in which the candidate country needs to make progress in order to prepare for accession and on the basis of which "National Programmes for the Adoption of the Acquis" provide a single framework for the programming of the EU pre-accession financial instruments. These include the "Phare" Programme to finance institution-building and for investment to help adopt the acquis, "ISPA" for the financing of large infrastructure projects in transport and environment sectors, and "SAPARD" to support agricultural and rural development. Community assistance for financing projects through these three pre-accession instruments is conditional on respect of commitments under the Europe Agreements, further steps towards satisfying the Copenhagen criteria and progress in meeting the specific priorities of the Accession Partnership. The financial framework 2000-2006 makes available €3.12 billion (in 1999 prices) per year for all three financial instruments of which about half is spent on Phare projects. The main objectives of Phare are to strengthen their public administrations and institutions to function effectively inside the Union, to promote convergence with the European Community’s extensive legislation and reduce the need for transition periods, as well as to promote economic and social cohesion (also to prepare the transition to Structural Funds). Around 70 per cent of Phare resources are allocated for investment in the regulatory framework and for economic and social cohesion, while approximately 30 per cent is being used to meet institution-building

needs.

• 4. 

  What can be the contribution from EU cohesion policy. No causality can be inferred from either

cohesion policy? the occurrence or the lack of convergence or from its speed which may result from many economic, social and

The EU spends about one third of its budget on policy factors other than the EU assistance.

supporting cohesion by assisting Member States and Two main methods have thus been adopted to assess the their regions in efforts to promote catching-up (see direct effect of the EU cohesion policy: model Box 2). Though the policy has a distributive dimension, simulations and econometric growth regressions evidenced by significant net transfers to the poorer incorporating the amount of cohesion funding as an

Member States, it aims primarily to enhance efficiency explanatory variable among other variables.

and growth. The goal of the Structural and Cohesion

Funds is precisely to support the main determinants of A variety of macroeconomic models, based on different catching-up highlighted above in Section 3. A number of theoretical foundations, have been used to assess the questions will help to identify the potential contribution impact of the Structural Funds. The Commission mainly of EU funds to the catching-up of the new Member relies on two combined demand-side and supply-side

States: Have Structural Funds contributed to real models, QUEST II and Hermin.

 159

convergence in Europe? What are the conditions under which they have an impact on growth and employment? How will and how should Structural Funds be changed?

4.1 Evidence of structural funds impact

159

Some authors have criticised the Structural Funds as For other models see e.g. Pereira (1994) or Bourguignon, having - if any - only a marginal impact on real Lolos and Zonzilos (1995). A complementary approach

convergence in Europe. 158 However, most of these used by the Commission is a predominantly demand-side

studies use growth regressions subject to model based on input-output techniques allowing to assess how Structural Funds expenditures affect the structure and

methodological, econometric and data weaknesses. level of final demand including investment and induce

Moreover, the role of the Structural Funds is, in essence, changes in imports, value-added, labour and capital use. See

to co-finance investments in physical and human capital, Beutel (2002).

using financial means coming mainly from other

economies. EU regional policy should therefore be

expected to have a positive impact on growth and

employment in the recipient regions and Member States.

Standard growth regressions testing for absolute or conditional ß-convergence cannot as such provide any evidence on the impact and effectiveness of the EU

158 See e. g. Boldrin and Canova (2001), Ederveen and Gorter

(2002) or Midelfart (2004).

Box 2: What is the EU cohesion policy?

In the less developed regions of the EU (“Objective 1”), the EU Structural Funds co-finance programmes in the fields of physical infrastructure, human resources development as well as aid to the private sector. Structural Funds also support the conversion of areas facing structural difficulties ("Objective 2") and policies and systems of education, training and employment outside Objective 1 regions ("Objective 3"). For the EU-15 in the period 2000 to 2006, about 70 per cent of the €195 billion (at 1999 prices) are allocated to Objective 1 regions. In the so-called "cohesion countries" (Greece, Spain, Portugal and, until 2003, Ireland), whose gross national income per capita is below 90 per cent of the EU average, the EU Cohesion Fund finances projects on the environment and on trans-European transport networks and has a volume of €18 billion (at 1999 prices) in the period 2000 to 2006. The Structural Funds and Cohesion Fund together have a certain macroeconomic importance in some countries, peaking at levels of around 3 per cent of GDP in Greece and Portugal at the end of the 1990s. The European Council meeting in Copenhagen in December 2002 decided, and this was later inserted into the Accession Treaty and implemented in programmes, that the 10 Acceding Countries would benefit from €14.2 billion of Structural Funds and €7.6 billion of Cohesion Fund (at 1999 prices) from 2004 to 2006. In addition, a transitional sub-heading on institution-building measures of €380 million has also been agreed. The draft Framework Regulation for the new programming period starting in 2007, adopted by the European Commission in July 2004, aims at reinforcing the financial focus on real convergence, thematic concentration and further simplifying the

management systems.

QUEST II embodies a neo-classical-Keynesian ending in 2000 according to the assumption made for the

synthesis. While in the short run the model is influenced calculation.

by standard Keynesian features, the behavioural

equations are based on microeconomic principles of The results of the ex-ante macroeconomic evaluations intertemporal optimising behaviour of households and for the new Member States are not easily comparable as firms and the supply side of the economy is modelled the applied methodologies are heterogeneous. However,

explicitly via a neo-classical production function. 160 they also show a substantial impact. In Poland, for example, according to the Hermin model’s impact

Hermin is basically a neo-Keynesian model with some assessment, real GDP would be higher in 2010 by

161

neo-classical features in the supply-side. The model approximately 1 per cent due to the support provided in

attempts to capture the effect of public investment by the period 2004 to 2006.

incorporating the beneficial externalities associated with

increased stocks of infrastructure and human capital.

The elasticities used are taken from existing empirical

studies.

The Hermin results of the ex-post evaluation for the last programming period (1994-1999) identify their continuing supply-side effects by assuming that funding terminates after the programming period. The results for the cohesion countries (see Graph 25) range from a relatively modest long-term impact in the cases of Greece and Spain to a real GDP level in Portugal that is more than 2 per cent higher in 2010 than in the absence of Structural Funds and national co-financing, both

160 Real interest and exchange rates are determined

endogenously. Thus, the initial positive effect of the cohesion policy through an increase in the public capital stock may be reduced by a temporary crowding-out of private investment. In the longer run, the increase in GDP is higher than the induced short term demand effect due to positive supply-side effects which continue beyond the period of aid payments. For a description of the model, see Roeger and in’t Veld (1997).

161 Two sectors are modelled behaviourally: a tradable sector

(manufacturing) and a non-tradable sector (market services). Output is primarily driven by world demand and cost and price competitiveness in the former and by final demand in the latter. Wages are determined in the traded sector in a bargaining model and are sensitive to the tax wedge, unemployment and productivity. Expectations are auto-regressive and interest and exchange rates are exogenous to the model. For a description of the model, see Bradley et al. (1995).

Graph 25: Impact of the European regional policy, programming period 1994-99

5 5

4.5 Greece Expenditure 4.5 Ireland Expenditure

4 Greece GDP 4 Ireland GDP

3.5 3.5

3 3

2.5 2.5

2 2

1.5 1.5

1 1

0.5 0.5

0 0

1993 1995 1997 1999 2001 2003 2005 2007 2009 1993 1995 1997 1999 2001 2003 2005 2007 2009

5 5

4.5 Spain Expenditure 4.5 Portugal Expenditure 4 Spain GDP 4 Portugal GDP

3.5 3.5

3 3

2.5 2.5

2 2

1.5 1.5

1 1

0.5 0.5

0 0

1993 1995 1997 1999 2001 2003 2005 2007 2009 1993 1995 1997 1999 2001 2003 2005 2007 2009

Note: Bars: CSF expenditure as percentage of GDP of the programming period 1994 – 1999, i.e. under the assumption of ending support in 2000; lines: CSF induced change of GDP level against baseline in percent. Source: European Commission 2003d.

Model simulations all conclude that cohesion support cohesion assistance are plagued with methodological, contributes significantly to growth and employment at econometric and data weaknesses. No structural model national and, when analysed, at regional level. The of such a complex mechanism as growth can be magnitude of the impact may vary depending on the represented by a single equation linking the former to model specifications, the economy’s characteristics, the one variable i.e. the amount of Structural Funds transfers amount of assistance and the types of public investments as done in Boldrin and Canova (2001) or two variables targeted. Modelling has two main advantages. It shows if initial income per capita is also considered. Such how the policy affects the demand and supply sides of regressions are not exempt from econometric problems. the domestic economy depending on a wide range of For instance, since the beneficiaries of EU cohesion other factors and allows for a counterfactual (i.e. without policy are poor economies, the amount of EU assistance policy) situation. On the other hand, simulations tend to works as a proxy for the omitted variables that assume that cohesion support is fully turned into presumably explain why they have below average

productive public investment, overlooking possible incomes. 163 As a result, the estimated coefficient on the

weaknesses in policy delivery. They may thus assess the volume of aid is negative while the inclusion of potential rather than the actual impact of the cohesion additional variables in the equation, even in a simple

policy. 162 form, leads to a positive impact of EU assistance on

Econometric regressions would be expected to give a growth.

 164

better “ex-post” assessment. However, attempts to link national and regional GDP or productivity growth to

163

  See de la Fuente (2003).

164

  This is illustrated by Ederveen et al. (2001) Their results, at

162 See Ederveen et al. (2002). NUTS II regional level for the period 1981-1996, suggest a

In other words, imposing the assumption of absolute • Rules on financial management and control help to

convergence creates a downward bias on the estimated improve the quality and efficiency of public

impact of cohesion support while it can be significant administration. 170

and positive if convergence is only conditional, which

seems to be the consensus view today. • Inter-regional and international exchange of good practices for regional policy can be a helpful tool

In addition, such regressions, when performed at for better targeting and a more efficient regional

regional (NUTS II) level, are faced with acute problems policy.

of data availability and reliability. Not only is the bulk

of cohesion support national or transregional and thus EU Structural Funds can thus have an important impact difficult to attribute to regions. But available statistics not only on the efficiency of regional policy, but also on hardly allow controlling for other factors that can national administration and overall public spending in

influence growth. the corresponding Member States.

Against this background, results have to be considered

with caution as they are very sensitive to the different 4.2 Conditions for maximising the impact

methods, time periods and data sets on which they are Several of the above mentioned studies give also based. With few exceptions, 165 most econometric studies interesting insights on the conditions that can affect the tend to find a significantly positive effect of cohesion effectiveness of EU cohesion policy. If public support on national growth and convergence. 166 At the investment has an impact on productivity and growth regional level, across the EU and in some case within and a leverage rather than a crowding-out effect on countries, many studies also identify a positive private investment, EU cohesion policy can be expected impact. 167 under both the neo-classical and the endogenous growth

models to be effective since it adds to physical and

In addition to their impact on growth and convergence, human capital stocks and promotes technological the implementation methods of Structural Funds have an progress. There is nevertheless a range of factors that effect on governance, i.e. they improve the efficiency of could hamper such effectiveness. Some factors may go public administration and public expenditure: beyond the control of policy-makers. Others, such as • The bottom-up approach and the partnership domestic policies and the design of the development principle between all actors involved allow strategy co-financed by the EU may, however, be

programmes to better reflect the real needs in the targeted for improvement.

regions. In view of the very limited budgetary means of EU

• The set-up of an integrated development strategy in cohesion policy, representing less than 0.5 per cent of a multi-annual framework enforces the planning the EU-15 GDP, the following conditions can be capacity and strategic thinking for regional identified to be important for a significant impact: First,

development. 168 sound and supportive national policies, including macroeconomic policies, national regional policies and

• The introduction or strengthening of the monitoring good governance, are an essential precondition for the

and evaluation culture leads to a more efficient achievement of a real impact. Second, the scarce selection of projects and a better targeting of financial means must be concentrated spatially, i.e. on

spending. 169 the poorest Member States and regions, and two issues

have to be considered: (a) whether to concentrate on national growth or on equalising living conditions across the country and (b) whether to focus on growth poles and cluster or target more dispersion of economic activity. Third, the strategic design of Structural Funds

negative impact of the cohesion policy when other factors programmes must allow for a concentration on those

than initial productivity and cohesion support are not types of expenditures most likely leading to growth and

controlled for. When they are, the estimated impact is employment. Fourth, ways have to be found to achieve positive and significant. An additional amount of cohesion the most effective use of EU Structural Funds.

support of 1 per cent of GDP leads to an annual increase in GDP per capita of 0.7 per cent.

165 See e.g. Ederveen and Goerter (2002). 4.2.1 The role of national policies

166 See e.g. Bosca et al. (1999); Garcia Solanes and Maria Since the effects of Structural Funds depend to a large

Dolores (2001); Beugelsdijk and Eijffinger (2003). extent on triggering additional private investment, a

167 See e.g. Fayolle and Lecuyer (2000); Garcia Solanes sound and supportive national economic and political (2001); de la Fuente (2003), although some do not find a environment can be regarded as a necessary condition

positive impact such as Boldrin and Canova (2001); Basile for maximising the impact of Structural Funds. In this

et al. (2002).

168 See e.g. Fitzgerald (1999).

169 See Fitzgerald (1999) and Barry (2003). 170 See Barca (2003).

context, the importance of the national political mechanisms reducing labour costs and improving environment has three main aspects: flexibility (and mobility) in the labour market. They can

• Macroeconomic and regulatory framework, translate into concrete requirements e.g. on the pursuit of labour market reforms including the obligation to

• National regional policies, and report to the Commission on progress and results.

• Governance including the administrative capacity. Besides the macroeconomic environment, an effective

national regional policy is needed for the achievement of

In the general and country-specific recommendations of real convergence between European regions. Regional the Broad Economic Policy Guidelines (BEPGs) several policy instruments used by the Member States can be countries have also been given specific classified mainly into two categories: on the one hand recommendations concerning their regional labour instruments with a rather redistributive character, aiming market. In particular, measures allowing wages to better at an equalisation of public finance resources or living reflect productivity and skill differentials would conditions among regions; on the other hand pro-active facilitate the attraction of investment flows into higher policy measures aiming at achieving economic unemployment areas. However, the 2004 report on the development in the poorest regions. However, even if a implementation of the 2003-2005 BEPGs indicates that “tendency for the policy focus to shift to wealth creation progress made by the EU-15 Member States remains from wealth distribution” can be observed, 171 national insufficient and uneven. regional policies, if compared with the pro-active design Empirical studies show that a sound economic-political of EU Structural Funds, are still rather redistributive in environment not only increases the growth and nature (for a discussion on the investment mix of employment perspectives of the corresponding country Structural Funds see Section 4.2.3). In Germany, for and its regions, but is also crucial for the effectiveness of example, estimates on the gross transfer to eastern international support. Based on an econometric analysis Germany arrive at € 116 billion in 2003 and net transfers including a set of policy indicators into a neoclassical representing nearly one third of eastern German GDP. growth model, Burnside and Dollar (2000) find that “aid The main part of these transfers is redistributive as has a positive impact on growth in developing countries transfers via the social security system or unconditional with good fiscal, monetary, and trade policies but has grants represent 45 per cent and 21 per cent of gross little effect in the presence of poor policies”, concluding transfers respectively. In contrast, only 9 per cent of that “aid would be more effective if it were more gross transfers are spent for support to the private sector systematically conditioned on good policies” (p. 847). and 13 per cent for infrastructure investment.

172 Also in

While EU Member States are not comparable with the other Member States like Spain a mix of fiscal transfer developing countries analysed, the underlying idea schemes and active regional policy exists.

173

  Active

remains valid and is supported by other empirical regional national policy has in some Member States studies. Drawing on Burnside and Dollar (2000), shifted its focus from large infrastructure investments Ederveen et al. (2002) perform cross-country regressions and sectoral state aid to selected large enterprises with panel data for 13 EU countries and 7 year-periods towards more human resource development (HRD) and from 1960 to 1995, based on a standard neo-classical technological progress related projects.

174

  Nevertheless,

growth model as introduced by Mankiw, Romer and even if expenditures are dedicated to an active regional Weil (1992). Testing only part of the Structural Funds, policy, this does not automatically mean that the projects they find a non-significant impact. The result is directly impact on growth and employment.

175

markedly different when they introduce a variable that A further factor of crucial importance for the impact of proxies openness; the interaction is significantly Structural Funds is a sound institutional and public positive. Similar results are obtained with some administrations environment. One of the expected variables which proxy the institutional context, namely a effects of Structural Funds is the improvement of the corruption perception index and an index of institutional administrative capacity due to capacity-building quality. These results, in line with previous studies on measures and the introduction of corresponding the determinants of long-term growth, tend to confirm legislations. This is of particular importance to the new that the effectiveness of the cohesion policy is highly Member States as first their institutional quality is in dependent on the growth-orientation of national policies. general poorer than in the old Member States (see EU Structural Funds have to a certain extent internalised Section 3.2.3) and second because they still have to some of the implications: First, the payments of the adapt to the management system of the Structural Funds

Cohesion Fund are conditional upon sound public finances. Second, a reference to the key role of national

policies for the impact of Structural Funds has been 171 See Yuill and Wishlade (2001).

introduced, in particular, in the programming documents 172 See Institut für Wirtschaftsforschung Halle (2003).

2004-2006 for the new Member States. These include, 173 See e.g. Davies and Hallet (2001).

inter alia, macroeconomic stability, the continuation of 174 See e.g. for Italy IMF (2003a).

privatisation and restructuring, a reduction and re 175

orientation of state aid, the implementation of See e.g. Wurzel (2001); on parallel issues on EU regional policy see Section 4.2.4.

as most incumbent Member States have done more than focus and on the other hand, that at the same time a decade ago. Consequently, guaranteeing a substantial relatively rich countries, well above the EU average, absorption of the Structural Funds can be seen as one of also receive substantial Structural Funds support. This the crucial challenges for the new Member States (see has led to strong criticisms and proposals to grant also Section 4.2.4). Structural Funds only to poorer Member States, while

comparatively rich Member States should support their

4.2.2 Achieving spatial concentration poor regions by own financial means and reduce their contributions to the EU budget accordingly. 176

For the achievement of a significant impact on

convergence in Europe, cohesion policy has in the first Enlargement has not only increased the diversity within place to concentrate its scarce financial means on those the EU substantially but also the average level of GDP regions and Member States most in need. In addition, per head has decreased statistically by nearly 10 per two strategic decisions have to be made: first, addressing cent. Consequently, the need to spatially concentrate national growth or trying to increase growth in poorer Structural Funds has become even more urgent with the regions; second, trying to support concentration through accession of ten countries that have income levels below clusters and growth centres or dispersion of economic – and often far below - the EU average.

activity in areas of slow growth. An additional effect of accession is that some regions in

Eligibility criteria for Cohesion and Structural Funds try EU-15 Member States having a GDP per capita in PPS to achieve a spatial focus on those regions and Member below the ceiling of 75 per cent surpass this threshold States in need. While the Cohesion Fund is supporting when measured against the EU-25, exclusively due to Member States having in the reference period a Gross the inclusion of poorer Member States. On the one hand, National Income (GNI) per capita in Purchasing Power it can be argued that their economic situation has not Standards (PPS) below 90 per cent of the EU, some 65 changed through the purely statistical effect and per cent of Structural Funds (SF) are allocated to the therefore support has to be continued. On the other hand, poorest, so-called Objective 1 regions with a GDP per allocation of scarce financial means requires capita in PPS below 75 per cent of the EU average. Over prioritisation and Structural Funds should favour only the period 2000 to 2006, Structural funds transfers to the poorest, i.e. nearly exclusively new Member States EU-15 Objective 1 regions are equivalent to € 127.5 and their regions.

billion at 1999 prices (€18.2 billion p. a.), amounting According to the Kuznets-Williamson hypothesis (see approximately to 2.3 per cent of GDP in Portugal, 2.2 Section 2.3.2) the possibility of an equity/efficiency per cent in Greece and 0.9 per cent in Spain. trade-off exists. Particularly in earlier stages of a

country’s catching-up process the maximisation of

Table 11: GDP per capita (EU-15=100) and national growth can be accompanied by a (temporary) Structural Funds (all objectives average 2000-2006) rise in regional inequalities as economic growth is in % of GDP driven by only few growth poles. Current experience of

the new Member States supports this argument as

GDP SF national growth in these countries seems to be largely EL 67 2.9 localised in the most dynamic areas around the capital PT 71 2.9 cities and other major agglomerations where investment, ES 84 1.4 including public investment, is likely to be more IT 100 0.4 productive.

DE 100 0.2

FI 104 0.2 These findings have implications for regional policy. FR 105 0.2 Namely, consideration should be given to proper UK 105 0.2 sequencing when designing the strategy for EU regional SE 106 0.1 policy by taking into account the differences between BE 107 0.1 the stages of development achieved in the catch-up AT 112 0.1 process. In those countries where the convergence gap is NL 113 0.1 highest, in particular when the territory is completely DK 115 0.1 covered under Objective 1 like in most new Member IE 118 0.6 States, more emphasis should be given to national LU 194 0.1 growth as trying to counteract market forces would be inefficient if not even unsuccessful. In the incumbent

Notes: GDP per capita in PPS in relation to the average of the EU-15

in 2001; SF: all Objectives in relation to national GDP by country, Member States, which have already reached an income

2000-2006. level which is closer to the EU average, relatively more

focus can be given to the reduction of regional income

Source: European Commission (2004c). dispersion.

Table 11 shows on the one hand that these eligibility 176 See Ederveen et al. (2002), Weise (2002) or Sapir et al.

criteria have been instrumental in achieving a spatial (2003).

Sequencing and prioritisation have, to some extent, been through efficient pricing and environmental taxes may implemented in the EU-15 cohesion countries. In be more efficient instruments than regional policy to

Ireland, the country with the most impressive growth divert activities towards other areas. performance, the main objective since the 1960s has been the maximisation of national growth. It is only

towards the end of the 1990s that a specific regional 4.2.3 The strategy and the investment mix

policy has emerged and more emphasis has been given EU regional policy is based on a pro-active, allocative to the reduction of regional inequality. In the other approach which targets the determinants of long-term

countries and southern Italy a “mixed” but prioritised sustainable growth with the aim of: strategy has been pursued. Structural expenditures have

initially been focused on national/interregional measures • improving the availability of public goods, i.e.

with specific regional programmes accounting for a mainly basic infrastructure,

small share of total funding. Only from 2000 on there • enhancing human capital, and

has been a shift towards more regional expenditures,

notably in Portugal and southern Italy. Similarly, in the • improving the business environment for investment

2004-2006 period, structural expenditures in the new and offering investment support.

Member States have been mainly focused on national,

interregional measures. However, empirical evidence indicates that not all of these investments are equally effective under all

Linked to the trade-off between equity and efficiency circumstances. Rodriguez-Pose and Fratesi (2002) test within a country is the issue of the intra-regional focus the design of the development strategies co-financed by of regional policies. According to the New Economic the Structural Funds. They regress Structural Funds Geography (NEG), enterprises tend to locate in clusters commitments for each of the four main areas of and areas with high purchasing power and close to other intervention (infrastructure – business/tourism – human enterprises in order to benefit from agglomeration resources – agriculture/rural development) on regional economies. In particular in the new Member States, growth in all NUTS 2 and Objective 1 regions for three business activities tend to locate in the most developed periods from 1989 to 1999, also taking into account a areas (see Section 2.3.1). number of structural variables. They find that

In this context a strategic decision has to be made agricultural/rural support has a strong immediate effect on growth in Objective 1 regions but this impact

between supporting on the one hand the development of vanishes almost immediately and turns negative in later clusters and growth poles and therewith increasing years, suggesting that it fulfils an income support rather overall growth or trying on the other hand to favour the than a sustainable development objective. Returns to dispersion of economic activities. The latter may be infrastructure in transport and environment as well as particularly inefficient at early development stages and business/tourism are relatively disappointing having

may run counter to market forces. For instance, the

relocation of public enterprises to southern Italy from little or no short-term or medium-term impact. However, for infrastructure, this result may be due to a too short

the 1960s to the mid-1970s with national support under period to assess its full impact. Human resources, on the the form of capital grants and wage subsidies did not other hand, have both short-term and medium-term succeed in attracting small and medium-sized private impacts if some characteristics of the labour market are

firms and thus in enlarging the industrial basis in the

South. While clusters have developed in the Centrecontrolled for. On the whole, regions with a balanced distribution of funds have performed well while those

North, no similar agglomeration effects can be found in with unbalanced strategies (e.g. emphasis on business the Mezzogiorno. On the other hand, the promotion of support or agricultural/rural preferences) have not. Such

clusters has been a major feature of the Irish

development strategy since the 1970s and horizontal and results contribute to highlight the importance of adequate regional development strategies.

vertical linkages between industries and research centres

are promoted in Portugal. However, as has been argued Consequently, the effectiveness of EU cohesion policy by some authors, creating artificially comparative in enhancing productivity growth and employment advantages has in most cases proved to have little depends on the national or regional strategy, i.e. the impact. 177 Therefore, regional policy should rather try to investment mix chosen for co-financing. Evidence on build upon existing clusters than try to create new ones. the effectiveness of different types of investment is first

discussed before analysing the strategy chosen for Dispersion of activities is more an issue in relatively

wealthy member states where costs of agglomeration, Structural Funds support in the old and the new Member States.

such as high factor prices, pollution, and congestion tend

to overwhelm agglomeration benefits. However, a more Infrastructure projects are one of the main areas of complete internalisation of negative externalities Structural Funds co-financed investment. A relatively

abundant literature argues that enhanced endowments in transport infrastructure raise the total factor productivity

177 See e. g. Midelfart-Knarvik, Overman (2002) and Midelfart of all inputs (i.e. via reduced transaction costs for

2004). enterprises and also improving workers’ labour

mobility) and thus the growth perspectives of regional or incentives for individuals to invest are found to be national economies. This is supported by evaluations of adequate. More important may be the elimination of Structural Funds programmes and numerous empirical implicit barriers to access to higher education such as

studies. 178 However, the available empirical evidence is liquidity constraints and lower basic skills levels among

still subject to debate as causality and econometric individuals from disadvantaged backgrounds. In issues have not been fully clarified. Three main points addition, guidance on the most productive types of seem to emerge from the existing literature. First, the investments include giving technology-related skills to a provision of transport infrastructure can be regarded as a broad segment of the population, supporting life-long necessary precondition for economic development, but learning and improving conditions for the accumulation

will not per se solve all problems of lagging regions, of research-related human capital. especially if they lack adequate factors of production.

Second, the returns to such investments are probably Although some part of the Structural Funds are used to high when infrastructure is scarce and basic networks co-finance the provision of technical and business have not been completed but may be decreasing if a services (mainly to SMEs), technology diffusion and

179

certain threshold has been reached. This is to be taken more market-based forms of investment financing, the into account in the context of EU enlargement, where co-financing of direct state aid to enterprises remains a

regions with a substantial lack of infrastructure (in most main area of intervention.

184

  Such aid can have

new Member States) co-exist with regions with higher important deadweight, displacement or substitution endowments. Finally, according to the New Economic effects which can question the impact of support and

Geography, infrastructure opening up interregional trade subsequently the effectiveness of EU cohesion policy.

185

may have the paradoxical effect of concentrating Evaluations of state aid are relatively scarce. production in the wealthier regions. However, the Nevertheless, the extent of such effects has been evidence is quite mixed. Concentration has been assessed by some studies, in most cases concluding that

highlighted in some cases 180 while a positive effect on only 10 per cent to 20 per cent of the projects are not

181

disadvantaged regions has been evidenced for others. subject to deadweight. 186 There is thus some evidence,

Besides transport infrastructure, increasing support is though quantitatively limited, that co-financing of state given to environmental infrastructure like waste water aid may not be the most effective channel for EU treatment plants. cohesion policy. Therefore, EU cohesion policy should

Recent theories of economic growth, in particular the be targeted to those investments where deadweight literature on endogenous growth, point to the important seems lower according to existing studies, namely in role of human capital. The result that economies only start-up companies, in small businesses and for grow fast if they have high levels of human capital technological upgrading, research and development and

182

seems robust both theoretically and empirically. human capital training.

However, studies tend to assess human capital at a very Besides these types of investment, support for rural aggregate level without precisely defining the development, mainly for the agricultural sector, is mechanisms through which it influences growth. The quantitatively important. However, the economic specific types of educational and training expenditures importance of primary agriculture for the economy as a to be undertaken by policy-makers are thus less clear. whole is limited. Even in predominantly rural NUTS3

183

A recent study provides policy suggestions, to be areas within the enlarged EU, the largest part of adapted to the specific national and regional conditions, economic activities stems from service (62 per cent) and in favour of a moderate increase in human capital industry (32 per cent) activities. In addition, the trends investment but not in favour of an across-the-board clearly indicate a further decline in the agricultural share increase in subsidies for post-compulsory education as in gross value added and employment. Thus, in order to

178 See e.g. Moreno et al. (2002) or Del Mar Salinas-Jiménez 184 However, it is incorrect to assume, EU Structural Funds

(2004). would mainly distribute state aid, and conclude, based on

179 A non-monotonic relationship between infrastructure and this assumption that Structural Funds are ineffective like in long-run growth is found e.g. by Bougheas et al. (2000). Midelfart-Knarvik and Overman 2002 or Midelfart 2004.

180 185 See Combes and Lafourcade (2001), Faini (1983). A deadweight effect is if the enterprise would have invested

181 even without support; a displacement effect is if it would See e.g. Martin and Rogers (1995).

have invested anyway but in a different region and a

182 This is confirmed by cross-country empirical evidence, see substitution effect is if a different enterprise would have

Mankiw, Romer and Weil (1992) and Barro and Lee (1994). undertaken the investment.

Some studies (e.g. Pritchett (1998) or Caselli et al. (1996)) 186 For a literature review incl. a discussion of the methologies using different (panel data) techniques have questioned the applied see Gerling (2002). For empirical studies, applying link between education and productivity, but recent heterogeneous methodologies and analysing different kinds investigations explain their negative results by poor data of aid schemes see e.g. Honohan (1998), Barry (2003) and and econometric problems. Lenihan (2004) for Ireland, Arup Economics and Planning

183 De la Fuente and Ciccone (2002). See also Chapter 3 in the (2000) for the UK or Gerling (2002) and Ragnitz (2003) for

EU E CONOMY R EVIEW 2003. Germany.

help lagging rural areas, it seems necessary to Table 12: Financial allocation of public spending

concentrate the efforts increasingly outside the eligible under Objective 1 in % of total

agricultural sector.

Old MS Old MS New MS

The standard measures targeted to the agricultural sector 1994/99 2000/02 2004/06 are, furthermore, not exempt from criticism. Early

retirement schemes for instance have little proven National INFR 53.8 45.2 58.4

effects on the restructuring of the sector and run counter HRD 30.0 36.9 19.8 to the Community employment strategy by reducing the

participation rate. The lump sum support to farmers in PROD 16.2 17.9 21.8

rural areas is neither targeted nor supportive to a positive National co INFR 40.9 40.1 43.3 sectoral restructuring. Finally, farm investment support

seems not to be implemented efficiently. 187 HRD 18.2 23.9 24.4

Furthermore, Structural Funds also offer co-financing of PROD 40.9 36.0 32.3

projects where the link to economic growth and EU INFR 31.8 36.9 44.9 employment is at least doubtful. For example, a positive Structural impact on regional development will be difficult to find Funds

HRD 31.8 28.4 25.8

for cultural projects or sport facilities. . PROD 36.5 34.7 29.3

The investment mix in the EU-15 and the new Member Notes: Percentage share of investment area in expenditures on

States infrastructure (INFR), human resources development (HRD) and aid to the productive sector (PROD), excluding other spending of each As the list of eligible expenditures for EU Structural source of finance, national eligible expenditure without co-financing, national co-financing and EU funds. Graphs for the new Member

Funds support is long and not all eligible expenditures States are ex-ante graphs. Calculations based on tables submitted for can be regarded as equally effective, the strategy and the verification of additionality of Objective 1 programmes. main areas of support have to be adapted to the needs of the corresponding Member States and regions. Regional

and national authorities present development plans Table 13: Financial allocation of EU Structural which are then negotiated with the European Funds in EU-15 in % of total Commission and adopted as multi-annual programmes. Non

Objective 1 Obj.1

For the EU-15 Objective 1 regions (see Tables 12 and 1994-99 2000-06 2000-06

• 13) 

  there is mixed evidence on whether financial support Infrastructure 29.8 41.3 14.1

is shifting over time towards investments that are more Transport 15.7 19.8 3.5

conducive to growth and employment or not. Using very

rough categories and only considering Structural Funds, ICT 1.6 3.5 1.7

the share of basic infrastructure has increased in the first Energy 2.3 1.2 0.4

years of the current programming period compared with Environment &

the late 1990s. In contrast, the share of the support for water 7.5 12.8 7.5 Health & social

human resource development has been reduced. inf. 1.7 3.9 0.7

However, as Table 13 displays, this is not only due to Other 1.1 0 0.3 investments in “concrete rather than brain”, but it is also Human due to a stronger focus on environmental and ICT resources 24.5 23.1 53.3

investments. In addition, Structural Funds can be more Education 6.9 n/a n/a easily absorbed by large projects, such as infrastructure Training 17.4 n/a n/a

investments, than by smaller and more complex projects,

such as in the area of human resources. Other 0.1 n/a n/a Productive

Environment 41 33.8 29.1 Industry and services 19.9 11.3 15.8

RDTI 3.5 6 4.5 Agric./rural dev./ fishery 15.2 13.7 5.1

Tourism 2.4 2.7 3.7

Other 4.6 1.8 3.4

Total 100 100 100

Notes: n/a = not available.

Source: European Commission (2003d).

187 See e.g. studies by Striewe, Loy, Koester (1996), Ebers

(1998) and Forstner and Clemens (1998).

In the recent process of Objective 1 programming for the infrastructure) 189 and to a lesser extent for the Czech

new Member States, the focus was on the main Republic. Since the development of human resources is determinants of higher productivity and, in those key to long-term growth, the allocations to the countries where the labour market situation is a key corresponding programmes were increased both where challenge, on a rapid improvement in the use of human employment is a major challenge as in Poland and where resources. Growth and employment have thus been the higher qualifications are called for by the upgrading of two main criteria against which priorities, investments economic activity and by the need to activate and measures were selected. The approach was to participation in the labour market as in Hungary. maximise measures with higher growth and employment

potential, promote concentration by avoiding a The final allocations are thus significantly different from scattering of resources into numerous small projects, the ones of the National Development Plans (see suppress or at least reduce redistributive types of Table 15).

190

  Even if the graphs on the financial

measures and to avoid the creation of distortions in allocation between priorities are not directly economic activity. Against the background of uneven comparable,

191 they indicate that the higher investment

effectiveness of different investment areas as need, compared to the EU-15, in the area of basic highlighted by available evidence, the aim was to select infrastructure has been reflected in the programmes and both adequate priorities and an effective mix of that more emphasis has been given to human resource measures within each priority. This, in turn, has development.

translated into shifts in financial allocations between and Table 14: Comparison of the financial allocation within priorities. in the National Development Plans (NDP) and the

The major adjustments between priorities in the initially Community Support Frameworks (CSFs)

submitted development plans and the finally adopted programmes are illustrated by Table 14 for the largest

four new Member States. Even if agriculture is still of Priorities /OPs PL HU CZ SK

major importance for some rural areas in the new

Member States, it is questionable if this sector will be a Competitiveness NDP 17.8 23.3 15.0 14.5 driving force for growth and employment. In contrast, and enterprises CSF 15.1 21.5 17.9 14.5 major restructuring and labour adjustment are still Human NDP 17.4 23.9 21.0 27.5

needed in some countries which will add to the expected resources

decrease of the share of agriculture in gross value added CSF 17.8 28.2 21.9 27.3

and employment. Consequently, assistance for Agri/food/rural NDP 16.8 18.2 12.0 27.7 agriculture was reduced. The highest reduction was (incl. fishery) CSF 16.8 15.9 12.0 17.6

agreed on in the case of Slovakia where the Structural

Funds allocation was reduced from 27.7 per cent to 17.6 Infrastructure*

NDP 8.6 16.5 13.5 30.4

per cent. As mentioned above, there is no evidence on CSF 14.1 16.4 16.9 40.6

the contribution to national growth and employment of Regional NDP 39.2 17.9 38.5 some regional and local measures such as cultural development CSF 35.9 18.0 31.2

investment or sport facilities. Therefore, it was agreed

with several countries to scale down such programmes. Notes: Figures given in % of total, Cohesion Fund excluded. The

In addition, due to the high deadweight and figures for the Technical Assistance Priority are not included in the table. *Excluding regional and local infrastructure, except for

displacement effects of state aid and because of the Slovakia. Calculations based on National Development Plans

already high level of state aid in most new Member (NDP) and Community Support Frameworks (CSF).

States, 188 it was in most cases agreed to reduce the

support of EU Structural Funds to this area. This resulted, if not counterbalanced by increasing support

for the business environment like in the Czech Republic, 189 into a reduction of the competitiveness/enterprises These shares do not include infrastructure like Transfinancial

 allocations like in Poland and Hungary. In European Networks (TENs) financed by the Cohesion

contrast, more emphasis was put on infrastructure as this Fund. 190

is regarded as a major weakness impeding higher growth Note that Table 14 gives only a partial picture of the in several new Member States. This was particularly the reallocation of funds agreed on between the national

case for Poland where the allocation was increased from authorities in charge and the Commission services as

8.6 per cent to 14.1 per cent (excluding regional already in the officially submitted NDPs major shifts had been included compared to the preliminary draft plans

infrastructure) and for the Slovak Republic from 30.4 submitted informally end 2001/early 2002.

per cent to 40.6 per cent (including regional 191 Programmes with similar objectives are in different

countries not identically designed. For example, a major part of the Polish Integrated Regional Development Programme is devoted to infrastructure, largely explaining the differences compared to other new Member States in the percentage shares of the corresponding two Operational

188

  European Commission (2002c). Programmes.

For competitiveness, not only was EU co-financing of transport infrastructure was substantially reduced (e.g. in direct state aid reduced. Simultaneously, state aid was Poland and the Czech Republic). The numerous requests re-oriented towards SMEs and targets ensuring that for regional/local cultural or sport facilities were priority is given to SMEs in the financial allocation have reduced in terms of financial allocations and made been set for example in Poland, Hungary and the Czech subject to conditions, in particular economic

Republic. All sectoral preferences were suppressed to sustainability and significant regional economic impact. avoid “protecting” declining industries or trying to pick

up winners by targeting manufacturing or specific “high In agriculture, finally, efforts were made to give higher

192

tech” sectors. The remaining measures in this priority importance to rural development aimed at offering are thus more focused on soft aid for knowledge, alternative employment at the expense of state aids for innovation and technology and the business the processing industry and on-farm investment support.

environment. For example in Poland, the financial allocation to rural development has more than tripled at the expense of

Especially for human resources the measures have to be direct aid measures. tailored to the country’s situation. For example in Hungary where both unemployment and the

participation rate are low and where in some sectors and 4.2.4 Effective use of funds

regions shortages of highly skilled workers can be The extent to which EU cohesion policy will be turned observed, the focus was put on those measures likely to into capital formation depends on the magnitude of the increase participation and on education and training. In administrative costs as these divert expenditures from contrast, for example in Poland and Slovakia where productive investments. Costs can result form unemployment is a key challenge, measures for social insufficient management and can be improved by inclusion were granted limited financial allocation to the capacity-building measures increasing public benefit of active labour market policies and in the latter administrative efficiency. Though necessary, such support was shifted towards groups with the highest measures will in turn diminish resources for investment. possibility to (re-)enter the labour market like youth. They can also result from regulatory complexity. The

In transport, a hierarchy of priorities for the period requirements of the Structural Funds regulations imply 2004-2006 were followed with a view to maximising somewhat complex procedures and thus transaction investments that yield higher returns in terms of costs for programming, monitoring, evaluations and enterprises competitiveness while facilitating labour control systems. Simplifications have been introduced, mobility. This has lead, depending on the situation in the but there is a trade-off between simplicity and country, to giving international and interregional accountability. All the more so since the final transport infrastructure clear priority like in the Czech accountability for the use of Structural Funds lies in the

Republic and Hungary and to suppress (Hungary) or hands of the European Commission.

condition (Czech Republic) aid for regional airports. The regulation for the current 2000-2006 programming

For regional programmes the aim was to avoid that they period has tried to set incentives to achieve high quality mimic the CSFs at regional level and widely disperse in the implementation of Structural Funds programmes resources into numerous priorities and measures with by introducing the performance reserve as a new most likely little effect on long-run growth and instrument. The allocation of the reserve of about 4 per employment. Consequently, an even distribution of the cent of total funding 2000-2006, which took place in Structural Funds across the whole territory (like in 2004, has led to rather heterogeneous results. Three Slovakia) as well as one favouring the most backward groups of indicators have been used to determine which regions (like in Poland) had to be avoided. Focus was programmes can be regarded as performing: indicators given on investment in areas and urban centres with related to output, to management and to financial growth potential while providing the necessary absorption. The use of these indicators is regarded as an infrastructure to allow for their inter-connections and incentive to improve the administrative situation. In connections with major transit routes, notably in practice, however, the necessary information was not

Slovakia and Poland. Financing of small-scale regional always available and a variety of methods have been used for the allocation ofthe reserve in different Member

States so that in some cases also less-performing

192 E.g. in the case of Poland it was originally envisaged to programmes benefited.

give preference to projects in “high technology sectors” and in “traditional industrial branches, which have potential for efficient export and may become competitive in the future (e.g. steel industry, …, ship-building industry, heavy machine-building industry, heavy chemistry, industry of copper …”). Ministry of the Economy, Labour and Social Policy: Sectoral Operational Programme Improvement, of Competitiveness of the Economy for 2004-2006, adopted by the Committee for European Integration 14. 02. 2003, p. 48.

Table 15: Compliance with the principle of • Stronger spatial concentration,

additionality for Objective 1 • Better thematic concentration,

Ex- post Ex-ante Mid-term

1994/1999 2000/2006 2000/2002 • More effective use of funds.

(1) (2) (3)

Belgium 118 98 117

Germany 80 93 88 Spatial concentration means concentrating Structural Greece 124 146 139 Funds on those regions and Member States most in

Spain 98 101 104 need. This implies, first, a decision whether to continue France 120 105 99 supporting regions in relatively rich Member States;

Ireland 166 200 189 and, second, if and to what extent to continue the Italy 80 104 98 support in regions whose eligibility is affected

Netherlands 124 231 253 negatively by the statistical effect of enlargement. These

Austria 136 103 103 issues are considered in the Draft Framework Regulation

Portugal 118 116 119

Finland 127 108 110 of Structural Funds for the programming period 2007 –

Sweden 114 249 264 2013 proposing to strengthen the focus on the new UK n/a n/a n/a Convergence Objective by allocating 78.5 per cent of

Notes: Indices for annual averages of national public eligible the resources to this objective, in comparison to 72 per

expenditures. cent for the Objective 1 regions in the current (1) Ex-post 1994/1999 compared with ex-ante 1994/99. programming period. Structural Funds in “statistical

(2) Ex-ante 2000/2006 compared with ex-post 1994/99. effect regions” would be continued, but only on a

(3) Ex-post 2000/2002 compared to ex-post 1994/99. 193 transitional and decreasing basis that cannot be

Calculations based on tables submitted for the verification of prolonged for the years after 2013.

additionality of Objective 1 programmes. Spatial concentration also means not counteracting

market forces in the selection of areas for support. As a

The second condition for effectiveness is that transfers response to the possible equity-efficiency trade-off, i.e. contribute to increase investment and do not lead to that high catch-up growth might temporarily be crowding-out. The principle of additionality enshrined in accompanied by higher inequalities between regions, a the Structural Funds regulations requires for Objective 1 sequencing approach initially emphasising growth of the programmes that Member States agree ex-ante with the national economy as a whole and at a later stage giving European Commission on a target for national public more prominence to addressing regional disparities eligible expenditure that generally should not be lower could be followed in order to make regional policy more than the level achieved during the former programming efficient. In parallel, the catching-up process of poorer period. Ex-post and mid-term verifications for the regions might be accelerated by supporting their growth periods 1994-99 and 2000-06 show that in most Member poles and by building on existing clusters. But one States additionality has at least nearly been met and that should avoid any artificial dispersion of economic

this result can be expected as well for the current period activities or creation of new clusters.

(see Table 15). Thematic concentration, in turn, means choosing an

4.3 Policy challenges effective investment mix. The question what an effective investment mix is can only be answered on a case by

In spite of its limited financial means, EU cohesion case basis after a sound analysis of the situation in the policy can have a substantial impact on catching-up - as corresponding Member State and region. However, has been shown by impact assessments based on some general arguments can be made. First, macroeconomic modelling. However, it can only have infrastructure endowment can be seen as a precondition significant effects if several conditions are fulfilled, and for growth, though not as a growth-enhancing here experience in the recent years shows that room for investment per se. Second, even if it generally takes time improvement exists. Among the various factors to achieve a needed enhancement of human capital, this influencing the effectiveness of Structural Funds in can be regarded as key to long-term growth. Third, aid achieving convergence, particularly against the to the productive sector should be limited to specific background of enlargement, the following aspects are projects enhancing the business environment, and important: support for start-ups and SMEs. Thus, in the draft new

ERDF regulation business support is always directly linked to SMEs. Fourth, support for rural areas should

193 Note that in the mid-term and ex-post verifications the take into account the limited and declining importance

graphs are compared with the (in some cases modified) exof agriculture in the process of catching-up, and should ante graphs for the same period and not with the ex-post be focused on providing alternative employment and

graphs of the previous programming period. Note also that development opportunities. Fifth, projects of doubtful in exceptional cases a reduction of national eligible economic benefit – such as for example cultural projects

expenditures can be accepted if the former expenditures –- should not be financed. Finally, and in the light of the

have been of an exceptional magnitude.

subsidiarity principle, thematic concentration implies While the draft new regulation for Structural Funds aims concentrating EU Structural Funds on larger projects, in to introduce a stronger regional and thematic order to achieve a significant impact - while leaving concentration, the contribution of EU cohesion policy to smaller projects to national, regional and local financing. real convergence will depend predominantly on Member

In order to guarantee the effective use of Structural States’ own national and regional policies. The role of Funds, two areas will have to be addressed. First, regional and national authorities in setting up strategies simplifications for the management of Structural Funds to support and implement Structural Funds programs will help to reduce administrative problems and costs. will be of key importance. More broadly, for the Second, particularly in the new Member States, building Structural Funds to have a favourable impact, it will be up the necessary administrative capacity will be of important to assure a stable macroeconomic setting,

crucial importance. effective structural reforms, and good governance practices.

References

A IGINGER , K. AND S. W. D AVIES (2001), “Industrial specialisation and geographic concentration: the two sides of the same coin?” University of Linz Working Paper 23.

A HN , S., AND P. H EMMINGS (2000), “Policy influences on economic growth in OECD countries: an evaluation of the evidence”, OECD Economics Department Working Paper No 246, Paris.

A LONSO , W. (1969), “Urban and Regional Imbalances in Economic Developments”, in: Economic Development and Cultural Change 17, pp. 1-14.

A LVAREZ -P LATA , P., H. B RÜCKER , B. S ILIVERSTOVS , AND DIW B ERLIN (2003), “Potential Migration from Central and Eastern Europe into the EU-15 – An Update”, Report for the European Commission, DG Employment and Social Affairs, Berlin.

A MITI , M. (1999), “Specialization patterns in Europe”, in: Weltwirtschaftliches Archiv 135, pp. 1-21.

A RUP E CONOMICS AND P LANNING (Contractor: Ecotec Research and Consulting Ltd. in association with GVA Grimley and the Centre for Urban and Regional Development Studies, University of Newcastle-upon-Tyne) (2000), “Planning for Clusters”, London.

A UDRETSCH , D. B. AND M. P. F ELDMAN . (1996), “Knowledge Spillovers and the Geography of Innovation and Production”, in: American Economic Review, 8, pp. 630-640.

B ALDWIN , R. (1989), “The growth effects of 1992“, in: Economic Policy, pp. 248-281.

B ALDWIN , R. (1994), “Towards an integrated Europe“, London.

B ALDWIN , R. E., P. M ARTIN , AND G. I. P. O TTAVIANO , (2001). “Global income divergence, trade and industrialization: the geography of growth take-offs”, in: Journal of Economic Growth No 6, pp. 5-37.

B ALDWIN , R. E. AND P. M ARTIN (2004), “Agglomeration and regional growth”, in: Henderson, V. and J.-F. Thisse, (eds.), Handbook of Regional and Urban Economics, forthcoming.

B ARCA , F. (2003), “Regional Policy Experience in Southern Italy”, in: Bernard Funck and Lodovico Pizzati (eds.): European Integration, Regional Policy, and Growth, Washington D.C., pp. 129–133.

B ARRELL , R. ET AL . (2001), “Openness, integration and transition: prospects and policies for economies in transition”, NIESR Discussion Paper No 177.

B ARRIOS , S. AND E. S TROBL , (2004a), “Industry mobility and geographic concentration in the European Union”, in: Economics Letters 82, pp. 71-75.

B ARRIOS , S. AND E. S TROBL , (2004b), “Revisiting the link between national development and regional inequalities: evidence for Europe”, Université catholique de Louvain, forthcoming.

B ARRO , R. J. AND J.-W. L EE (1994), ”Sources of Economic Growth”, Carnegie-Rochester Conference Series on Public Policy, 40, pp. 1-46.

B ARRO , R. AND X. S ALA - I -M ARTIN (1992), “Convergence”, in: Journal of Political Economy 100(2), pp. 223-51.

B ARRO , R. AND X. S ALA - I -M ARTIN (1995), “Economic Growth”, New York, Mc Graw Hill.

B ARRY , F. (2003), “European Union Regional Aid and Irish Economic Development”, in: Bernard Funck and Lodovico Pizzati (eds.): European Integration, Regional Policy, and Growth; Washington D.C., pp. 135–151.

B ASILE R. AND K OSTORIS P ADOA S CHIOPPA F. (2002), “Unemployment dynamics of the Mezzogiornos of Europe: lessons for the Mezzogiorno of Italy”, ISAE, Istituto di Studi e Analisi Economica and ISAE, Istituto di Studi e Analisi Economica.

B ASSANINI , A., S. S ARPETTA AND P. H EMMINGS (2001), “Economic growth: The role of policies and institutions. Panel data evidence from OECD countries”, OECD Economics Department Working Paper No 283, Paris.

B EUGELSDIJK M. AND E IJFFINGER S. C. W. (2003), “The Effectiveness of Structural Policy in the European Union: An Empirical Analysis for the EU-15 during the Period 1995-2001”; CEPR (Centre for Economic Policy Research) Discussion Paper 3879; London.

B EUTEL J. (2002) “The economic impact of objective 1 interventions for the period 2000 – 2006; Final report to the Directorate General for Regional Policies”, http://europa.eu.int/comm/regional_policy/sources/docgener/studies/pdf/objective1/final_report.pdf .

B LUNDELL R. AND D UNCAN A. (1998), “Kernel Regression in Empirical Microeconomics”, The Journal of Human Resources, Vol. 33 (1), pp. 62-87.

B ODE , E. (2004), “The spatial pattern of localized R&D spillovers: an empirical investigation for Germany”, in: Journal of Economic Geography 4, pp. 43-64.

B OERI , T., B RÜCKER , H. ET . AL . (2000), “Study on the impact of eastern enlargement on employment and labour markets in the EU Member States”, study commissioned by the European Commission, DG Employment and Social Affairs, Berlin/Milano 2000.

B OLDRIN M. AND F. C ANOVA (2001), “Inequality and Convergence in Europe’s Regions: Reconsidering European Regional Policies”, in: Economic Policy, No 32.

B ORNER , S. ET AL . (2004), “Institutional efficiency and its determinants – The role of political factors in economic growth”, OECD Development Centre Studies, Paris.

B OTTAZZI L. AND G. P ERI (2003), ”Innovation and Spillovers in Regions: Evidence from European Patent Data”, in: European Economic Review 47(4), pp. 687-710.

B OUGHEAS S., P. O. D EMETRIADES AND T.P. M AMUNEAS (2000), “Infrastructure, Specialization and Economic Growth”, in: Canadian Journal of Economics, Vol. 33, No 3, pp. 506-522.

B OURGUIGNON F., L OLOS S., Z ONZILOS N., S UWA -E ISENMANN A. (1995), “Evaluating the CSF With an Extended Computable General Equilibrium Model: The Case of Greece (1988-1995)”, Journal of Policy Modeling , Vol. 17, Issue 2, pp. 177-197.

B RADA , A. C. AND A. M. K UTAN (2002), “The end of moderate inflation in three transition economies?”, David Williamson Institute Working Paper No 433.

B RADLEY J., M ODESTO L. AND S OSVILLA -R IVERO S. (1995), “HERMIN: A macroeconometric modelling framework for the EU periphery”, Economic Modelling, 12(3), pp. 221-47.

B RÜLHART , M. (1998), “Trading places: Industrial specialization in the European Union”, in: Journal of Common Market Studies No 36, pp. 319-346.

B RUNO , M. AND W. R. E ASTERLY (1998), “Inflation Crises and Long-Run Growth”, in: Journal of Monetary Economics, Vol. 41, No 1, pp. 3-26.

B URNSIDE , C. AND D. D OLLAR (2000), “Aid, Policies, and Growth”, in: American Economic Review, Vol. 40, pp. 289- 329.

C AMPOS , N. F. AND F. C ORICELLI (2002), “Growth in transition: What we know, what we don’t and what we should”, in: Journal of Economic Literature, Vol. 40, pp. 793-836.

C AMPOS , N. F. AND Y. K INOSHITA (2003). “Why does FDI go where it goes? New evidence from the transition economies”, IMF Working Paper No 03/228.

C ANOVA , F. AND A. M ARCET , (1995). “The poor stay poor: non-convergence across countries and regions”, CEPR discussion paper No 1265.

C ARSTENSEN , K. AND F. T OURBAL (2003), “Foreign direct investment in Central and Eastern European countries: A dynamic panel analysis”, Kiel Working Paper No 1143, forthcoming in: Journal of Comparative Economics.

C ASELLI , F., G. E SQUIVEL AND F. L EFORT (1996): “Reopening the Convergence Debate: a New Look at Crosscountry Growth Empirics”, in: Journal of Economic Growth, Vol. 1 No 3, pp. 363-389.

C ASTELLS , A. AND A. S OLLÉ -O LLÉ (2004), “The regional allocation of infrastructure investment: The role of equity, efficiency and political factors”, forthcoming in: European Economic Review.

C HATTERJI , M. (1992), “Convergence clubs and endogenous growth”, in: Oxford Review of Economic Policy 8(4), pp. 57-69.

C OMBES , P.-P. AND M. L AFOURCADE (2001), “Transportation Cost Decline and Regional Inequalities: Evidence from France”, CEPR Discussion Paper No 2894.

C OMBES , P.-P. AND H. O VERMAN , (2004). “The spatial distribution of economic activities in the European Union”, in: Henderson, V. and Thisse, J.-F. (eds.), Handbook of Urban and Regional Economics, Vol. 4. Elsevier-North Holland, Amsterdam; forthcoming.

D AVIES , S. AND M. H ALLET , (2001). “Policy responses to regional unemployment: Lessons from Germany, Spain and Italy”, DG ECFIN Economic Paper No 161, Brussels.

D AVIES , S. AND M. H ALLET (2002), “Interactions between National and Regional Development”, HWWA Discussion Paper 207, Hamburg Institute of International Economics.

D E LA F UENTE , A., (2002), “Convergence across countries and regions: theory and empirics”, Insituto de Análisis Económico (CSIC), forthcoming.

D E LA F UENTE , A. (2003), “Second-best Redistribution Through Public investment: A Characterization, an Empirical Test and an Application to the Case of Spain”, Regional Science and Urban Economics, forthcoming.

D E LA F UENTE , A. AND X. V IVES (1995), “Infrastructure and Education as instruments of regional policy: evidence from Spain”, in: Economic Policy No 20, pp. 13-51.

D E LA F UENTE , A. AND A. C ICCONE (2002), “Human Capital in a Global and Knowledge-Based Economy”, Report for the European Commission, DG Employment and Social Affairs, Brussels.

D EL M AR S ALINAS -J IMÉNEZ , M (2004), “Public infrastructure and private productivity in the Spanish regions”, in: Journal of Policy Modelling, Vol. 26, issue 1, pp. 47-64

D ENIS , C., K. M C M ORROW AND W. R OEGER (2002), “Production function approach to calculating potential growth and output gaps – estimates for the EU Member States and the US”, DG ECFIN Economic Paper No 176, Brussels.

D EVEREUX , M., R. G RIFFITH , AND H. S IMPSON (2003), “The geographic distribution of productive activity in the UK”, in: Regional Science and Urban Economics, forthcoming.

D JANKOV , S. AND B. M. H OEKMANN (2000), “Foreign investment and productivity growth in Czech Enterprises”, World Bank Review, Vol. 14, pp. 49-64.

D JANKOV , S. AND P. M URRELL (2000), “Enterprise restructuring in transition: A quantitative survey”, in: Journal of Economic Literature, Vol. 40, pp. 739-792.

D OLLAR , D. AND A. K RAAY (2003), “Institutions, Trade, and Growth”, in: Journal of Monetary Economics No 50, pp.133-162.

D URLAUF , S AND D. Q UAH (2002), “The new empirics of economic growth”, NBER Working Paper No 6422.

D URANTON , G. AND D. P UGA , (2004), “Micro-foundations of urban agglomeration economies”, in: Henderson, V. and Thisse, J.F. (eds.), Handbook of Urban and Regional Economics, Vol. 4. Elsevier-North Holland, Amsterdam, forthcoming.

D URO , J.A., (2001), “Regional income inequalities in Europe: an updated measurement and some decomposition results”, Instituto de Anlásis Económico, CSIC, forthcoming.

E ASTERLY , W. AND R. L EVINE (2003), “Tropics, germs and crops: how endowments influence economic development”, in: Journal of Monetary Economics No 50, pp. 3-39.

E BERS , H. (1998) “Erfolgskontrolle investiv geförderter landwirtschaftlicher Unternehmen im Bereich der Landwirtschaftskammer Hannover“, Diplomarbeit, Göttingen.

E DERVEEN , S., J. G ORTER , R. DE M OOIJ AND R. N AHUIS (2002), “Funds and Games: The Economics of European Cohesion Policy”, The Hague.

E DERVEEN , S. AND J. G ORTER (2002): “Does European cohesion policy reduce regional disparities?” CPB Discussion Paper No 15, The Hague.

É GERT , B. ET AL . (2004), “Inflation differentials in Europe: Past experience and future prospects”, Österreichische Nationalbank, Monetary policy and the economy, Q1/04, pp. 47-72.

E STEBAN , J. M. (1999), “L’euro y la desigualtat territorial: implications per a Catalunya”, in: J.M. Esteban and J. Gual (eds.), Catalunya dins l’Euro, Barcelona: Antoni Bosh Editor, pp. 165-210.

E UROPEAN C OMMISSION (1996), “European System of Accounts ESA95”, Eurostat, Luxembourg.

E UROPEAN C OMMISSION (2002a), “Employment in Europe 2002”, Luxembourg.

E UROPEAN C OMMISSION (2002b), “Public Finance Report 2002”, Luxembourg.

E UROPEAN C OMMISSION (2002c): “State Aid Scoreboard, autumn 2002 update, Special edition on the candidate countries”, Luxembourg

E UROPEAN C OMMISSION (2003a), “2003 European innovation scoreboard”, Commission Staff Working Paper SEC(2003) 1255.

E UROPEAN C OMMISSION (2003b): “Public Finances in EMU – 2003”; Luxembourg (= European Economy 3/2003).

E UROPEAN C OMMISSION (2003c): “Second progress report on economic and social cohesion”, Luxembourg.

E UROPEAN C OMMISSION (2003d): “Ex-Post Evaluation of Objective 1 1994-1999; A Final Report to the Directorate General for Regional Policy, European Commission”, Brussels.

E UROPEAN C OMMISSION (2004a), “Employment in Europe 2004”, Luxembourg.

E UROPEAN C OMMISSION (2004b), “EMU after five years”, European Economy Special Report 1/2004, Brussels.

E UROPEAN C OMMISSION (2004c): “A new partnership for cohesion: convergence competitiveness cooperation, Third report on economic and social cohesion”, Luxembourg.

E UROSTAT (2002): “Statistics in Focus”, Theme 3, 56/2002, Luxembourg.

E VENETT , S. J. AND W. K ELLER (2002), “On Theories Explaining the Success of the Gravity Equation”, in: Journal of Political Economy 110 (2), pp. 281-316.

F AINI , R. (1983): “Cumulative Process of Deindustrialisation in an Open Region: The Case of Southern Italy”, 1953 – 1971; in: Journal of Development Economics, Vol. 12, pp. 277–301.

F AYOLLE J. AND L ECUYER A. (2000), “Regional growth, national membership and European structural funds: an empirical appraisal”; OFCE (Observatoire Français des Conjonctures Èconomiques) Working Paper No 00-02, Paris.

F ELDMAN , M. P. AND D. B. A UDRETSCH . (1998), “Innovation in Cities: Science-based Diversity, Specialization, and Localized Competition”, in: European Economic Review, 43: pp. 409-429.

F ELDSTEIN , M. (1996), “The costs and benefits of going from low inflation to price stability”, NBER Working Paper No 5469.

F ITZ _G ERALD J. (1999), “Impact of NDP on the Economy", in: Stationery Office: Ex Ante Evaluation of the National Development Plan, 2000-2006; Dublin.

F ORSTNER , B. AND D. C LEMENS (1998), “Einzelbetriebliche Investitionsförderung in Schleswig-Holstein effizient und nachhaltig?“, in: Bauernblatt für Schleswig-Holstein und Hamburg, issue 51/52, pp. 74-78.

G ALLUP , J. L., J. S ACHS AND A. D. M ELLINGER (1998), “Geography and Economic Growth”, Paper prepared for the Annual Bank Conference on Development Economics, Washington, D.C., April 20-21, 1998.

G ARIBALDI , P. ET AL . (2001), “What moves capital to transition economies?”, IMF Staff Papers, Vol. 48, pp. 109- 145.

G ERLING , K. M. (2002), “Subsidization and Structural Change in Eastern Germany“, Berlin.

G IANNETTI , M. (2002), “The effects of integration of regional disparities: Convergence, divergence or both?”, in: European Economic Review No 46, pp. 539-567.

H AERDLE , W. (1992), “Applied nonparametric regression”, Econometric Society Monographs Series 19, Cambridge University Press.

H ALLET , M. (2000), “Regional Specialisation and Concentration in the EU”, DG ECFIN Economic Paper No 141, Brussels.

H ECHT , Y., A. R AZIN AND N. G. S HINAR (2002), “Interactions between capital inflows and domestic investment: international panel data”, Central Bank of Israel, Mini Conference, May 2002.

H IRSCHHAUSEN , C. VON (2002), “Infrastrukturentwicklung in den mittel- und osteuropäischen EU-Beitrittsländern: Auf dem Weg nach Europa“, DIW Wochenbericht 37/02 2002.

H OLLAND , D. AND N. P AIN (1998), “The diffusion of innovation in Central and Eastern Europe, A study of determinants and impact of foreign direct investment”, NIESR Discussion Paper No 137.

H ONOHAN , P. (1998), “Key Issues of Cost-Benefit Methodology for Irish Industrial Policy”, General Research Series 172, The Economic and Social Research Institute, Dublin.

I NSTITUT FÜR W IRTSCHAFTSFORSCHUNG H ALLE (2003), “Wie hoch sind die Transferleistungen für die neuen Länder?“, IWH-Pressemitteilung 21/2003.

I NTERNATIONAL M ONETARY F UND (IMF) (2003a), “World Economic Outlook 2003”, Washington D. C.

I NTERNATIONAL M ONETARY F UND (IMF) (2003b), “Italy: Selected Issues”, Country Report No 03/352.

I SLAM , N. (1995), “Growth empirics: a panel data approach”, in: Quarterly Journal of Economics No 4, pp. 1127- 1170.

J ENKINS , S. P. (1991). “The measurement of income inequality”, in: Osberg (ed.), Economic Inequality and Poverty: International Perspectives, Armonk NY, M.E. Sharpe.

J ENKINS , S. P. (1995), “Accounting for inequality trends: decomposition analyses for the UK, 1971-86”, in: Economica 62, pp. 29-63.

K AUFMANN , D., A. K RAAY AND M. M ASTRUZZI (2003), “Governance Matters III: Governance Indicators for 1996- 2002“, World Bank Policy Research Department Working Paper No 3106.

K ELLER , W. (2002). “Geographic Localization of International Technology Diffusion”, in: American Economic Review 92(1), pp. 120-142.

K RUGMAN , P. (1991), “Increasing returns and economic geography”, in: Journal of Political Economy 99 (3), pp. 483-499.

K RUGMAN , P. AND A. J. V ENABLES (1996), “Integration, specialization, and adjustment”, in: European Economic Review 40 (3-5), pp. 959-967.

K UZNETS , S., (1955), “Economic Growth and Income Inequality”, in: American Economic Review 65, pp. 1-28.

L ANDESMANN , M. A. (2003), “Structural features of economic integration in an enlarged Europe: patterns of catching-up and industrial specialisation”, European Commission, DG for Economic and Financial Affairs, Economic Paper No 142, Brussels.

L EE , H. Y., L. A. R ICCI AND R. R IGOBON (2004), “Once Again, is Openness Good for Growth?” IMF Working Paper WP/04/134, Washington D.C.

L ENIHAN , H. (2004), “Evaluating Irish industrial policy in terms of deadweight and displacement: A quantitative methodological approach”, in: Applied Economics, Vol. 36 No 3, pp. 229-252.

L OVINO , I. (2003), “Acceding countries: still attractive to foreign direct investors”, Eurostat, Statistics in Focus, Theme 2, No 51.

L UCAS , R. E. Jr. (1988), “On the Mechanics of Economic Development”, in: Journal of Monetary Economics No 22: pp. 3-42.

L UCAS , R. E. Jr. (2000). “Some Macroeconomics for the 21 st Century”, in: Journal of Economic Perspectives 14(1),

pp. 159-168.

M AGRINI , S., (2004), “Regional (Di)convergence”, in: Henderson, V. and Thisse, J.-F. (eds.), Handbook of Urban and Regional Economics, Vol. 4. Elsevier-North Holland, Amsterdam, forthcoming.

M ANKIW , N. G., D. R OMER , AND D. N. W EIL (1992), “A Contribution to the Empirics of Economic Growth“, in: Quarterly Journal of Economics, Vol. 107, pp. 407-437.

M ARTIN , P. AND C. A. R OGERS (1995), “Industrial location and public infrastructure”, in: Journal of International Economics, Vol. 39, pp. 335-351.

M ARTIN , P. AND G. O TTAVIANO (1999), “Growing Locations: Industry Location in a Model of Endogenous Growth”, in: European Economic Review No 43, pp. 281-302.

M AUREL , F. AND B. S ÉDILLOT (1999), “A measure of the geographic concentration in French manufacturing industries”, in: Regional Science and Urban Economics 29, pp. 575-604

M IDELFART K. H. (2004), “Regional Policy Design: An analysis of relocation, efficiency and equity”, CEPR Discussion Paper No 4321, London.

M IDELFART -K NARVIK , K. H. AND H. G. O VERMAN (2002), “Delocation and European Integration: Is structural spending justified?”, in: Economic Policy, Vol. 35, pp. 322–359.

M IDELFART -K NARVIK , K.H., H. G. O VERMAN , S. R. R EDDING AND A. J. V ENABLES , (2002), ”The location of European industry”, European Economy 2, pp. 216-273

M ODY , A., A. R AZIN AND E. S HADKA (2003), “The role of information in driving FDI flows: Host country transparency and source-country specialisation”, NBER Working Paper No 9662.

M OLLE , W. (1997), “The regional economic structure of the European Union: an analysis of long term developments” in: K. Peschel, ed., Regional Growth and Regional Policy within the Framework of European Integration (Physica-Verlag, Heidelberg) pp. 66-86.

M ORENO , R., E. L ÓPEZ -B AZO AND M. A RTÍS (2002), “Public infrastructure and the performance of manufacturing industries: short- and long-run effects”, in: Regional Science and Urban Economics, Vol. 32, issue 1, pp. 97-121.

N ADARAYA , E. (1964), “On Estimating Regression”, in: Theory of Probability and its Applications 9, PP . 141-42.

N ORTH , D. C. (1990): “Institutions, Institutional Change and Economic Performance”, New York.

O RLOWSKI , W. M. (2004), “Accelerate change”, IMF Finance and Development, June 2004, pp. 34-35.

P ETRAKOS , G. AND J. C. B RADA (1989): ”Metropolitan Concentration in Developing Countries”, in: KYKLOS: The International Review of Social Sciences, Vol.42, Fasc.4, pp. 556-578.

P ETRAKOS , G. AND Y. S ARATSIS (2000), “Regional Inequalities in Greece”, in: Papers in Regional Science 79(1), pp. 57-74.

P ETRAKOS , G., A. R ODRÍGUEZ -P OSE AND A. R OVOLIS (2003), “Growth, integration and regional inequality in Europe”, in: Research papers in Environmental and Spatial Analysis No 81, London School of Economics.

P RITCHETT L. (1995), “Patterns of Economic Growth: Hills, Plateaus, Mountains, and Plains”, World Bank Working Paper No 1947; Washington D.C.

P UGA , D. (1999), “The rise and fall of regional inequalities”, in: European Economic Review 43(2), February 1999, pp. 303-334.

P UGA , D. (2002), “European regional policies in light of recent location theories”, in: Journal of Economic Geography No 2, pp. 373-406.

Q UAH , D. (1997), “Empirics for Growth and Distribution: Stratification, polarization, and Convergence Clubs”. In: Journal of Economic Growth 2(1), pp. 27-59.

Q UAH , D. (1996), “Regional convergence clusters across Europe”, in: European Economic Review, 40 (3-5), pp. 951- 958.

Q UAH , D. (1999), “Models of explicit distribution dynamics”, in: European Commission (ed), The Socio-Economic Impact of Projects Financed by the Cohesion Fund: a modelling approach, 79-219. Luxembourg.

R AGNITZ , J. (2003), “Wirkungen der Investitionsförderung in Ostdeutschland“, Institut für Wirtschaftsforschung Halle: Diskussionspapiere No 186; Halle.

R AHMAN M. A. (1963), “Regional allocation of investment”, in: Quarterly Journal of Economics 77, pp. 26-39.

R AISER M. ET AL . (2001), “Social capital in transition, a first look at the evidence”, EBRD Working Paper No 61.

R AISER M., M. S CHAFFER AND J. S CHUCHART (2003), “Benchmarking structural change in transition”, EBRD Working Paper No 79.

R AZIN , A. (2002), “FDI contribution to capital flows and investment in capacity”, NBER Working Paper No 9204.

R IVERA -B ATIZ , L. A. AND P. R OMER (1991), “Economic Integration and Endogenous Growth”, in: The Quarterly Journal of Economics 106, pp. 531-555.

R ODRIGUEZ , F . AND D. R ODRIK (2000), “Trade policy and economic growth: A skeptic’s guide to cross-national evidence” in: B. Bernanke and K. S. Rogoff (eds.): Macroeconomics Annual 2000, MIT Press, Cambridge, MA, 2001.

R ODRIK , D., A. S UBRAMANIAN , F. T REBBI (2002), “Institutions Rule: The Primacy of Institutions over Geography and Integration in Economic Development“, IMF Working Paper No 189.

R OEGER , W. AND J. IN ’ T V ELD (1997), “Quest II – A multi country business cycle and growth model”, DG ECFIN Economic Paper No 123.

R OMER , D. (2001), “Advanced macroeconomics”, 2 nd edition, McGraw-Hill, Irwin.

R OSE , A. K. (2000), “One money, one market: The effect of currency unions on trade”, in: Economic Policy April 2000, pp. 9-45.

S ACHS , J. D. (2003), “Institutions don't rule: direct effects of geography on per capita income”; NBER Working Paper 9490.

S ALA -I-M ARTIN , X. (1997), “I just ran 4 million regressions”, NBER Working Paper No 6252.

S ALA -I-M ARTIN , X. (2003), “Executive summary to the world competitiveness report 2003/04”.

S APIR , A. (1996), “The effects of Europe’s internal market programme on production and trade: a first assessment“, in: Weltwirtschaftliches Archiv 132 (3), pp. 457-475.

S APIR , A. ET AL . (2003), “An Agenda for a growing Europe: Making the EU Economic System Deliver”, Report of an Independent High-Level Study Group established on the imitative of the President of the European Commission, Brussels.

S ILBER , J. (1989), “Factor Components, Population Subgroups and the Computation of the Gini Index of Inequality”, in: Review of Economics and Statistics, No 71: pp. 107-115.

S TRIEWE , L., J.-P. L OY AND U. K OESTER (1996), “Analyse und Beurteilung der einzelbetrieblichen Investitionsförderung in Schleswig-Holstein“, in: Agrarwirtschaft, Vol.45, No 12, pp. 423-434.

S OLANES J.G. AND R AMÓN M.D. (2001), “The Impact of European Structural Funds on Economic Convergence in European Countries and Regions; in: W. Meeusen and J. Villaverde Castro (eds): Convergence Issues In The European Union, Edward Elgar Publishing Ltd, 2002, pp. 61–82.

S OLOW , R., (2000), “Toward a Macroeconomics of the Medium run”, in: Journal of Economic Perspectives 14(1), pp. 151-158.

S VEJNAR , J. (2002), “Transition economies: performance and challenges”, in: Journal of Economic Perspectives, Vol. 16, No 1, pp. 3-28.

T ONDL , G. (1997), “The ups and downs of regional income convergence in Europe”, Robert Schuman Centre working paper No 97/53, European University Institute, Florence.

T ONDL , G., (2001), “Convergence after divergence? Regional growth in Europe“, Springer-Verlag: Wien.

T ONDL , G. AND G. V UKSIC (2003), “What makes regions in Eastern Europe catching-up? The role of foreign investment, human resources and geography”, ZEI Working Paper B12.

T RAISTARU , I., P. N IJKAMP AND S. L ONGHI (2002), “Regional specialization and concentration of industrial activity in Accession Countries”, ZEI working Paper No 16, Bonn.

V ON S CHÜTZ , U. AND M. S TIERLE (2003), ”Convergence in an enlarged EU? An empirical analysis of regional specialisation and sectoral concentration”, in: C. Hausen, M. S. Resinek, N. Schürmann and M. Stierle (eds.) Determinants of Growth and Business Cycles: Theory, Empirical Evidence and Policy Implications INFER Annual Conference 2003 Reihe: INFER Research Edition, Vol. 9

W ACHTEL , P. AND I. K URHONEN (2004), “Observations on dis-inflation in transition economies“, Bank of Finland Discussion Paper No 5/2004.

W ÄLDE , K. AND C. W OOD (2004), “The empirics of trade and growth: where are the policy recommendations?” DG ECFIN Economic Paper No 204, Brussels.

W ATSON , G. (1964), “Smooth Regression Analysis”, in: Sankhya 26, PP . 359-72.

W EISE , Christian (2002): “How to finance Eastern Enlargement of the EU: The need to reform EU policies and the consequences for the net contributor balance“, DIW Discussion Paper No 287, Berlin.

WIFO (1999), “Specialisation and (Geographic) Concentration for European Manufacturing – Background Paper for ’The Competitiveness of European Industry 1999 report’ ”, European Commission, DG Enterprise working paper 1, Brussels.

W ILLIAMSON , J., (1965), “Regional inequality and the process of national development”, in: Economic Development and Cultural Change 14, pp. 3-45.

W ORLD B ANK (2002), “World Development Report 2002”, Washington D.C.

W ORLD B ANK (2004), “Doing Business in 2005: Removing obstacles to growth”, Washington D.C.

W URZEL , E. (2001), “The Economic Integration of Germany’s New Länder”, OECD Economics Department Working Papers, No 307, Paris.

Y UILL , D. AND F. W ISHLADE (2001), “Regional Policy Developments in the Member States: A Comparative Overview of Change“, European Policies Research Centre, University of Strathclyde, Regional and Industrial Policy Research Paper No 45, Glasgow.

ANNEX I: M ETHODOLOGICAL CONCEPTS OF CONVERGENCE

Three concepts of convergence are generally found in the literature. The first one concerns the evolution of the distribution of income per capita, i.e., the so-called σ-convergence. The other two ask whether poorer countries tend to catch-up with the richer ones or whether the relative position of each country within the income distribution, considering its fundamentals, tends to stabilize over time. In the first case we talk about absolute β-convergence while in the second we talk about conditional β-convergence. While the concept of σ-convergence refers to a single statistic, the other two deal with the causality between two variables: the growth rate of income per capita and its initial level. Considering the following expression:

Δy it = αx i - βy it (i)

where y it is the level of income per capita in country (or region) i and Δy it is the rate of growth of this variable between t=T and t=0. The variable x i , is assumed, for simplicity, to be constant over time and represents the

fundamentals of each country i (i.e., change in population, investment rate, technological capability, etc.) that are likely to determine the steady-state level of per capita income of each economy. With Δy it =0 in the steady state we have:

y* =(αx i ) / b (ii)

Then if 0<β<1, there is conditional β-convergence. If, in addition, x i is the same across all i, that is, all countries

(regions) converge to the same income per capita, then there is absolute β-convergence. This is equivalent to estimate econometrically (i) with common intercept and no other explanatory variable besides the initial level of per capita income. Starting from the neo-classical model of convergence described, for instance, in Barro and Sala-i-Martin (1995), one can derive the corresponding convergence time, i.e., the time it takes for a given country to converge to the steady-state. Using the logarithm expression of the convergence equation:

ln(y t ) = (1-e -βt ) ln(y * ) + e -βt ln (y 0 ) (iii)

where y* denotes the steady state level of income and y o the initial level of income, the convergence time t can be derived by plugging the estimate of β into the following expression:

e -βt = H

where H denotes the position of y t compared to y o and y*. For instance, in order to know how long it takes for a country’s GDP per head y t to be half-way between y o and y t , the corresponding convergence time will be:

T = ln(1/2)/β

Note also that, while β-convergence is a necessary condition for σ-convergence, it is not sufficient for convergence to actually take place since a positive value of β is compatible with a transitory rise of income dispersion (due, for instance, to transitory shocks to the economy). It is only when poor economies grow faster than richer ones that the reduction of income disparities will in fact happen. It follows that a negative value of β does not guarantee that the dispersion of incomes is smaller at the end of a period than at the beginning or even that regions converge to a common steady state. In particular, Chatterji (1992) showed that for both β-convergence and σ-convergence to take place, the value of β must be such that -2 < β <0.

Note also that, related to this latter point, the concepts of absolute and conditional β-convergence have not the same implications in terms of inequality since the first implies that all economies will, in the long run, converge exactly to the same level of income while in the second case, each economy converges toward its own steady-state. Hence, in the latter case, inequalities could persist even if conditional convergence is taking place.

Several indices can be used to describe income disparities across countries and regions. Three main indices have generally been used in the literature: the Gini index, the Theil index and the coefficient of variation. These inequality indices differ in their sensitivities to income differences in different parts of the distribution, in particular, the Gini coefficient being most sensitive to income differences at the mode of the distribution while the coefficient of variation is more sensitive to high incomes. An advantage of the Theil index and of the coefficient of variation, however, is that they are easily decomposable into group-contribution. In particular, EU countries can be considered

as specific groups with regions belonging to the same country sharing common features in terms of GDP per head. 194

One can thus use the coefficient of variation and the Theil index to derive the relative contribution of within-country

variation and between-country variation in explaining the total variation in GDP per head across EU regions.

194 The Gini index can also, in principle be decomposed into within and between groups components. However, while such

decomposition is not straightforward, it also involves an interaction terms which may capture a large part of income variability across regions, see Silber (1989).

ANNEX II: S EMI - PARAMETRIC TECHNIQUES

Non-parametric techniques are especially suited when considering the possibility for non-linear relationship between a set of variables, see for an introduction. Robinson (1988) shows in addition that these techniques can also allow for the effect of other conditioning variables by using the Kernel regression estimator. This second class of estimator is

195

often termed semi-parametric estimator. Accordingly, the following equation is estimated:

Y it = α + g(X it ) + βZ it + u it (1)

where Z is a set of explanatory variables that are assumed to have a linear effect on Y. The variable Y represents the level of regional inequalities measured as before by the Gini index. The function g() is smooth and continuous while X is the level of GDP per head measured in PPS and u is a random error term. Time and country indices are represented by i and t respectively. In addition, both the dependent and explanatory variables are measured with respect to the EU average. Note also that the set of control variables Z contains time and country dummies in order to control for time and country-specific characteristics that can influence the relationship between national GDP per

head and the level of regional inequalities. 196

A commonly used non-parametric estimator of an unknown function like g(X) without allowing for the effect of

other conditioning variables is the well-known Nadaraya-Watson estimator: 197

n

∑ K h ( X − X i ) yY i

m ˆ ( ) n − 1 i = 1 h X =

− n

 (2)

n 1 ∑ K h ( X − X i )

i = 1 such that i=1…n are the n number of observations, K h () is the shape function, commonly referred to as the Kernel, that is a continuous, bounded and real function that integrates to one and acts as a weighting function of observations around X and depends on the choice of bandwidth h. This technique corresponds to estimating the regression function at a particular point by locally fitting constants to the data via weighted least squares, where those observations closer to the chosen point have more influence on the regression estimate than those further away, as determined by the choice of h and K. This allows avoiding any parametric assumptions regarding the conditional mean function m(X), and thus about its functional form or error structure. Furthermore, Robinson (1988) showed that in controlling for other conditioning variables the (semi-parametric) Kernel regression estimator for g(X) simply

198

becomes:

g ˆ ( ) X = m ˆ ~ ( ) X − δ ˆ ( ) y m ˆ Z X (3) where m ˆ ~ ( ) y X and m ˆ ( )  Z X are the (non-parametric) Kernel regression estimates of E( y  X) and E(Z X), and δ ˆ

is the OLS estimator of:

Y − m ˆ ~ ( ) δ ( ( ) ) ε y X = Z − m ˆ Z X + (4)

where ε is a random error term. Intuitively, ĝ(X) is the estimate of g(X) after the independent effect(s) of Z on Y has

been removed.

The semiparametric estimator presents a number of limitations. First, given that the estimate of gˆ ( ) X is at least in

part based on non-parametric estimation techniques, one cannot subject it to the standard statistical type tests (e.g., ttest). A possibility, adopted here, is to calculate upper and lower pointwise confidence bands as shown by Haerdle (1990). Another limitation comes from the fact that the shape function K h is a weighting function of observation around X and depends on the choice of bandwidth which, again, limits the possibility of hypothesis testing. Finally, the estimator tends to be biased at sudden peaks of the estimation of g(X) and at the left and right boundaries of the data, simply because observations at the neighbourhood of these points are necessarily less informative. For this reason, estimates at the extreme points of the distribution are less reliable.

195

  See Blundell and Duncan (1998) for a useful introduction.

196

  This is especially important given that, for instance, the number of regions could have an influence on the value of the Gini index; see Barrios and Strobl (2004b). Time dummies can also allow controlling for annual specific shocks due to business cycle fluctuations.

197

  See Nadaraya (1964) and Watson (1964).

198

  The fact that δ is in part estimated using OLS makes this a semi- rather than non-parametric estimator.

• 3. 

  LABOUR MARKETS IN THE EU: AN

ECONOMIC ANALYSIS OF RECENT

PERFORMANCE AND PROSPECTS

Summary

Halfway through the first decade of the Lisbon strategy, the targets for employment rates in the EU now look more challenging than they did. When one looks at the key demographic groups from which most of the increase in employment must come, it is difficult to see how the overall target of a 70 per cent employment rate can be achieved by 2010, even in the EU-15, let alone the EU-25. The 50 per cent target for older workers also appears out of reach, though substantial progress has been registered. There is still a chance that the 60 per cent target for female workers will be reached. The macroeconomic slowdown has not helped, but more importantly, a great deal remains to be done in the area of structural reforms.

Nevertheless, there is evidence that much of the improvement in labour market performance over the 1990s was structural, and that significant progress has continued in some areas, such as tax and benefit reforms and early retirement. Also on the positive side, there is no great mystery about the main determinants of labour market performance, or about the kinds of measures Member States need to take in order to permanently raise employment rates. The economic evidence – on the determinants of both overall labour market performance and employment in specific demographic groups – suggests that the right strategy has been set out in the EU’s Broad Economic Policy Guidelines for 2003-05, in the Employment Guidelines based on the reformed European Employment Strategy and in the recent report of the European Employment Taskforce chaired by Wim Kok.

The strategy includes, inter alia, steps to ensure that wages better reflect productivity and local labour market conditions, tax and benefit reforms in conjunction with well-targeted active labour market policies that favour labour market participation, labour market regulations that are conducive to job creation and policies to improve education and training, especially for the low-skilled and older workers. Detailed reform strategies have to be country-specific, looking at the ensemble of labour market and social protection institutions. Appropriate measures in one country might differ from what is required in another. But, in most EU Member States, there is ample scope for improvements in the design of institutions so as to improve incentives to take up employment while tackling deadweight costs and distortions that provide very little in the way of genuine social insurance. A comparison of country-specific priorities as identified in the EU Employment Recommendations and the Broad Economic Policy Guidelines with progress made in the last few years points to areas for urgent action at the Member State level.

Although a rising employment rate may temporarily depress productivity growth, simply because the number of workers per unit of capital is increasing, and because those who move from unemployment or inactivity into employment are likely, on average, to have a relatively low level of productivity to start with, there are three reasons why this does not give cause for concern. First, the temporary negative effect on productivity growth is estimated to be rather small. Secondly, even if growth in productivity – GDP per employed person – is negatively affected, a higher employment rate unambiguously raises growth in GDP per capita. Newly employed people clearly contribute more to GDP than they used to, even if their productivity is below average. Thirdly, both economic theory and evidence suggest that a higher employment rate has no significant negative implications for longer-term productivity growth, which is what really matters for the competitiveness and dynamism of the EU economy. These points – important ones for the Lisbon strategy – are supported by two separate pieces of analysis: an econometric analysis of the dynamic response of productivity to structural employment shocks, and a simulation based on the Commission’s macroeconomic model. This suggests that there is no genuine trade-off between policies to raise the employment rate and policies to foster productivity growth.

TABLE OF CONTENTS

• 1. 

  I NTRODUCTION ............................................................................................................................................116

• 2. 

  L ISBON AT M ID - TERM : A N O VERVIEW ........................................................................................................117

2.1 Labour market performance since 2000 .............................................................................................................. 117

2.2 Employment and labour productivity: reconsidering the potential trade-off ...................................................... 121

• 3. 

  T HE DETERMINANTS OF LABOUR MARKET PERFORMANCE ...........................................................................128

3.1 Determinants of overall performance ................................................................................................................. 128

3.2 Employment and participation in specific groups ............................................................................................... 136

• 4. 

  L ABOUR MARKET REFORM IN THE EU: PRIORITIES AND PROGRESS ..............................................................144

4.1 What is required to meet the Lisbon targets? ...................................................................................................... 144

4.2 Priorities for and progress with labour market reforms in the EU-15 ................................................................. 146

4.3 Labour markets in the enlarged EU .................................................................................................................... 152

• 5. 

  G ENERAL CONCLUSIONS ..............................................................................................................................153

• 6. 

  R EFERENCES ................................................................................................................................................155

LABOUR MARKETS IN THE EU: AN

ECONOMIC ANALYSIS OF RECENT

PERFORMANCE AND PROSPECTS

• 1. 

  Introduction as a signal that the EU was taking economic reform seriously. 199

The Lisbon strategy involves efforts on several fronts Two clear advantages of the Lisbon strategy, and both to improve labour market performance and to raise especially the employment rate targets, are often productivity growth. This twin aspiration is neatly overlooked in these discussions. First, the commitment summed up in the phrase ‘more and better jobs’, which to raising employment rates, i.e. raising labour force implies higher employment rates but also more participation as well as reducing unemployment, productive, higher-quality employment. represents a clear rejection of an idea that has been one The strategy set explicit targets for ‘more jobs’: an of the great weaknesses of some Member States’ employment rate of as close as possible to 70 per cent employment policies in recent decades, namely that high and a female employment rate of over 60 per cent by unemployment can be cured by discouraging labour 2010. The Stockholm summit a year later added a supply. If this seems obvious today, it is not so long ago further target of an employment rate of 50 per cent for in some countries that married women were discouraged older working-age people. Given the rate of employment from working, while older workers were actively growth required to meet these targets, the Lisbon encouraged to quit the labour market through early conclusions also established an implicit target for retirement schemes, partly in response to high productivity growth with the statement that – if the unemployment. Even more recently, governments in recommended measures were implemented against a some EU Member States were entertaining a similar sound macroeconomic background – it should be notion – that employment in persons might be boosted possible to achieve 3 per cent GDP growth. by means of regulatory restrictions on hours worked.

These targets have met with criticism in some quarters Secondly, the Lisbon strategy embodies the idea that on several counts. Some regarded them as overstructural improvements in the functioning of markets ambitious, particularly since the European Council – as are required for a sustained increase in employment opposed to individual Member States – lacks full control rates and higher productivity growth. Clearly, at any of the necessary instruments to meet its objectives. given moment, output and (un-)employment are There were doubts about whether a credible strategy had determined by real demand in the economy. However, been set out, or even whether EU leaders realised the over the longer term, real demand will generally tend extent of reforms that would be required. Others pointed towards a level consistent with stable inflation, this level to the risk of policy distortions – there are many ways to being determined by overall supply conditions in the raise employment rates, for example, but not all of them economy. By focusing on the functioning of labour, are fully consistent with raising economic welfare. On product and capital markets, as well as investments in the other hand, the Lisbon targets appeared to score an R&D and human capital, the Lisbon strategy seeks to initial public relations success, being widely interpreted raise employment and growth potential in a sustainable

manner.

199

  Even then, though, it was noted that this might damage the credibility of similar exercises were the targets to be missed by a wide margin.

In addition, while one may ask whether the employment 2. Lisbon at Mid-term: An Overview

rate is the ideal variable to target, there is no doubt that low employment rates in several EU Member States are

a symptom of poor labour market performance, and that 2.1 Labour market performance since 2000

improving labour market performance would lead both

to higher employment rates and to greater economic In 2000, the Lisbon employment targets seemed within welfare. The benefits of higher employment rates for the easy reach to many. The EU-15 employment rate for the sustainability of public finances, at least in the short-topopulation aged 15-64 was 63.4 per cent, which meant

medium term, were also noted. that in order to reach 70 per cent by 2010, employment had to grow at an annual rate of 1 per cent. 200 This

The Lisbon strategy also addresses two much-discussed would be well above the long-term average of 0.4 per ‘trade-offs’ – one between employment growth and cent since 1960, but half the rate of employment growth productivity growth, the other between employment and in 2000.

social cohesion. On the former, the position is clear:

provided the necessary reforms are undertaken, it is As it turned out, employment growth peaked at 2.0 per possible to have both a substantial increase in the cent in 2000, fell to almost zero in 2003, and is expected employment rate and higher productivity growth. to remain relatively weak in 2004 and 2005: 0.3 per cent Whether this is actually the case is the special subject of and 0.8 per cent respectively, according to the

Section 2.2 of this chapter. Commission’s Spring 2004 economic forecasts. On this basis, the employment rate would rise to 65 per cent in

On social cohesion, Lisbon also takes quite a clear line, 2005, and the annual rate of employment growth calling for modernisation of social protection systems in required to hit the 70 per cent target by the end of 2010 order to ensure that work pays, and stating that “the best would then be 1.5 per cent. In other words, an safeguard against social exclusion is a job”. At the same immediate return to the economic performance of the time, the strategy underlines the need to improve late 1990s would be needed in order to hit the overall working conditions and skill levels – i.e. the quality of Lisbon target. Even then, the higher the employment jobs. This is not to say that uncomfortable choices may rate rises, the more difficult it becomes to sustain rapid not sometimes have to be made between social employment growth.

protection and economic efficiency. However, in many

cases, there is substantial scope for improving the design The disappointing performance over the past few years of labour market institutions in such a way as to improve can be partly explained by the macroeconomic employment performance without weakening social slowdown. Up to 2001, there were signs that structural

protection (see Section 3.1.3) reforms of product and labour markets, together with wage moderation, were beginning to pay off. The

This chapter takes stock of the implementation of the Commission services’ assessment was that these factors Lisbon strategy as far as labour markets are concerned. might be behind as many as 5-6 million additional jobs

Section 2 reviews labour market performance since 2000 since 1995. 201 However, the remaining 6-7 million jobs

and considers the extent to which disappointing progress created during the same period could then be put down can be put down to the less than favourable to cyclical or macroeconomic factors, and the removal macroeconomic environment. It then focuses on a of this stimulus clearly makes it much more difficult to crucial question for the strategy of ‘more and better reach the Lisbon employment targets on schedule.

jobs’: whether and in what sense there are trade-offs

between employment growth and productivity growth. Nevertheless, disappointing performance cannot be ascribed entirely to the cyclical downturn. First, EU

Section 3 reviews the best available evidence on the leaders were presumably aware that the favourable determinants of labour market performance, and conditions of 1999 and 2000 might not continue for a compares this with actual performance and policywhole decade. Secondly, had employment continued to making in the EU-15 since 2000. It then turns to critical grow rapidly, further structural improvements in labour groups in the labour force in which significantly lower market performance would still have been required in unemployment and/or higher labour force participation order to avoid inflationary pressures. Labour markets in would appear to be necessary conditions for approaching some EU countries were already showing signs of the Lisbon employment targets. tightness in 2000. Thirdly, the cyclical downturn in the

labour market has not been particularly severe. Indeed as Section 4 reviews the priorities established for labour

market reforms in EU Member States and compares Graph 1 shows, the employment rate appears to be at progress achieved since 2000 to what is likely to be about its equilibrium level.

required in order to hit the Lisbon targets. Most of the chapter focuses on the former EU-15, to which the

Lisbon strategy initially applied. However, Section 4 also looks at the labour market challenges faced by the ten new Member States, and at how they will fit into the 200 Assuming, as projected by Eurostat, a roughly constant Lisbon strategy. working-age population.

201

  See the EU E CONOMY 2002 R EVIEW , Chapter 2.

Graph 1: Structural employment and the employment gap

65 74.5

64 74

73.5 63

73

62 72.5

61 72

71.5 60

71

59 70.5 1990 1992 1994 1996 1998 2000 2002 2004 1990 1992 1994 1996 1998 2000 2002 2004

Employment rate EU-15 Employment rate trend EU-15 Employment rate USA Employment rate trend USA

3 3

2 2

1 1

0 0

-1 -1

-2 -2

-3 -3 1990 1992 1994 1996 1998 2000 2002 2004 1990 1992 1994 1996 1998 2000 2002 2004

Gap between actual and employment rate trend EU-15 Gap between actual and employment rate trend USA

Gap between actual and potential gross domestic product EU-15 Gap between actual and potential gross domestic product USA

Notes: The trend employment rate is calculated using Commission services and OECD estimates of the NAIRU (for the EU and the US respectively) and HP-filtered labour force participation. For both the EU and the US, employment and participation rates are calculated for the population aged 15- 64. Source: Commission services, OECD for US data.

The upper panel of Graph 1 shows estimated ‘structural cyclical response of employment appears more moderate employment’ in the EU and the US, which combines in the EU than in the US, even after controlling for the

estimates of the NAIRU 202 and trend labour force wider fluctuations of the output gap in the US. This is

participation. The lower panel compares the despite the growing use of temporary contracts in the ‘employment rate gap’ – the deviation of the actual EU, which many expected to lead to an increase in the employment rate from its estimated structural level – sensitivity of labour demand to the business cycle (see

with the output gap, or the deviation of actual GDP from Box 1). its potential level. While for the EU the employment rate gap is almost zero, in the USA it is around minus 1.4 percentage points – i.e. the actual employment rate is 1.4

percentage points below its structural level. 203 The

202 Non-accelerating inflation rate of unemployment: the

‘structural’ rate of unemployment that is consistent with constant inflation.

203 With a US working age population of 192 million, this

implies a gap between potential and actual employment of 2.7 million jobs, not far from the “low” estimate (3.5 million) recently obtained by the Federal Reserve Bank of Boston (2004) assuming a NAIRU of 5.5 and a participation rate of 66.3 per cent. When the same calculation is done ‘baseline’ and ‘high’ estimates (assuming, respectively, a using OECD data on the population aged 15 years and over NAIRU of 5 per cent and 4.5 per cent and a participation (as opposed to aged 15-64), which corresponds more rate of 66.8 per cent and 67.3 per cent) indicate a somewhat closely to the usual US definition of the employment rate, larger employment gap, of 5.2 million and 6.9 million jobs the implied gap is 2.8 million persons. The Boston Fed’s respectively.

Box 1: The impact of temporary contracts on the cyclicality of employment rates

The use of temporary contracts in EU countries has grown rapidly over recent decades. These types of contracts, once a relatively rare exception to the rule of permanent employment, now represent a significant share of total employment in the EU-15: around 13 per cent in 2003 compared with 7 per cent in 1987. However, the share varies a lot across countries, ranging from 2 per cent in Luxembourg to 31 per cent in Spain in 2003.

Whether or not the growth of temporary contracts affects the overall level of employment, one might expect it to modify the response of participation and employment to the business cycle. Firms can choose not to renew temporary contracts in downturns, while, in periods of recovery or expansion, they can quickly hire new staff without running the risks associated with high firing costs. However, the legal restrictions on the renewal of temporary contracts may limit their use as a cyclical buffer. As surveyed by Portugal and Varejão (2003), other reasons to use temporary contracts include screening and ‘churning’. Screening occurs when employees are offered temporary contracts to see if they are suitable for permanent positions, while churning refers to the practice of successively hiring different employees on fixed-term contracts to fill the same permanent position.

To shed light on this issue, we checked in a panel of EU-15 countries whether the sensitivity of employment to the output gap is modified by the introduction of temporary contracts. The employment rate, E it , is explained by the output gap, OG it , the interaction between the output gap and the share of temporary contracts, S it , country dummies (fixed effects) α i and countryspecific trends, t i , capturing the heterogeneity across EU-15 countries. The equation is estimated for both male and female employment rates, where γ is the parameter of interest that captures the impact of the temporary contracts on the cyclical components of employment rate:

E = α + λ + β + γ * − + ε = α + λ + ( + ) + i t i i t OG i t OG i t S i t 2 i t i i t β γ S i t − 2 OG ε i t i t

Initial results (Table below) suggest that temporary contracts have a small pro-cyclical effect on male employment but are insignificant for female and overall employment. For men, the magnitude of this effect is fairly modest, albeit not negligible: an increase of 10 percentage points in the share of temporary contracts would raise the (positive) impact of a 1 per cent output gap on the male employment rate by around 0.2 percentage points.

Table: Panel data results 1985-2003 for EU-15 countries

total female male Output gap 0.200** 0.192* 0.159*

(2.51) (1.81) (1.91) Output gap * share of temporary jobs 0.009 -0.001 0.019**

(1.25) (-0.13) (2.42) Observations 239 239 239

Notes: Absolute value of t statistics in parentheses. Significance at 10% and 5% denoted by * and ** respectively. The equations are estimated by GLS allowing for heteroskedastic errors and common-across-group first order serial correlation. The share of temporary contracts jobs is lagged by two years to avoid any problem of endogeneity between the share of temporary contracts and the employment rate. Data are annual and start from 1986 for Portugal, from 1987 for Spain, for 1995 only for Denmark, Sweden and Finland. Source: Eurostat Labour Force Survey; DG-ECFIN, AMECO database.

When the same regression is run separately for periods of slowdown and expansion, the procyclical impact of temporary contracts on male employment appears to be slightly stronger in downturns. This may point to the role of temporary contracts as a cyclical buffer, though the smaller procyclical effect in periods of expansion is consistent with their use as a screening device as well. Conversely, for women, the effect of the share of temporary contracts is insignificant throughout the business cycle. This could indicate churning of temporary contracts in services, where female employment is concentrated.

In any event, temporary contracts do not, at first sight, appear to have a large impact on the cyclicality of total employment, although their effect on male employment might not be negligible. Further analysis, perhaps disaggregating further by economic sector or by age group and including missing explanatory variables, would be required to draw firm conclusions.

Table 1 : Labour market performance in EU-15 Member States

Employment rate Unemployment

All Female Older workers All Youth Long term Regional disparities *

Country 2003 2003 2003 2003 2003 2003 2002 BE 59.6 51.8 28.1 8.1 21.5 3.7 0.48

DK 75.1 70.5 60.2 5.6 10.3 1.1 :

DE 65.0 59.0 39.5 9.6 11.1 4.6 0.63

EL 57.8 43.8 42.1 9.3 26.3 5.1 0.16

ES 59.7 46.0 40.8 11.3 22.7 3.9 0.38

FR 63.2 57.2 36.8 9.4 20.9 3.5 0.22

IE 65.4 55.8 49.0 4.6 8.3 1.5 :

IT 56.1 42.7 30.3 8.6 27.0 4.9 0.78

LU 61.8 50.0 34.2 3.7 10.4 0.9 :

NL 73.5 65.8 44.8 3.8 6.7 1.0 0.16

AT 69.2 62.8 30.4 4.1 7.0 1.1 0.43

PT 67.2 60.6 51.1 6.3 14.4 2.2 0.27

FI 67.7 65.7 49.6 9.0 21.8 2.3 0.31

SE 72.9 71.5 68.6 5.6 13.4 1.0 0.17

UK 71.8 65.3 55.5 5.0 12.3 1.1 0.30

EU25 62.9 55.1 40.2 9.1 18.4 4.0 :

EU15 64.4 56.0 41.7 8.1 15.9 3.3 0.63

• * 

  Coefficient of variation = Standard deviation of NUTS 2 regional unemployment rates / National average unemployment rate.

Source: Commission services.

Regarding the immediate prospects for employment, increased by less than one percentage point between there are several different possible interpretations. First, 2000 and 2003, the female employment rate rose by two and most pessimistically, the scenario of a delayed percentage points, in line with strong growth in reaction to the downturn in output still cannot be women’s labour force participation. Indeed, there is a excluded. This was the case in the early 1990s, when the chance that the Lisbon target for a female employment employment gap continued to fall even after GDP rate of more than 60 per cent by 2010 could still be growth had resumed. Secondly, if the employment rate reached. The contribution of males to employment is indeed at around its equilibrium level, then the growth was actually negative in Denmark, Germany, nascent recovery might soon encounter the obstacles of Austria, Portugal and Finland. This is explained mainly low labour force participation and high structural be falling employment of younger people, partly due to unemployment. In this scenario, employment growth increased enrolment in education, though even the might remain relatively flat for several more years, employment rate of prime-aged men fell in some assuming that there are no further increases in structural countries. employment in the pipeline, due for instance to the

delayed effects of earlier reforms. Older workers have made a remarkable contribution in recent years. The employment rate of 55-64 year-olds

Thirdly, however, the experience of the 1990s suggests rose by as much as four percentage points between 2000 that structural improvements to accommodate sustained and 2003, accounting for around half of total employment growth without excessive inflationary employment growth. Reforms of pension systems and pressures are a feasible scenario. This depends, of early retirement schemes – in some cases decided in the course, on sufficient progress on reforms having been early 1990s, but with phased-in implementation – have made, or at least on expectations of continued growth in begun to take effect. Particularly strong increases were the structural employment rate in the EU (see Graph 1, recorded in Finland, France and the Netherlands: 8, 7.4 top-left chart). This is consistent with the idea that and 6.6 percentage points respectively. Nevertheless, the employers in the EU have hoarded labour during the employment rate of older workers, at just under 42 per recent downturn, anticipating that employment rates cent in 2003, remains far short of the Stockholm target would continue to grow and that recruitment difficulties of 50 per cent by 2010.

might quickly re-emerge. So far, this chapter has looked at labour market Graph 2 breaks down progress since 2000 by age group performance for the EU-15 as a whole. But of course

and gender. While the overall employment rate performance varies a great deal among Member States, ranging from good to exemplary in the cases of

Denmark, Ireland, the Netherlands, Austria and Then, obviously, higher employment will be inevitably Portugal, Sweden and the UK, and from bad to worse in associated with lower output per worker and vice versa. Belgium, Germany, Greece, Spain, France and Italy, Thus, in such a comparative-static setting it is easy to with Finland and Luxembourg somewhere in between. construe a situation where, for example, regulations and Table 1 provides a snapshot of performance by Member restrictions excluding low productivity workers from State. Although many Member States are still far from employment result in a higher level of actual labour complying individually with the Lisbon employment productivity, but it will come at the price of lower targets, it should be recognised that some, including employment; similarly, reform efforts to price back low Spain and Greece, have made significant progress in productivity workers into employment will mean more

204

recent years. jobs, but this will be associated with lower overall productivity.

Graph 2: Increased employment rates, 2000-2003, EU-15 In comparing labour productivity levels across countries,

10 such considerations of a comparative-static nature can be useful. There appears to be widespread agreement

8 that measured labour productivity in Europe relative to

the US may be upward biased as a result of the

ts 6 in exclusion of more low productivity workers. Indeed, the

e po 4 EU employment rate falls short of the US level by some

nt ag 10 percentage points, with lower participation rates and

ce 2

P er higher unemployment rates disproportionately affecting

low skill workers. In a similar vein, the capital-labour

0 ratio appears to be typically higher in the EU than in the

-2 US, driving up measured labour productivity in Europe.

BE DK DE GR ES FR IE IT LU NL AT PT FI SE UK EU15 Thus, both economic theory and quick inspection of a Total working Age Population (y15_64) Older workers (y55_64) few aggregate figures suggest that one should control for Female empl. Rate (y15_64) these effects in productivity comparisons. Obviously, in

Source: Commission services. consequence, a Europe at full employment may well see a significantly larger labour productivity gap vis-à-vis

the US than the current actual figures suggest.

2.2 Employment and labour productivity: By how much could the productivity gap rise? A simple reconsidering the potential trade-off calculation could be performed focussing on

comparisons of total factor productivity levels, using the 2.2.1 Introduction following relationship:

At the moment, EU GDP per capita in purchasing power (1) Y/L = (K/L) 1-α TFP

parities is around 70 per cent of the US level, with 1/3 of

the gap due to productivity differentials and 2/3 due to a where Y/L denotes measured labour productivity, TFP is lower labour input, i.e. a lower employment rate and total factor productivity, K/L is the capital intensity of fewer hours worked compared with the US. production and 1-α is the capital-elasticity of output in Consequently, improving the EU’s productivity the constant-returns Cobb-Douglas case. For the

performance and raising employment is fundamental to increasing the long-term growth potential of the EU economy. However, several observers have argued that the twin goals of raising both employment rates and productivity growth may be difficult, or even impossible to pursue simultaneously, given a perceived negative trade-off between employment and productivity.

The basic argument for the existence of a negative relationship between employment and productivity is derived from straightforward comparative-static reasoning. For any standard production function, average factor productivity will decrease with rising output as the expansion of production will require to bring less and less productive factors into operation – less fertile soil, older and less efficient equipment and machinery, workers with lower abilities and skills, etc.

204 See Chapter 1 in European Commission (2004a) for further

details.

Graph 3: TFP and labour productivity gap Actual labour productivity growth can of course deviate from the balanced labour productivity growth rate over 200 the short-to-medium term due to capital-labour substitution; faster than “balanced” productivity growth

180 Hourly labour productivity indicates labour shedding, and a shortfall of actual

160 Total factor productivity relative to “balanced” productivity growth is a

140 characteristic of what is loosely called labour-intensive 00 120 growth. Obviously, then, the employment neutrality

 1

 = 100 hypothesis will not hold over the short-to medium

U S

A

80 term.

206

  In consequence, pressing ahead with labour

60 market reforms may entail a temporary reduction in

40 measured productivity growth below full potential, but this should not be regarded as a trade-off in any sense. A

20 higher employment rate implies an unambiguous

0 increase in GDP per capita with no negative BE DK DE EL ES FR IE IT LU NL AT PT FI SE UK US EAEU15 implications for the long-run productivity growth of the

Source: Commission services. existing workforce. Thus, there is no inherent problem to act on both fronts simultaneously, raising the

“balanced” rate of productivity growth using all the

calculation, GDP and capital stock in PPP are taken available instruments to stimulate TFP growth, whilst at from AMECO. Employment is civilian employment the same time encouraging the labour-intensive growth (LFS). Hours worked come from the GGDC (Groningen in the medium term that is needed to move towards full

Growth and Development Centre). The results of this employment. 207

simple exercise, shown in the graph below, suggest that

the productivity gap between the euro area and the US 2.2.2 The dynamic employment-productivity may be by some 6 percentage points wider than the relationship in recent years

actual figures indicate. EU employment and productivity growth patterns have

However, the notion of a negative relationship between diverged sharply over recent years. Compared with the employment and productivity levels emerging in first half of the 1990’s, the period since then has comparative-static considerations should not be witnessed a significant increase in the contribution of confused with a genuine trade-off between employment labour to EU GDP growth but unfortunately this has and productivity in a long-run dynamic sense. One of been accompanied by a reduction in the contribution the “big” stylised facts in economics is that in the long from labour productivity, with labour productivity run technical progress is neutral with respect to growth having come down by about one percentage employment. History has told us that the process of point. By comparison, over the same timeframe, the capital accumulation and technological innovation has USA has been able to combine a strong employment not meant the “end of work” and despite notions of performance with acceleration in labour productivity “factories without workers”, it is clear that from an growth. Against this background, this section overall perspective workers have not been replaced by investigates to what extent the recent slowdown in machines. In standard economic growth theory this labour productivity growth may merely reflect a long-run neutrality proposition has been captured by the response to a series of positive shocks to labour supply

concept of labour-augmenting technical progress. 205 and jobs emanating from structural reforms and

Along this balanced growth path, labour productivity, employment-friendly wage developments.

real wages and the capital intensity of production grow

at the same rate, driven by (exogenous) technical Graph 4, as a starting point for the analysis, shows the progress. Technical progress is called total factor contribution to growth from employment measured in productivity growth, indicating that this concept should total hours worked and from labour productivity, with not be seen in a narrow “engineering” sense. Given that the US included for comparison purposes. Evidently, TFP determines our standards of living in the long run, productivity growth has further slowed down over the

clearly policy makers want it to grow faster than in recent years.

206

  Gordon (1995) provides a neat theoretical and empirical

205 Labour augmenting technical progress is equal to Harrodinvestigation how a productivity-unemployment trade-off neutral technical progress when the capital stock grows at might emerge and how it will subsequently be eliminated the same rate as output, thus leaving the capital-output ratio through a dynamic path of capital adjustment. constant. For a Cobb-Douglas production function this 207 Obviously, misguided policies attempting to exploit a

“balanced” labour productivity growth rate is defined as perceived trade-off have to be avoided, for example

TFP growth divided by the labour share. unnecessary regulations leading to “overmanning”.

Graph 4: Labour input (employment + hours worked) and labour productivity per hour trends (1966-2002)

2.5 7.0 EU labour input

2 US labour input 6.0 EU productivity

1.5 US productivity 5.0

1 .

4.0

 p .a

.

0.5 p

.a

 c hg c hg

% % 3.0 0

2.0 -0.5

-1 1.0

-1.5 0.0

1966 1978 1990 2002 1966 1978 1990 2002

Source: Commission services.

1990s, with the EU’s long established superiority in Thus, the analysis distinguishes between three shocks, terms of labour productivity growth having disappeared shocks to employment, shocks to productivity and over recent years. It is also striking that the contribution shocks to aggregate demand and makes an effort to of employment to growth in the US, albeit partly due to measure their relative importance for productivity and immigration, has been consistently higher than in the employment. What is of specific interest in the context EU, even during the recent period of ‘jobless growth’ in of this section is the dynamic response of productivity to the USA. structural employment shocks. In technical terms, a

The benign interpretation of the observed productivity structural VAR methodology is used to estimate a model growth trends sees the recent performance deterioration in the three variables employment, productivity and mainly as the mirror image of structural labour market inflation for the euro area, applying a procedure improvements. Under this view the EU may now simply suggested by Stock and Watson (1988) and Blanchard be in a transition phase whereby wage moderation and and Quah (1990) to identify the shocks and estimate positive labour supply shocks may have initially created structural relationships. The identifying restrictions a negative trade-off between employment and implied by a standard neoclassical growth model and productivity growth, basically via a temporary decline in used in the present analysis are the following.

208

capital-labour substitution; however, the dynamic • The labour market shock can have short and long run adjustment path towards a new equilibrium with higher effects on employment, productivity and inflation. employment and lower structural unemployment will

also involve capital accumulation that should eliminate • The productivity shock can have long run effects on the trade-off over the medium-term. The more productivity and inflation, but only short and pessimistic view, on the other hand, is that the labour medium run effects on employment.

productivity growth slowdown reflects a genuine • The demand shock can have a long run effect on negative shock, either in the form of a decline in total inflation only, but not on employment and factor productivity growth or additional pressures on productivity. capital productivity; clearly, in such a scenario,

prospects for a recovery of labour productivity growth The empirical results are presented in two steps. First,

are much bleaker. the impulse responses from the estimated VAR are discussed. These responses give the impact on

Obviously, both interpretations are likely to contain an employment and productivity of a unit shock to element of truth, posing the analytical challenge to employment, productivity and demand. Recall that the derive inference on the relative magnitude of the identifying restrictions imply that temporary unit shocks employment and the productivity shock and their to employment can have permanent effects on respective consequences for overall productivity and employment and productivity, while a unit shock to employment developments. The picture is complicated demand (inflation) can only have temporary effects. In by a third possible factor, namely aggregate demand. order to evaluate the quantitative magnitudes of these Indeed, a comprehensive analysis has to allow for the shocks, they are compared to similar shocks simulated possibility of positive or negative shocks to demand with the Euro area QUEST model. This comparison is affecting output, employment and productivity in recent useful since it shows whether orders of magnitudes of years.

208

  See Box 2 for a formal description.

these shocks are similar when two very distinct 2003Q4 211 in order to derive an estimate for the

empirical tools are used, with the VAR model imposing structural component in employment growth and its very little economic structure apart from the long run likely impact on productivity. The results of this constraints, while QUEST consists of explicitly exercise are depicted in Graph 7. The cumulated size of estimated structural equations and estimated adjustment the employment shock over the period 1995-2003 is lags. estimated at about 5 per cent. Thus, roughly one half of

Employment shock: A positive employment shock the overall observed employment expansion over that initially leads to an increase in productivity; however, period is attributed to structural trend improvements. this short run positive effect in the VAR model is partly According to the VAR approach the cumulated spurious. 209 In the medium and long run the effect on productivity cost of this structural employment productivity is negative, i.e. an increase in employment expansion may have amounted to ¾ of a percent; the is associated with a decrease in labour quality. Note, QUEST model simulations would put the productivity though, that this negative long run effect is estimated to cost somewhat higher at 1 ½ per cent; roughly translated be small: a shock which leads to a permanent increase in into year-on-year figures, this implies a reduction in the level of employment of about 1 per cent is associated annual productivity growth of around two tenth of a with a long run productivity level effect of about - percentage point, equivalent to some 20 per cent of the 0.1 per cent. 210 Analysis based on QUEST model observed total productivity growth slowdown, which simulations yields fairly similar result to the VAR could be attributed to positive structural shocks in the

approach, but the negative impact upon the long run labour market.

productivity level is slightly stronger (-0.3 instead of - The empirical results presented above are quantitatively 0.1); moreover, the QUEST model analysis does not broadly in line with other available evidence on reveal any short run increase in productivity. The structural labour market improvements as indicated by a productivity effect remains negative in the QUEST trend increase in participation and a reduction in model over the entire simulation period of 10 years, structural unemployment. Moreover, relating the reflecting the long lasting dynamics of capital-labour productivity effect to real wage moderation also substitution induced by labour market reforms. suggests that the estimated impact on short-run However, the QUEST model does not distinguish productivity developments is of a reasonable order of between different skill levels and, in consequence, there magnitude. A stylised number for real wage moderation will be no productivity impact over the very long term. in the past 10 years or so would put the average annual

Productivity shock: A positive productivity shock is reduction in real efficiency wages at slightly less than ½ associated, in the short-run, with a small negative of a per cent. Thus, back-of the envelope calculations employment effect. The order of magnitude of the would suggest that real wage moderation could, on employment effect is only about one tenth of the size of average, have reduced annual actual labour productivity the productivity shock. Again, in the QUEST model growth relative to its balanced steady-state rate by about analysis a qualitatively similar pattern to the VAR two tenth of a percentage point, which is well within the emerges, but the short-run negative employment range derived from the VAR and QUEST model

response appears to be somewhat stronger. approaches. Further corroborating evidence stems from growth regressions suggesting that about 25 per cent of

Demand shock: The demand shock is initially associated the productivity decline is due to the increase in

with a positive employment and productivity effect. This employment. 212 In summary, and recalling that the

result appears quite plausible, since a demand shock is overall slowdown in average annual productivity growth likely to lead to better capacity utilisation in the short has amounted to about one percentage point, it emerges run. As the demand effect fades away and employment as a fairly robust result that only some 20 per cent of this is slow to adjust, the productivity effect turns negative reduction can be attributed to the dynamic response of and dies out within a year. productivity to positive structural shocks in the labour

The identifying restrictions of the VAR model allow market.

calculating the structural shocks from the estimated residuals. In the second step of the empirical analysis the shocks are cumulated over the period 1995Q1 to

209 This results from the labour input series being employment

rather than hours worked. A positive employment shock is likely to be correlated with a short run increase in hours

worked, which by definition will be associated with an 211 It should be noted that this provides an estimate for the

increase in labour productivity per worker. overall magnitude of the shocks, but not of the impact these

210 In fact, the hypothesis of a zero long run productivity effect shocks have had on the macroeconomic aggregates. cannot be rejected at standard significance levels. 212 See EU E CONOMY R EVIEW 2003, Chapter 2, Table A3.

Graph 5: Impulse response analysis

0.9 0.15

0.8

0.7 0.10

0.6 t

en 0.05 m 0.5

oy ct

iv ity Employment shock: productivity response

0.4 du

E m

pl

P ro 0.00 0.3 Employment shock

0.2 -0.05

0.1

0.0 -0.10

0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 Quarters Quarters

0.00 1.00

-0.01 0.99

0.98

-0.02 0.97

t -0.03 0.96 en m

oy -0.04 ct

iv ity 0.95

du 0.94

E m

pl

-0.05 Productivity shock: employment response P ro 0.93

Productivity shock

-0.06 0.92

0.91

-0.07 0.90

-0.08 0.89

0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76

Quarters Quarters

0.04 0.20

0.04 0.15

0.10

0.03 0.05

t 0.03 0.00 en

m -0.05 oy 0.02 ct

iv ity

du -0.10

E m

pl

0.02 P ro -0.15

0.01 Demand shock: employment response -0.20

-0.25 Demand shock: productivity response

0.01 -0.30

0.00 -0.35 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76

Quarters Quarters

Source: Commission services.

Graph 6: QUEST analysis

Employment shock Productivity shock

1.0 1.00

0.8 0.80

0.6 0.60 0.4 Employment

0.40

0.2 Capital labour ratio

Labour productivity

0.0 0.20

-0.2 0.00

-0.4 -0.20

Employment -0.6

Capital labour ratio

-0.8 -0.40 Labour productivity

-1.0 -0.60 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Years Years

Source: Commission services.

`Graph 7: Cumulated Shocks (based on VAR Analysis): Historical decompositions 1995-2003

0.06 0.0080

0.05 0.0060

0.04 0.0040

0.0020 0.03

Employment 0.0000

0.02

-0.0020 Productivity 0.01 -0.0040

0.00 -0.0060

-0.01 -0.0080

1995 1996 1997 1998 1999 2000 2001 2002 2003 1995 1996 1997 1998 1999 2000 2001 2002 2003

Source: Commission services.

2.2.3 Conclusions • The dynamic response of productivity to positive labour supply and wage shocks may entail a

In a nutshell, the analysis in this section dismisses the temporary reduction in productivity growth rates, notion of a genuine trade-off between employment and which, in principle could be considered as benign; productivity growth. Obviously, misguided policies to anyway, the size of a negative effect of this type is exploit such a trade-off have to be avoided. However, estimated to be fairly small. there are no reasons to think that structural labour

market reforms boosting employment will typically • The increase in employment since the mid 1990s has entail negative implications for longer-term productivity indeed been to a significant extent the result of such growth. In particular, this section reaches the following positive labour market shocks, with about one half of

conclusions: the additional jobs attributed to structural improvements.

• The negative relationship between productivity and

employment in comparative-static considerations • Positive employment shocks can only account for a

should not be interpreted as a genuine trade-off. very small fraction of the observed productivity slowdown in recent years; consequently, the decline

• However, all else equal, a move towards full of labour productivity growth must be considered as employment is likely to see a widening of the labour predominantly caused by other factors and probably productivity gap between Europe and the USA. not just a temporary phenomenon.

Box 2: Identifying structural shocks for Employment, Productivity and Demand with a Structural VAR

Model

We use a structural VAR (SVAR) methodology, based on Stock and Watson (1988) and Blanchard and Quah (1990), for the identification of structural shocks. The intuition for shock identification in Blanchard and Quah is based on the idea that demand shocks only have temporary effects while supply shocks have permanent effects. Stock and Watson extend this approach and allow for separate supply contributions from labour and productivity (TFP). In order to identify different supply contributions, namely those coming from employment and those coming from productivity, additional identification criteria must be introduced. Stock and Watson use long run restrictions implied by the neoclassical growth model for that task. The neoclassical growth model appears to be suitable, since there are at least three important features in the long run trends which are compatible with this model:

• 1. 

  There is a close trend correlation between the growth of labour productivity and capital intensity.

• 2. 

  Capital intensity and productivity grow at a similar rate in the long run.

• 3. 

  If one looks over long periods of time and across the EU and the USA, the employment rate appears to be unrelated with productivity growth.

If one uses the neoclassical growth model then one can impose the following long run structure on the data:

• 1. 

  The labour market shock can have short and long run effects on employment, productivity and inflation.

• 2. 

  The productivity shock can have long run effects on productivity and inflation but only short and medium run effects on employment. This constraint arises from the assumption that real wages are indexed to productivity in the long run.

• 3. 

  The demand shock can have a long run effect on inflation only but not on employment and productivity. No long run constraint is imposed on inflation.

These three types of restrictions imply a triangular long run structure between the growth rate of employment (Δh), productivity (Δ(y-h)), and inflation π on the one hand and the corresponding shocks to employment (v), productivity (e) and demand (d) on the other. If one defines the vector Δx t =[Δx t , (Δ(y t -h t ), Δπ t ] and the vector ζ t = [v t , ,e t , d t ] , then the moving

average representation of this model is given by :

 a 11 0 0 

Δ x t = A ( L ) ξ t A ( 1 ) =  a  with  21 a 22 0 



 a  31 a 32 a 33 

where the matrix A(1) shows the long run restrictions. Notice, this particular structure is particularly suited to test for the short, medium and long run effects of an employment shock. Allowing for a non-zero long run productivity effect of an employment shock allows one to test for labour quality effects associated with a permanent change in the employment rate. A similar analysis has been conducted by Galí (1999). He is mainly interested in the employment effects of productivity shocks.

The implications of the above findings for the Lisbon and, in consequence, the dismal productivity strategy are straightforward: Indeed, “the more jobs the performance in recent years cannot be attributed to better” may serve as a simple catch-phrase labour market reform efforts. characterising the principal goal of labour market reform efforts since there is no genuine trade-off – in the sense

of a difficult decision to be made – between policies to 3. The determinants of labour market

raise the employment rate and policies to foster performance

productivity growth. Of course, misguided policies

attempting to exploit such a trade-off have to be avoided 3.1 Determinants of overall performance

– if, for example, policy-makers promoted sectors with

low productivity growth prospects, if they introduced 3.1.1 The impact of labour market institutions on unnecessary regulations leading to “overmanning”, if labour market performance

they discouraged young people from pursuing further

education, or if they used funds for public training A realistic account of the role of labour market programmes in an unproductive manner, then institutions in influencing labour market performance employment might be raised at the expense of longermust start from the recognition that the assumptions term productivity potential. However, none of these behind the textbook model of a competitive economy – policies is advocated in the EU economic and complete markets, perfect information, atomistic and employment policy framework. There is no inherent homogeneous agents, perfect competition – are often problem with the logic of the European Employment violated where labour markets are concerned. Nominal Strategy, i.e. attempting simultaneously to raise wages are the result of negotiations between employers

employment growth and labour productivity growth, and employees, while firms set price as a mark-up over labour costs. While in the short run unemployment is

determined by real aggregate demand, in the long run it widely cited are summarised in Table 2. Among these, converges toward the level which is compatible with a three main strands may be identified.

stable inflation rate. 213 In this framework, labour market

policies influence the stock of employment and In a first group of studies, indicators of labour market unemployment in three ways: by modifying the wage institutions are used to explain cross-country differences

formation mechanism; by changing the price elasticity in unemployment rates.

217

  Unemployment is positively

of product demand; and by stimulating technological associated with generous unemployment benefits, a high

progress. 214 tax wedge, and high union coverage and negatively associated with active labour market policies

Another way to look at labour market performance is as (ALMPs) 218 and a high degree of coordination in wage

the outcome of a process of matching between workers bargaining. The role of employment protection

and job vacancies. 215 The heterogeneity of workers and legislation and union density is uncertain. However, a

jobs, imperfect information about the characteristics of large part of the change in structural unemployment potential employers and employees and restrictions on remains unexplained. Moreover, some of the labour mobility all generate labour market frictions. unfavourable institutions were already in place in the These in turn influence labour market flows. In the 1960s in many EU countries, when European

steady state, inflows into unemployment are equal to unemployment was lower than in the USA. outflows from unemployment, and there is an inverse

relationship (the Beveridge curve) between the number A second group of studies focuses on the interactions of vacancies and the unemployment rate. Anything that between labour market institutions and macroeconomic

improves the efficiency of the matching between shocks.

219

  The essence of these is that transitory

unemployed people and vacancies and/or increases the increases in unemployment due to shocks may be exit rate from unemployment will shift this curve prolonged by labour market institutions that restrict inwards and reduce the steady-state level of labour market flows and protract the adjustment of

unemployment for a given number of vacancies. wages. There is not a full consensus on the impact of different institutions. In Blanchard and Wolfers (2000),

In both the stock and the flow approaches to equilibrium for instance, all the ‘usual suspects’ except union density unemployment, labour market institutions affect firms’ are significant and with the expected sign. In Nickell et hiring and firing decisions and individuals’ readiness al. (2003), benefit duration, union density and low and willingness to take up a job, as well as the extent to labour mobility shift the Beveridge curve outwards, which unemployment reins in inflationary pressures. which implies higher equilibrium unemployment, while

Institutions such as unemployment and welfare-related employment protection legislation shifts it inwards. benefits, wage bargaining, labour market regulation and

labour taxation thereby influence the equilibrium rate of A third important strand is studies, including some of

unemployment. those already cited, that look at interactions between different labour market institutions. Coe and Snower

However, the impact of labour market policies and (1997) argued theoretically that a wide range of reforms on labour market performance is often institutions may have complementary effects on ambiguous, at least in theory. For example, it is well unemployment. In Belot and Van Ours (2001, 2004), known that unemployment benefits are subject to moral institutions strongly influence performance when they hazard, since job-search effort cannot be fully observed. reinforce each other. This means that it is harder to Benefits thereby reduce the incentive to find a job and predict the response of equilibrium employment to raise reservation wages. But, in search models with riskchanges in a single institutional variable in isolation. averse workers and imperfect capital markets, the Belot and Van Ours find, for example, find that high absence of unemployment insurance may lead people to labour taxes and benefit replacement rates combine to accept jobs too quickly, in the sense that further search weaken the financial incentives for employment, and for a higher productivity match would increase overall that this interaction has driven the evolution of

welfare. In this case, unemployment benefits do not unemployment rates in several countries. work as a search subsidy but as a way to deal with

imperfect insurance. 216

During the 1990s there has been a wealth of studies focussing on the effects of institutions on employment performance. The main results of several of the most

217

  See Elmeskov et al. (1998); Nickell and Layard (1999).

218

  It should be noted that placement of the unemployed in labour market programmes automatically reduces the

213 See, e.g., Blanchard (1986); Layard et al. (1991). number of people registered as unemployed. When one

214 includes in the definition of unemployment also those See the EU E CONOMY R EVIEW 2002, Chapter 2.

participating in such programmes, the impact of ALMPs is

215 See, for example, Mortensen and Pissarides (1999). usually more uncertain.

216 See Acemoglu (1999) (2001); Acemoglu and Shimer 219 See Blanchard and Wolfers (2000); Fitoussi et al. (2000);

(2000). Bertola et al. (2001); Nickell et al. (2002).

Table 2: Labour market institutions and labour market performance

Study Countries and Periods Institutions considered Results

• 1. 

  Aggregate performance

Tax wedge (TW) Small positive effects. Positive and significant only in countries

Gross replacement rate (GRR) with intermediate co-ordination Positive effects, larger in countries that spend more on ALMPs

Static Panel data on 19 Spending on ALMPs (ALMPU) Negative effects if Sweden is excluded

Elmeskov et OECD countries over the EPL Positive effects. Positive and significant only in countries with al. (1998) period 1983-1995 (GLS intermediate co-ordination

random effects). Minimum wage (MW) Insignificant effects Co-ordination/Centralisation Negative effects in high centralised/co-ordinated and decentralised (CO) countries

Insignificant effects Union density (UD)

Effects on total unemployment Effects on long-term

Positive effects unemployment

Positive effects Positive effects Tax wedge (TW)

Positive effects Insignificant Gross replacement rate (GRR)

Negative effects Positive effects Nickell and Cross Section on 20 OECD Benefits Duration (BD) Negative effects

Layard countries (GLS random Spending on ALMPs (ALMPU) Negative effects Insignificant (1999) effects) EPL Positive effects Negative effects Co-ordination (CO) Positive effects Insignificant Union density (UD) Positive effects Positive effects

Union Coverage (UC) Insignificant Owner Occupation rate Effects on employment rate Similar effects. UD, UC, GRR ALMP insignificant Tax wedge (TW)

Gross replacement rate (GRR) Positive effects

Static Panel data on 20 Positive effects. Among most significant when interacted with

OECD countries over the Benefits Duration (BD) shocks

period 1960-1995. Positive effects. Among most significant when interacted with

Blanchard Spending on ALMPs (ALMP)

shocks

and Wolfers Interactions of time fixed EPL

Positive effects Positive effects but weaker when Spain is dropped from sample

(2000) institutions with TFP, real interest rate and labour Minimum wage (MW) Positive effects

demand shocks are Co-ordination/Centralisation Positive effects. Among most significant when interacted with

considered with non-linear shocks

least squares Union density (UD) Positive effects. Among most significant when interacted with shocks

Union Coverage (UC) Insignificant effects

Two steps approach. Macro-variables: world real First step: Over the period interest rate , trend labour 1960-1998 for 19 OECD productivity growth, ratio of non At least 50% of cross country differences in unemployment and in countries, a dynamic panel wage support to labour sensitivity to shocks are explained by labour market institutions (fixed effects) estimate of productivity, direct taxes, payroll

Fitoussi et al. unemployment persistency taxes, inflation rate

(2000) and sensitivity to macro cross country differences in unemployment are a positive function shocks is obtained. Labour market institutions: of GRR, UD, CO and a negative of UC

Second step: Cross section Replacement rate (GRR), benefit of (short- and long-run) duration (BD), union density, cross country differences in sensitivity of shocks are a positive

fixed effects and sensitivity (UD) union coordination (CO), function of BD, UD and a negative CO and ALMP coefficients to labour union coverage (UC), active

market institutions labour market expenditure (ALMP)

Effects on unemployment rate

Tax wedge (TW) Positive effects. Larger in countries with high degree of bargaining co-ordination

Gross replacement rate (GRR) Positive effects. Larger in countries where the duration of unemployment benefits is high

Nickell et al. Dynamic Panel data on 20

(2003) countries over the period Benefits Duration (BD)

Positive effects

1961-1995. (GLS estimates) EPL Insignificant effects

Co-ordination (CO) Negative effects

Union density (UD) Positive effects, reduced when co-ordination is bargaining is high Insignificant effects

Owner Occupation rate Effects on employment rate Similar effects. Only Benefits duration are insignificant

Effects on unemployment rate

Tax rate Insignificant effects. Gross replacement rate (GRR) Negative effects. The effect of GRR is larger in countries with a

high tax rate

Belot and Static Panel data on 17 EPL Insignificant. Effect of the interaction with centralisation Van Ours OECD countries over the ambiguous

(2004) period 1960-1999 Centralization Insignificant effects

Union density (UD) Insignificant effects Union density* Centralization Positive

Effects on non-employment rate Similar results

• 2. 

  Relative performance

  Effect on employment rate of middle- relative to low-skilled Static panel data over the Men Women

Kahn (2000) period 1985-1994 for 14 Co-ordination Positive Insignificant OECD countries Union density (UD) Positive Insignificant Union Coverage (UC) Positive Insignificant Effect on relative employment rate

Prime age vs. youth Prime age vs. older Men Women Men Women Tax wedge (TW) Insignificant Insignificant Negative Insignificant replacement rate year 1 Insignificant Negative Negative Insignificant replacement rate year 5 Insignificant Insignificant Negative Negative

EPL Insignificant Positive Positive Insignificant Co-ordination (CO) Positive Positive Insignificant Positive

Union density (UD) Negative Negative Positive Positive Union coverage (UC) Positive Insignificant Positive Positive Public pension replac. Rate Insignificant Positive Insignificant Positive Replac. rate older workers Insignificant Positive Positive Insignificant

Disabil. Replac. rate Insignificant Insignificant Insignificant Negative Female retirement age Insignificant Negative Negative Insignificant Male retirement age Negative Insignificant Negative Insignificant

Bertola, Blau Static Panel data on 17 Accrual rate 10 yrs age 55 Insignificant Insignificant Positive Insignificant

and Kahn OECD countries over the Effect on the relative unemployment rate

(2002) period 1960-1999. (GLS Prime age vs. young Prime age vs. older estimates)

Men Women Men Women

Tax wedge (TW) Insignificant Insignificant Negative Negative replacement rate year 1 Insignificant Insignificant Negative Insignificant replacement rate year 5 Insignificant Insignificant Negative Insignificant

EPL Negative Negative Insignificant Insignificant Co-ordination (CO) Insignificant Insignificant Insignificant Positive

Union density (UD) Positive Insignificant Insignificant Negative Union coverage (UC) Insignificant Positive Negative Insignificant Public pension replac. Rate Positive Positive Positive Positive Replac. rate older workers Insignificant Insignificant Negative Insignificant

Disabil. Replac. rate Negative Insignificant Negative Negative Female retirement age Positive Positive Insignificant Insignificant Male retirement age Insignificant Insignificant Positive Insignificant Accrual rate 10 yrs age 55 Insignificant Positive Negative Negative

Effect on the relative unemployment rate

Young Prime age Young -Prime age

Men Women Men M W

Jimeno and Tax wedge (TW) Positive Positive Positive Pos Pos Gross replacement rate (GRR) Positive Insignificant Positive Ins Neg

Rodrìiguez Static unbalanced panel data Benefits Duration (BD) Positive Insignificant Positive Ins Neg Palenzuela on 19 OECD countries Spending on ALMPs (ALMP) Insignificant Insignificant Negative Ins Neg (2003) EPL Insignificant Positive Insignificant Pos Pos

Strictness of temporary contracts Insignificant Positive Positive Ins Pos Relative Minimum wage (MW) Negative Negative Insignificant Pos Neg Co-ordination/Centralisation Negative Negative Negative Neg Pos Union density (UD) Positive Negative Insignificant Pos Neg

Union Coverage (UC) Insignificant Positive Positive Ins Pos

Taken together, these studies suggest that labour market contends that institutions are in practice largely shaped institutions can explain a significant share of crossby political interests. In particular, anything that raises country differences in labour market performance. average wages and reduces the likelihood of dismissal

Moreover, this is so even though the available indicators will benefit the typical labour market ‘insider’. 223 The

of time-varying institutions are far from perfect – in resulting institutions are liable to introduce a wedge other words, there is a degree of measurement error. The between labour supply and labour demand, interfere studies considered do not reach a complete consensus on with the reallocation of labour, compress the wage the role of each and every labour market institution. This distribution and restrict mobility. This results in lower is perhaps unsurprising given the different specifications employment rates, especially for those with high labour and methodologies employed, the scope for omitted supply elasticity, including many women, young people

variables 220 and the fact that it is difficult to take full and older workers, and quite possibly reduced overall

account of interactions between different policies and welfare. institutions. Nevertheless, as Table 2 shows, there is a

reasonable degree of consensus on some of the In practice, both views have some validity. The question

institutions that clearly make a difference. is how to design labour market institutions in order to secure the benefits, while avoiding as far as possible the

distortions that provide little benefit in terms of 3.1.2 The determinants of labour market institutions insurance or social protection. The optimal design of

The next relevant question is why labour market institutions depends on several factors, including the institutions are as they are, and whether the current state characteristics of financial markets and the frequency of affairs might be desirable despite the likely and nature of labour demand shocks, both of which

unfavourable impact of some labour market institutions influence the need for insurance.

224 In addition, different

on employment and unemployment. Here, two basic labour market institutions, and social protection more

views may be outlined. generally, may act as substitutes. The apparent trade-off between the stringency of EPL and the generosity

The ‘normative as positive’ view emphasises the role of (levels and coverage) of unemployment insurance is

labour market institutions in insuring risk-averse agents well-known, for example. 225

against income volatility, where capital markets fail to

provide adequate insurance against employment and Both views also imply that institutions cannot be income risks. 221 Institutions such as unemployment regarded as exogenous in practice. Changes in one area, insurance, employment protection legislation and wage or in external conditions, are likely to create demands compression in collective bargaining entail deadweight for changes elsewhere. Moreover, the optimal losses and information costs, but may still be welfareconfiguration of institutions will change over time. improving when markets are incomplete. In addition, the Increased competition in product markets, which aims of some policies and institutions are clearly related increases the elasticity of demand for labour, and to pure redistribution and social protection. The arguably more turbulent technological progress, for argument here is that, even though these may example, change the labour market response to existing

unavoidably reduce labour market efficiency, the institutions.

226

  Institutions that performed reasonably

benefits in terms of distribution mean that social welfare well in the past may entail large employment losses in is still greater. 222 more competitive markets.

227

These normative arguments for labour market .

institutions are important, since they show clearly why ‘flexibility’ of labour markets is not an end in itself.

However, an alternative, “purely positive”, view 223 That is, established worker, probably on a permanent

contract and well-represented by labour unions, see

220 Including on aspects known to be important in theory , such Lindbeck and Snower (1988).

224

as enforcement of benefit eligibility criteria, but on which Bertola and Koeniger (2004) show that there is a significant few data are available. correlation between EPL and borrowing constraints, which

221 The consequences of incomplete insurance markets have suggests a greater need for institutions to reduce labour been explored in the case of redistributive taxation (Varian, income fluctuations in countries where under-developed (1980)), of unemployment insurance (Acemoglu and financial systems reduce consumption smoothing Shimer (1999)), of employment protection (Bertola (2004a) opportunities. Hassler et al. (2001) argue that relatively and Bertola and Koeniger (2004)) and institutional wage immobile workers who acquire specialised skills tend to compression (Agell (2002)). With insurance arguments the prefer more generous unemployment insurance. The data benefits of insurance should be trade-off with the cost due indeed strongly suggest a negative relationship between the

to reduced efficiency. mobility rate and unemployment insurance. 225

222 Higher wages for those remaining employed and social See Buti et al. (1998); Boeri, Conde-Ruiz and Galasso

transfers for non-employed individuals have a first order (2002). 226

effect on the welfare of risk-averse workers who prefer to See Bertola, Boeri and Nicoletti (2001); Mortensen and smooth consumption inter-temporally across different states Pissarides (1999). of the world, Bertola and Koeniger (2004); Bertola (2004a). 227 See Boeri (2001); Bertola (2004b, c).

Box 3: Centralised vs. decentralised wage bargaining

Both theoretical and empirical analyses have suggested that bargaining systems which are either highly centralised at national or multi-industry level or decentralised at the level of firms perform better than intermediate systems where bargaining takes place at the level of industries (Calmfors, 1993). According to this literature, the relationship between wages and centralisation is hump-shaped, implying lower employment in intermediate bargaining systems. In practice, the key requirement is that wages should reflect productivity and local labour market conditions, and this might be achieved under different bargaining systems, depending partly on factors such as the size of the country, the extent of regional productivity disparities and whether bargaining tends to be constructive or conflictual.

The main advantage of centralised bargaining is that it allows labour representatives to take into account the negative impact that excessive wage claims would have on overall employment. Decentralised bargaining, on the other hand, means that wages are restrained by market forces and adjust better to local productivity and labour market conditions.

Evidence from OECD countries (Boeri et al. (2001)) suggests that highly coordinated, centralised systems tend to be associated with lower unemployment and, moreover, that the degree of coordination between different bargaining levels is a much more significant influence on performance than union density or coverage, i.e. the share of workers who belong to a union or are covered by collective agreements.

However, coordinated bargaining also entails greater wage compression, with negative effects on relative employment – especially at the bottom of the wage distribution (Blau and Kahn, 1996). Bargaining institutions tend to raise the relative wages of the young and less-educated, which results in lower employment, especially for men, though possibly higher employment for women, since higher relative wages encourage female labour supply (Kahn 2000). Wage compression also modifies the industry distribution of employment, shifting employment away from industries with low wages (Davis and Henrekson, 2000), and is liable to widen regional employment disparities. In contrast, decentralised bargaining allows higher relative wage flexibility and leaves wider room for bargaining on working conditions more generally. It also makes possible the introduction of performance-related pay schemes where wages are used to motivate and improve workers’ productivity.

In practice, the distinction between centralised and decentralised systems is blurred, since bargaining often takes place at two or more levels. The kind of ‘decentralisation’ in two- or three-tier systems that involves local wage increases in excess of those agreed at higher levels, is liable to discourage wage moderation (Calmfors, 1993). In the context of monetary union and large regional disparities within several EU countries, a shift from centralised towards more decentralised bargaining appears desirable.

towards groups at higher risks of inactivity or

3.1.3 Principles for the design of labour market unemployment. 228 Eligibility conditions and job-search

institutions requirements may be even more important than the level of benefits. There is evidence that the threat of losing

Growth- and stability-oriented macroeconomic policies benefits if an employment offer is not accepted tend to are an essential underpinning for an improvement in raise the incentive to find a work (Jensen, Rosholm and labour market performance. Macroeconomic stability is Svarer, 2003). More generally, a system with monitoring supported by a wage formation mechanism that sets and sanctions restores search incentives most wage growth in line with both price stability and effectively, since it brings additional incentives to search productivity developments. There remains some room actively so as to avoid the sanction, allowing for higher

for debate over whether wage bargaining should in benefits than otherwise. 229

general be centralised or decentralised (see Box 3). In the context of monetary union, and in view of large regional employment disparities in several Member

States, wages, including minimum wages, must be adaptable to local productivity and labour market conditions At the microeconomic level, an improvement in the functioning of the labour market requires pricing in workers with low labour market attachment and improving the matching between unemployment and vacancies. The role of incentives – particularly in the design of unemployment benefits and in the targeting of

active labour market policies – has been highlighted in 228 See De Koning et al. (2004); Van Ours (2003); Madsen

countries where reforms appear to have led to improved (1999). The example of the British ‘New Deal for Young labour market performance. People’ (for those unemployed for 6 months or more) is

interesting. Before being offered subsidies training or

Successful reforms have included activation measures subsidies or government-provided employment, participants

serving partly to tighten the eligibility conditions of must pas through a ‘Gateway’ period where they are unemployment benefits, combined with more intensive assigned a personal adviser. The evidence suggests that, at

active measures – including subsidies to employers, this stage, a substancial share (40 per cent) move into direct job creation and training measures – targeted unsubsidised employement (Bell et al. (1999)).

229

  See Fredriksson and Holmlund (2004).

Box 4: Reforms in successful countries

As noted in Section 3.1.1, labour market institutions alone explain a relatively low share of the variation in employment and unemployment performance. Thus, a simple story about successful countries that have pursued reforms and unsuccessful ones that have not is unlikely to be fully convincing. In order to explain a large share of the variation in employment rates, researchers usually have to include macroeconomic factors, country dummy variables (fixed effects) and/or lagged employment rates among

the explanatory variables.

This means that, over a period of around 10 years, some countries will be doing better, and some worse, than one might expect just by looking at labour market institutions. In some cases, this will be due to unobserved factors – an example here would be the relatively cooperative relationship between unions and employers in the Nordic countries, which does not show up in standard indicators. In countries where bargaining is more conflictual, a similar bargaining structure may be more problematic. Another example is where the employment rate deviates from its structural level on account of macroeconomic shocks or short-term policy

effects. As noted in Section 2.1, part of the improvement in the EU since the mid-1990s has been cyclical rather than structural.

What can be said is that, in the few countries where more or less comprehensive labour market reforms have been undertaken over the past decade or more, these appear quite clearly to have led to improved performance. Within the EU, the leading examples are:

• the United Kingdom: substantial reforms of taxes, benefits and the collective bargaining system in the 1980s, continued tax and benefit reform, reinforced by active labour market policies focused on job-search in the 1990s;

• the Netherlands: wage moderation combined with reductions in the tax burden on labour; substantial tightening of benefit systems and job-search requirements, especially for younger unemployed people; and

• Denmark: reforms of unemployment benefits and active labour market measures in the 1990s, shifting the emphasis towards job-search rather than automatic benefit entitlement, were widely seen as instrumental in the large fall in unemployment, especially among young people.

Ireland is sometimes included, though the very particular macroeconomic circumstances affecting that country over the past decade make it a difficult case to judge. Spain, despite its beginning as the worst-performing EU Member State, and the fact that the fall in unemployment in recent years is partly a statistical phenomenon, might well be added to the list. A succession of reform packages since the mid-1990s addressing employment contracts, unemployment benefits and labour taxation has coincided with a

sustained rise in the employment rate.

In recent years, several EU countries have undertaken norms and the nature of contractual arrangements, partial reforms of their labour market institutions. perhaps combined with tax reforms; a rebalancing of Liberalising temporary contracts without addressing measures designed to protect workers from labour labour market regulation for other employees is perhaps demand shocks, such as employment protection or the most notable example, and may be a risky unemployment insurance; reform of unemployment and

strategy. 230 In addition, early reforms have sometimes other benefits, focusing especially on duration and

focused on politically ‘low-hanging fruit’, such as tax eligibility criteria, coupled with enhanced enforcement cuts in return for wage moderation, expenditure-based of job-search requirements and followed by a range of active labour market policies and liberalisation of parttargeted active measures for those unable to find work

time work. Remaining reform options may therefore be during the period of benefit entitlement. concentrated in politically more difficult areas such as

benefit entitlements, wage bargaining or employment A well-functioning labour market should also be protection legislation, and therefore tougher to inclusive, reducing the risks of marginalisation and of implement. Moreover, the experience of the most long-term unemployment. The debate on how to reform successful countries suggests that far-reaching labour the European labour market has often been dominated market reforms require major shifts at both macro and by the perception of an inescapable trade-off between micro levels. Thus, both theory and experience suggest efficiency and equity. Blanchard (2004) presents this as the need for a comprehensive package, or at least a close a production possibility frontier, with efficiency on one sequence, of reforms. axis and social insurance on the other. However, several

Policy efforts within the revised European Employment Strategy and the BEPGs might include a shift in the wage-setting mechanism through a redefinition of rules,

230 The liberalisation of temporary contracts facilitates hiring

and allows employers to screen staff before offering a permanent job. But, unless stringent employment protection for permanent contracts is relaxed, this may also result in excess staff turnover among temporary workers, whose contracts cannot be freely renewed. It may also strengthen the bargaining position of those on protected permanent contracts, since there is now a buffer of temporary staff who will be the first to be fired in a downturn.

Graph 8: Developments in labour force participation, employment and unemployment EU15, 1983-2003

 Total

80 Females 70

70 60

60 50

50

Participation rate 40 Participation rate

40

Employment rate Employment rate

30 Unemployment rate 30 Unemployment rate

20 20

10 10

0 0 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003

Older workers total Youth total

45 60

40

35 50

30 40

25 Participation rate Participation rate

Employment rate

20 30 Employment rate

Unemployment rate Unemployment rate 15 20

10 10

5

0 0

1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003

Source: Commission services.

countries are characterised as being located inside the in reducing youth unemployment, it seems difficult to frontier. The idea is that the design of labour market argue that equity or social cohesion overall have been institutions is below best practice, and might be adversely affected. Employment protection legislation in improved in ways that would improve both efficiency some countries has a clear impact on distribution, but and equity, or at least improve one without not necessarily in a way that everyone would agree is compromising the other. equitable. It protects established employees on

Clearly, the management of taxes and transfers entails permanent contracts, but partly at the expense of groups administrative costs and deadweight losses as well as who are worse-off, such as the unemployed or those on risks of welfare dependency. Nevertheless, Blank (2002) temporary contracts. Moreover, protection often takes notes three situations in which the equity-efficiency the form of administrative and legal costs and delays – trade-off may be low, or there may even be i.e. deadweight costs. The impact on efficiency and complementarities between equity and efficiency. These distribution could be strengthened, for example, by are: 1) when transfers go to segments of the population reducing deadweight costs while facilitating redundancy with no capacity of changing their behaviour; 2) when payments.

benefits are paid conditional to behavioural

requirements; 3) when payments change the behaviour 3.2 Employment and participation in specific or the opportunities in such a way that increase income groups

in the future. While the first condition holds only in the

case of social policies strictu sensu, such as genuine This section provides an alternative perspective on the incapacity benefits, the others are clearly relevant for determinants of labour market performance by looking labour market policies. at the structural determinants of employment in key

demographic groups. Graph 8 provides a memorandum

Reforms of unemployment benefit systems and active of longer-term developments in labour force labour market policies that withdraw entitlement to participation, employment and unemployment among benefits for those not actively seeking employment may women, older-working age people and young people, as be perceived as inequitable by some. But in cases where well as for the working-age population overall.

such reforms appear to have been effective, particularly

3.2.1 Female employment cultural changes, has enabled women to opt increasingly

The female employment rate has increased sharply in for market employment. Improved education may also recent years and decades, driven mainly by increased increase the returns to professional experience, as more female labour force participation. This in turn is due women access higher responsibilities and more qualified partly to institutional factors, but also largely to occupations. Olivetti (2001) shows for the USA that the changing social and cultural attitudes as well as the rise increase in total hours worked by married women in female educational attainment in recent decades. between 1970 and 1990 can be explained by a rise in Indeed, participation of young women in education or returns to experience. As the opportunity cost of the labour force is now barely below that of young men, temporarily leaving the labour market increases, married while young women are now on average better educated women increasingly avoid interruptions to their

than their male counterparts (see Table 4). professional life.

The change in cultural attitudes and social norms However, women – especially married women and those regarding gender roles is clearly a major influence on with children – are still less attached to the labour female employment. Participation in the labour market market on average than men. The opportunity cost of is increasingly the norm for women of all ages. In most employment is higher when there are viable alternatives European countries, women try to plan motherhood in in the form of home production or child care. Part-time order to reconcile family and professional life. They employment has risen rapidly in recent years, and tend to postpone the first child, have children at shorter

around one third of employed women are working partintervals and have fewer children in total. 233 The change

time in the EU-15 as a whole, although this varies in cultural attitudes is reflected in differences between substantially across countries, from 7 per cent in Greece age cohorts, with married women from younger to 74 per cent in the Netherlands. generations much more likely to participate in the labour

Women who participate in the labour force also remain force.

more likely to be unemployed than active men, though Some reduction in the gender wage gap might be an the gap between female and male unemployment rates additional factor behind the rise in female participation,

has declined since the mid-1990s. 231 The share of longalthough its magnitude should not be overstated. The

term unemployment in total unemployment is also narrowing in the wage gap is not fully explained by higher for women, at 42 per cent of total female convergence in experience and education but may also

unemployment in 2002 as compared with 38 per cent for be related to the decline in gender discrimination. 234

men. Gender segregation by sector and relatively low wages

Nevertheless, the fall in overall unemployment signals in female-dominated sectors nevertheless explain a greater employment opportunity for women, and may significant proportion of the remaining gender pay

thus prompt ‘discouraged workers’ to return to the gap.

235

labour market from inactivity or long-term The tax system distorts the labour market participation unemployment. This phenomenon also explains the prodecision of married women, who are more heavily taxed

cyclical pattern of labour force participation, with a as second earners than men in many EU countries. 236

positive output gap or higher economic growth There is sound evidence that high marginal tax rates enhancing women’s prospects of finding a job. reduce labour supply and, moreover, that labour supply

Main structural determinants of female employment is more elastic for women than for men. Only in a few countries – Finland, Sweden, Luxembourg and Greece –

The increase in female education in recent decades do second earners and single individuals face similar appears to be a major determinant of the positive trend marginal tax rates. Almost all countries now have

in female labour force participation. In 2002, a woman separate taxation for married individuals, 237 but the total

with tertiary-level education was more than twice as tax burden on second earners remains significantly likely (79 per cent) as a woman with lower-secondary higher than on single individuals owing to the loss of the

232

level or below (38 per cent) to be in employment. dependent spouse allowance. This does not guarantee that further increases in average

educational attainment will lead to further increases in Likewise, child benefits reduce female participation by aggregate female employment. Nevertheless, the rise in increasing the disposable income of families with female education relative to males over recent decades is children, by up to 10-20 percent on average in some EU no doubt a factor that, along with broader social and countries. This income effect on participation may be

231 In absolute terms and, to a lesser extent, in relative terms

232 Men with tertiary-level education are also more likely to be 233 See Vlasblom and Schippers (2004).

employed (86.4 per cent) than men with lower-secondary 234 See Pissarides et al. (2003).

level or below (61.8 per cent), but less so. Of course, the 235

causality may run both ways, better employment prospects See European Commission (2003). 236

leading individuals to invest more in education, as well as See Jaumotte (2003). education leading to improved employment prospects. 237 OECD (2001).

Table 3: Part-time employment as a share of total employment

in the EU-15

1983 1993 2003

Country Males Females Males Females Males Females

BE 1.8 19.6 2.2 28.6 6.1 39.0

DK 6.1 43.7 10.1 36.9 10.8 32.1

DE 1.2 29.7 2.6 31.7 5.5 40.4

GR 3.0 11.2 2.2 7.2 1.9 7.4

ES : : 2.2 14.6 2.5 16.7

FR 2.1 19.8 3.9 26.2 5.2 29.8

IE 2.6 15.3 4.7 21.2 6.1 30.5

IT 1.9 9.1 2.2 10.9 3.0 17.3

LU 1.0 17.0 1.0 18.6 0.9 25.4

NL 6.5 49.7 14.7 64.4 21.3 73.9

AT : : : : 4.3 35.1

PT : : 3.1 10.3 4.2 14.2

FI : : : : 8.0 17.4

SE : : : : 10.0 34.9

UK 2.3 40.8 5.4 43.2 8.7 43.3

EU15 : : : : : :

Source: Eurostat.

combined with an inactivity trap effect in the case where Nevertheless, a high share of part-time work is not a child benefits are means-tested and are likely to decrease necessary condition for a high female employment rate,

if the mother enters employment. In particular, child as the example of Finland shows.

238

benefits deter women from taking up part-time work.

Some of the evidence suggests that, other things equal, Anti-discrimination policies are expected to lead to the availability of long paid parental leave may also further increases in female participation and

lower female participation by encouraging women to employment. Although there is a relatively wellwithdraw from the labour market in the short run. 239 In developed legislative framework regarding equal pay turn, withdrawal from the labour market is liable to and employment opportunities within the European

reduce wage and career prospects in the longer run. Union, there remain substantial differences in enforcement, as measured for example by the number of

Conversely, female participation may benefit from lawsuits, and public awareness of these problems. 242

measures aimed at better reconciliation of work and

family life. In particular, childcare subsidies cut the The sectoral shift from manufacturing and agriculture to relative price of childcare, increasing the relative return services, which tend to be more female employmentof market work. The empirical evidence 240 indicates that intensive, is perhaps the main demand-related factor childcare subsidies raise female labour supply and that behind the rise in female employment. As women are the employment rate of married women is higher in disproportionately employed in smaller establishments countries providing for subsidised childcare. Maternity in service sub-sectors such as retailing, catering and leave or short paid parental leave helps women to personal services, product market regulations affecting reconcile working and family lives by reinforcing their the creation and expansion of such enterprises may be attachment to the labour market while allowing them to particularly relevant for female employment.

243

take care of newborn children. However, there is still some evidence of segmentation

Part-time work also appears to facilitate female labour by gender, especially in the southern Member States, force participation. Labour force surveys indicate that where women are over-represented in involuntary partfamily responsibilities are one of the main reasons for time, temporary or casual jobs. Since these jobs tend to working part-time and that only 14 per cent of female offer relatively poor pay, working conditions and part-time employees are seeking a full-time job. Partprospects, there is a risk that many women’s skills are time work is clearly associated with higher female under-utilised.

241

participation and higher employment in persons. As regards institutional influences on the demand side, Bertola, Blau and Kahn (2002) find that centralised

238 See Buddelmeyer et al. (2004). wage-bargaining together with a high degree of

239 See Jaumotte (2003).

240 See Jaumotte (2003). 242 See Pissarides et al. (2003).

241 See Jaumotte, 2003; Garibaldi and Mauro, 2002. 243 See Pissarides et al. (2003).

unionisation lowers the female employment rate, while from pensions, (v) actuarial fairness, or the incentive

preserving a high employment rate for prime-age men. structure, of pension systems, and (vi) social norms. 247

The idea is that unions purposely negotiate large wage

premiums for those whose opportunity cost of A broad range of theoretical and empirical literature employment is high, which results in wage compression, concludes that incentives embedded in tax and benefit

and increased female inactivity and unemployment. systems are a critical influence.

248

  Old-age and early retirement pension systems are the most important

3.2.2 Older working-age people benefit types in this respect, but unemployment insurance and disability benefits often serve as

The low employment rate for 55-64 year-olds is mainly substitutes. In addition, the interaction of benefits with

due to early exit from the labour force. 244 The the taxation of earnings is an important influence.

unemployment rate for this age group is less than half

the rate for the working-age population as a whole. More precisely, the key determinants of early retirement There is a broad agreement in the economic literature appear to be: (i) the first age at which the benefits are that the decline in participation and employment of older available; (ii) the generosity of benefit systems; and (iii) working-age people over recent decades has been the implicit tax rate imposed on continued employment

largely due to generous early retirement pathways. Such once an individual is eligible for benefits

249

. The

schemes were offered to redundant or unemployed availability of benefits seems to largely eclipse elderly workers in response to severe shocks in the incentives to postpone the take-up of benefits. An labour market. They comprised a wide array of early implicit tax on continued employment arises when the retirement and other benefit schemes such as cost of working one additional year in terms of foregone unemployment pensions or prolonged unemployment benefits and additional contributions paid is not offset benefits, special contracted pensions to redundant by higher future benefits. This represents a clear

workers, disability pensions awarded on the basis of distortion of the labour supply decision.

labour market considerations, benefits provided in In addition, OECD (2003a) finds that the prevalence of exchange for the employment of young workers. unemployment has an impact on the labour supply of

In addition, many reforms of old-age pension systems older workers, probably through the ‘discouraged

have contributed to disincentives for continued work, for worker’ effect.

instance, by lowering the standard retirement age, Education is often supposed to play a role in the increasing replacement rates of pensions, increasing employment and participation decisions of older pension contribution rates, and reducing the pension workers, less-educated people being prone to early benefit in case of early retirement by less than what withdrawal. As in the working-age population in

would be actuarially justified. 245 general, those with a high level of education are much

Main structural determinants of employment of older more likely to be employed than those with a medium or

working-age people low level. However, it is unclear whether changes in the average level of education over time influence older

In theory, rational individuals would withdraw from the working-age people’s employment. In countries where labour market at the age which maximises their expected mass upper-secondary and tertiary education spread welfare over the life cycle, given preferences for earliest, and where older workers are therefore almost as consumption and leisure. However, actual retirement well educated as the rest of the labour force (see decisions do not conform to the basic life cycle model, Table 4), the employment rate of older workers is not for several reasons. First, capital market imperfections necessarily high (see Table 1, third column). may prevent people from borrowing to finance Nevertheless, formal years of schooling provide at best retirement before they become eligible for benefits. an imperfect proxy for functional skills and adaptability, Secondly, pension systems often strongly discourage or which are likely to be among the true determinants of

even disallow work after the statutory retirement age. older workers’ employability. Thirdly, retirement at the customary age may be a strong

social norm. Fourth, people are usually not fully aware On the demand side, older workers are affected by of incentives to advance or defer retirement, which tends temporary demand shocks like any other workers. to result in their retiring as soon as benefits become However, they may be less adaptable than younger

available. 246 colleagues, in part because of lower education levels and/or obsolete skills, and therefore more difficult to

The retirement decision is influenced by many factors, retain in employment. This is of particular relevance in including: (i) health, (ii) mandatory retirement rules, (iii) times of rapid technological change. Employers may be rules on continued earnings, (iv) other benefits apart reluctant to provide training for older workers, both for

244 See European Commission (2004b). 247 See Casey (1998).

245 See e.g. OECD (1998) and (2000), Blöndal and Scarpetta 248 See Pestiau (2001); OECD (2003a).

(1998) and (1999), Conde-Ruiz and Galasso (2002). 249 See Gruber and Wise (1999); Blöndal and Scarpetta (1998), 246 OECD (2003a). (1999); OECD (2003a).

the above reasons and simply because the period over laws or are vulnerable to ‘last in, first out’ provisions. 251

which the investment can be amortised is relatively Thirdly, young people’s labour supply tends to be short. elastic: many are still searching for and choosing the job

These issues might be less problematic if wages closely that suits them best; the opportunity cost of working is reflected productivity and performance. However, wages high for those with parental support but without families tend to increase with seniority and are rigid, at least of their own; and publicly funded education provides a

downwards, even in the event of declining productivity, viable alternative to employment.

which reduces the demand for older workers’ labour. The situation improved over the mid-1990s, following a Moreover, to the extent that a monotonically increasing severe shock to youth employment during the recession age-earnings profile is seen as the norm, older workers of the early 1990s. Since 2000, the youth employment may be discouraged from supplying their labour under and unemployment rates in the EU-15 have remained different terms and conditions. broadly constant at around 40-41 per cent and

There is limited evidence that participation in training 15 per cent respectively.

may, in a sense, substitute for wage flexibility. Bassanini Main structural determinants of employment of young

(2004) finds low returns to training in the form of higher people

wages for older workers, but some evidence of returns in

the form of increased job security. One possible The population aged 15-24 has decreased in the EU by interpretation is that, other things equal, the productivity over 6 per cent between 1995 and 2002, from 48 to 45 of older workers fails to keep up with the growth in million, while total population has grown over the same wages due to seniority pay, thus reducing the likelihood period by over 3 per cent, from almost 363 million to of continued employment. Training, however, may almost 375 million. To the extent that youth arrest this decline in older workers’ unit labour costs, unemployment is partly due to over-supply of relatively which would raise the probability of continued low-skilled, inexperienced workers, the fall in the youth

employment, but without leading to wage increases. population, given a broadly constant participation rate, would be expected to result in lower youth

Employment protection legislation favours established unemployment. 252

workers by definition and is therefore likely to delay

redundancies of some older workers, with a positive The share of young people in education has increased impact on their employment relative to other groups. considerably in most EU countries over recent decades, However, it does not prevent redundancies or early which is reflected in the decline in average years of

retirement in the event of severe demand shocks or sharp schooling beyond age 25, as shown in Table 4.

253 This

declines in an individual’s productivity, and is likely to explains part of the decline in labour force participation make it more difficult for older working-age people and employment. It may also be a negative influence on seeking re-employment, at least on similar terms and the youth unemployment rate, since it is the relatively

conditions to established staff. low-skilled who tend to enter the labour force rather than pursuing further education. On the other hand, the 3.2.3 Young people increase in educational attainment may have a positive longer-term effect on overall employment (see Box 5).

The situation of young people in European labour markets appears to have steadily worsened over the past 30 years, with falling participation and rising unemployment. Moreover, earnings are not only much lower than those of prime-age workers, but have been

250

falling gradually in relative terms since the 1970s. These developments may be partly explained by increased enrolment in education since, as enrolment in education rises, labour market participation falls directly, while those young people who remain in the labour market tend to be the lowest-skilled. Nevertheless, with youth unemployment around or over 20 per cent in several EU countries, the picture is still bleak.

Employment of young people is also relatively volatile, for several reasons. First, almost 37 per cent of

employees aged 15-24 are on temporary contracts, 251 See also Pissarides (1986), O’Higgins (1995). compared to 9.5 per cent of 25-64 year-olds. Secondly, 252 See Korenman and Neumark (1997).

many others do not benefit from employment protection 253 15-24 year-olds have fewer average years of completed

schooling because many are still studying, and years are only counted once the level of education in question has

250 See Blanchflower and Freeman (1999). been successfully completed.

Table 4 : Years of schooling by age group and gender, 2002

15-24 25-34 35-44 45-54 55-64 25-64

F M F M F M F M F M F M

BE 10.7 10.3 12.4 11.9 11.4 11.5 10.6 10.8 9.4 10.0 11.0 11.1

DK 10.3 10.3 12.8 12.6 12.6 12.5 12.4 12.6 11.9 12.3 12.4 12.5

DE 10.0 9.8 12.9 13.1 12.9 13.2 12.8 13.3 12.2 13.1 12.7 13.2

EL 10.7 10.3 12.0 11.5 10.7 11.1 9.2 10.0 7.9 8.7 10.0 10.4

ES 10.2 9.6 11.1 10.6 9.7 9.7 7.9 8.5 6.4 7.3 9.1 9.3

FR 10.7 10.5 12.0 11.8 11.1 11.2 9.8 10.3 8.5 9.2 10.5 10.8

IE 10.6 10.2 11.9 11.6 11.1 10.9 10.1 9.9 9.0 8.8 10.8 10.5

IT 10.1 9.8 11.2 10.9 10.4 10.4 8.9 9.6 6.9 8.0 9.5 9.9

NL 10.5 10.1 12.5 12.3 12.0 12.3 11.3 12.0 10.6 11.5 11.7 12.1

AT 11.2 11.0 12.6 12.8 12.4 12.9 11.9 12.6 11.4 12.3 12.1 12.7

PT 8.8 8.0 9.0 8.2 7.6 7.3 6.7 6.9 4.5 6.1 7.1 7.2

FI 10.3 10.0 12.7 12.1 12.4 12.0 11.4 11.0 9.9 9.9 11.6 11.3

SE 10.8 10.7 12.4 12.3 12.2 11.9 11.9 11.5 10.9 10.6 11.9 11.6

UK 11.8 11.7 12.3 12.4 12.1 12.3 11.7 12.1 11.2 11.6 11.9 12.1

EU-15 10.6 10.3 12.0 11.9 11.5 11.6 10.5 11.0 9.3 10.2 10.9 11.3

EU-15

(both sexes) 10.5 11.9 11.5 10.8 9.8 11.1

Note: Estimated from Labour Force Survey data. The (self-reported) highest level of schooling attained is multiplied by the standard number of years required to reach that level; see the EU Economy 2003 Review Chapter 3 for further details.

Source: Commission services.

Box 5: Education and employment

For those of working age, education and labour force participation are strong substitutes, especially in countries where part-time work is hard to come by or where students traditionally live with parents during studies. Nevertheless, one may be both employed and in full-time education, since paid work of one hour or more per week counts as employment in the Labour Force Survey.

Rising enrolment in education is thus one reason for the long-term decline in labour force participation of young people. The EU E CONOMY 2003 R EVIEW (Chapter 3) estimates that, with a further rapid expansion in upper-secondary and tertiary enrolment over the next decade or so, reduced participation might lower the overall (i.e. 15-64) employment rate by up to 0.9 percentage points. This is not to be lamented, since the evidence suggests that education is a key driver of economic growth, so that reduced employment today will be compensated by more productive employment in future.

An important question is whether rising educational attainment also has a positive employment impact. Individuals with tertiary education are much more likely to be employed (82.8 per cent) than those with only lower-secondary education or less (49.4 per cent). But it does not necessarily follow that an increase in the average level of education leads to an increase in aggregate employment. This is because it is arguably one’s level of education and skills relative to others that influences labour supply and demand. In any event, it is striking that neither in the literature on the determinants of aggregate employment and unemployment (see Table 2) nor in studies of the impact of education on economic growth is there any solid evidence that education influences the aggregate employment rate. That may be partly because researchers have not had access to data – including data on functional competences, as opposed simply to years of schooling – that would allow them to investigate the links between education and employment more thoroughly.

It seems likely that education, along with broader social and cultural shifts, has been a factor behind rising female employment (see Section 3.2.1), and quite possible that training may help older people to remain longer in the workforce. For younger people, any aggregate employment effect is likely to depend on how increased attainment is distributed. It is far from obvious that sending more young people to university, for example, will raise aggregate employment. On the other hand, encouraging more people to complete upper-secondary education would have the effect of evening out the distribution of skills. This, given the impact of labour market institutions on the demand for young people’s labour, seems more likely to yield aggregate employment gains.

in most countries over the period 1985-1994, the changing industrial structure has a positive impact on

youth employment.

One reason why the labour market is tough for young

people is that they tend to have relatively low skilllevels

and, by definition, not much work experience.

While, as noted, educational attainment has risen rapidly

Over 22 per cent of young workers are in part-time jobs, in recent decades, the consensus view is that the demand

compared to 15.5 per cent of prime-age workers. There for highly educated workers has risen even faster. This

is a significant gender difference, though less has been good news for young people who complete

pronounced than in the case of prime-age workers: higher education, but leaves those who drop out early

16.3 per cent of young men work part-time, compared to from education in a difficult situation.

29 per cent of young women. Of course, many people

who combine studies with employment choose to work The wage-setting mechanism has been highlighted as part-time. Thus, the availability of part-time work is one key institutional factor affecting youth likely to be an important influence on youth labour unemployment. As noted, the labour supply of young

supply in persons. people is relatively elastic, which makes them vulnerable to bargained wage structures, by the same

A combination of unemployment benefit reforms and argument as in Section 3.2.1 on female employment. active labour market policies has had a clear impact on

effective labour supply among young people in recent Minimum wages might be expected to have a similar years. As noted in Section 3.1.3, stricter enforcement of impact, although theoretical and empirical studies are far benefit eligibility criteria in combination with from reaching conclusive agreement on this. Nobody personalised jobsearch assistance appears to have helped doubts that a very high and undifferentiated minimum many young people at risk of becoming long-term wage would have a negative impact on youth

unemployed to find unsubsidised employment. employment. But some argue that market-determined wages may be too low in some circumstances, especially

The evidence is less clear on whether more intensive where employers have monopsony power and workers active interventions – such as subsidised jobs and are in a weak bargaining position. In this case, a training programmes – have improved the employability minimum wage might even raise employment by of beneficiaries. People are not counted as unemployed encouraging increased labour supply; or it could at least while they participate in such programmes, so in this improve job quality without reducing employment. sense there is a statistical improvement. But evaluation Empirical studies do not give a clear indication of the results have been mixed. Some studies find that active direction of interactions between minimum wages and

programmes do nothing to raise – and in some cases youth unemployment. 255 This may be partly because

even reduce – the probability of unsubsidised minimum wages are frequently set at lower levels for employment, or that any enhancements in employability young people.

are prohibitively costly. But some programmes, where

well-designed and carefully targeted to the individual Employment protection legislation (EPL) clearly needs of participants, appear to have been more benefits prime-age, established workers at the expense

successful. of young people. Firms are more reluctant to fire established workers if this involves severance payments,

Youth unemployment is generally much higher than notice periods and costly procedures. They are also overall unemployment, but closely correlated with it. reluctant to take on new workers on standard permanent According to various studies, total unemployment is – contracts, since they must take into account the together with the relative size of youth population, possibility of having to pay firing costs in future. Young labour market institutions and macroeconomic shocks – people may be employed on apprenticeship contracts or the main factor explaining differences in youth temporary contracts, which involve lower firing costs,

254

unemployment. but more often than not these are not converted into

The sectoral shift from agriculture and industry to standard contracts. Empirical studies confirm that more services, and the expansion of many low-wage service stringent EPL is associated with higher youth industries that traditionally employ many youths, has unemployment relative to prime-age unemployment.

256

arguably been a positive factor for youth employment.

Many young people find their first jobs in retail trade or 3.2.4. Migrants

hotels and restaurants. Young males are often employed Third-country nationals are a small but increasingly in construction and young females in health care. significant group as far as EU employment rates are Blanchflower and Freeman (2000) find empirically that, concerned. In the EU-15, they accounted for 3.6 per cent

255 See Ghellab (1998).

254

  See Gaude (1997); Jimeno and Rodriguez-Palenzuela 256 See Bertola, Blau and Kahn (2002); Jimeno and

(2003). Rodríguez-Palenzuela (2003).

of total employment in 2002, but as much as 22 per cent Several empirical studies, mainly for the US, indicate of employment growth between 1997 and 2002. Spain, that the marital status and presence of children also play Italy and Ireland have seen particularly large increases, a role in explaining the difference in participation rates albeit beginning from low levels. Migrants make a among immigrant and native women; the foreign-born significant contribution to the labour force in several who had not graduated from high school were more Member States: Luxembourg (43.2 per cent of the likely to be married or to have children than their native labour force), Austria (9.9 per cent), Germany (8.9 per counterparts, which in turn reduced their likelihood of

cent), Belgium (8.2 per cent) and France (6.2 per cent). participating in the labour force. 260

In most countries, migrants of working age, especially A large body of literature for the US supports the women, are less likely than natives to participate in the hypothesis of economic assimilation, whereby wages or

labour force. 257 In a few Member States, namely Spain, employment prospects of migrants improve with time

Italy, Greece, Luxembourg and Austria, the participation spent in the host country. Empirical studies typically rate of foreigners is similar to or higher than that of find that the earnings gap between immigrants and natives. Foreign workers are more likely to be natives falls over time. The interpretation provided is unemployed than natives in all Member States. that, in the absence of any form of discrimination, wages

Disparities in labour market outcomes exist both reflect individual productivities, and only part of the between migrants and natives and among migrants human capital acquired in the country of origin can be themselves. Migrants are a very heterogeneous group transferred to the destination. Migrants who have lived according to their age, gender, skill level, country of longer in the country have had more time to adapt and

origin, reasons for immigration and timing thereof. Not learn the language and other country-specific skills.

261

surprisingly, women, young adults, older workers and Empirical evidence for the EU is more limited. Results those with lower skills find themselves in the worst for the UK indicate that, among non-white foreign-born position, even more so than the same groups in the men, a significant share of the initial disadvantage

native population. Empirical evidence available for diminishes with time and labour market experience. 262

some countries indicates that humanitarian migrants Employment rates rise sharply in the five years after tend to have worse labour market outcomes than other arrival and more slowly afterwards. The effects on migrants, and that the disadvantage of migrants relative unemployment probabilities are even more marked. to natives tends to be reduced with the time spent in the There is ample empirical evidence on the importance of host country. host-country language skills, notably on employment

Main structural determinants of employment of migrants probabilities.

263

The low level of education and skills – including The data available for Belgium, Germany, France, language and other host-country skills – among many Luxembourg, the Netherlands and the UK suggest that migrants is a major determinant of low labour force the skill level of foreigners tends to be higher on average

258

participation and high unemployment. In most for recently arrived migrants than for those who arrived Member States, over 40 per cent of foreigners aged 25 to a few years ago. Evidence for the USA and the UK 64 have no secondary education. Differences in the indicate an improvement in the educational attainment labour market performance of migrants across Member of migrants’ children relative to natives. Card (2004) States may also be linked to their education level, since finds above-average levels of educational attainment for the distribution of the foreign population by education immigrants’ children in the USA, even for children born level varies across receiving countries. The proportion to parents who had much lower educational attainment of foreigners with tertiary level education attainment is than native parents. Hatton et al. (1998) identify higher relatively high in Denmark, Luxembourg, Portugal, participation rates in full-time education among ethnic

Sweden and Spain. minorities than among whites in the UK, while their labour market participation rates are slightly lower and

Formal education is only part of the picture. Most of the their unemployment rates are significantly higher. difference in unemployment rates between foreigners

and natives is explained by the quality of initial training, Country-of-origin differences can have a strong impact professional experience, the transferability of skills, on the labour market performance of migrants. Nationals language skills and problems relating to from other EU Member States or the USA have an

discrimination. 259 average participation rate similar to natives in the host country, but migrants from, for example, Turkey,

Morocco, Sub-Saharan Africa or ex-Yugoslavia have lower participation rates than natives. The same applies

257 Nevertheless, because immigrants are over-represented in working-age cohorts, the average immigrant is still more 260 likely to be in the labour force than the average native, See Mosisa (2002). except in Sweden, Finland, Denmark and the Netherlands. 261 See Chiswick (1978).

258 See Bauer et al. (2003). 262 See Hatton et al. (1998).

259 OECD (2003b). 263 See Dustmann and Fabbri (2000).

to Mexicans in the US and it often reflects the economic quarter more than wage-earners. This is consistent with situation prevailing in the country of origin. However, the view that it is difficult for immigrants to find the labour market participation of foreigners also differs employment at going wages. Finally, a relatively high across host countries, depending partly on host country tax wedge on labour incomes tends to make it more characteristics and partly on when immigrants arrived. expensive to employ low-skilled workers, including

For example the participation rate of Moroccans is more migrants, in the provision of household services. than 73 per cent in Spain and less than 40 per cent in Belgium and nationals from former-Yugoslavia have lower participation rates than natives in Sweden and the

UK and higher rates than natives in Austria. 4. Labour market reform in the EU: priorities and progress

Observed wage differentials between immigrants and natives can be explained by the quality of education and

training received abroad, language acquisition, 4.1 What is required to meet the Lisbon targets?

discrimination and also unobserved individual Even without further policy measures, the employment characteristics. Empirical studies in the US indicate that, rate in the EU is expected to continue increasing over once education is controlled for, the wage gap falls to the next decade and beyond for two main reasons. First,

264

under 10 per cent for both males and females. younger women are much more likely to be employed Nevertheless, the possible role of discrimination remains than older women, owing to social and cultural changes, significant. UK evidence suggests that the wage rising educational attainment and the effects of previous disadvantage of foreign-born people relative to natives reforms, such as the liberalisation of part-time

seems to be smaller than that in access to jobs. 265 employment. This gives rise to a cohort effect which,

Labour market institutions are also likely to play a role according to Burniaux, Duval and Jaumotte (2004), in explaining the relatively low employment rates of would mechanically increase female participation from many immigrant groups, although there is insufficient 59.9 per cent in 2000 to 63.6 per cent in 2010. Secondly, evidence to draw robust conclusions. Migrants perform the early retirement tide has turned, with most relatively well in the labour market in Spain and Greece, governments having embarked on reforms to encourage where low-skilled jobs are relatively abundant. The later and more flexible retirement and to support the employment of migrant and foreign workers is employability of older workers. It will take some time concentrated in certain sectors employing a relatively for the full impact of these reforms on the employment

high share of unskilled workers such as agriculture, of older working-age people to materialise.

construction, hotels and restaurants, care for the elderly However, without further structural reforms, the EU is and other household services. In the past, many likely to miss the Lisbon target of a 70 per cent overall foreigners were employed in industry. In the last few employment rate by a considerable distance. If female years however, employment of foreigners in the services participation rises to 63.6 per cent, the female sector has gained importance, partly due to the unemployment rate would still need to be halved, from characteristics of new foreigner inflows. Foreigners who 8 per cent to 4 per cent, in order to reach a 60 per cent have arrived in the past five years are generally underfemale employment rate, which looks difficult though represented in sectors such as mining, manufacturing, not impossible. The target for older working-age people energy and construction, while they tend to be overlooks challenging even if further reforms are represented in the services sector: education, health and implemented. OECD (2003a) simulates the impact of other community services, household and other services. the following additional measures: (i) a removal of early

Institutions such as unemployment and other benefits or retirement schemes; (ii) a move towards actuarial minimum wages can be expected to affect migrants neutrality of old-age pension schemes; and (iii) a disproportionately, given their lower earnings potential convergence of standard retirement ages to 67. With on average, for reasons explained in previous subthese reforms and under the assumption of a high sections. Institutions leading to a high degree of wage elasticity of labour supply, a halving of the compression may also act as a disincentive for the unemployment rate for older workers would still be highly skilled, which may partly explain the required to reach the 50 per cent employment (see attractiveness of some Member States for highly skilled Table 6). Moreover, the labour force participation rate of migrants. Evidence from Denmark and the UK on self 15-24 year-olds is likely to continue to decline in view employment 266 provides partial support for this view. In of stated policy objectives at both EU and national levels Denmark, self-employment is much higher among for increased investment in human resources, in some migrant groups and they earn lower wages than wagecases involving targets for increased enrolment in higher

earners. Yet self-employed natives earn on average oneeducation.

264 See Card (2004).

265 See Hatton et al. (1998).

266 See Roseveare and Jorgensen (2004).

Table 5: Employment rate targets under alternative scenarios for participation and unemployment

Total Employment Participation rate Unemployment rate (PR) rate (UR)

2002 64.4 69.6 7.50

2010-OECD baseline scenario for 66.1 71.5 7.50 UR in 2002

PR

2010-OECD baseline scenario for

PR 70.1 71.5 2.00 UR to reach the target

2010-OECD high-case for PR 69.0 74.6 7.50 UR in 2002

2010-OECD high-case for PR 70.1 74.6 6.00 UR to reach the target

Females 2002 55.6 60.8 8.50

2010-OECD baseline scenario for 58.2 63.6 8.50 UR in 2002

PR

2010-OECD baseline scenario for 60.4 63.6 5.00 UR to reach the target

PR

61.8 67.5 8.50 UR in 2002 2010-OECD high-case for PR

2010-OECD high-case for PR .. .. .. UR to reach the target

Older workers (55-64)

2002 39.9 42.4 6.00

2010-OECD baseline scenario for PR 43.5 46.3 6.00 UR in 2002

2010-OECD baseline scenario for PR .. 46.3 .. UR to reach the target

2010-OECD high-case for PR 48.8 51.9 6.00 UR in 2002

2010-OECD high-case for PR 50.3 51.9 3.00 UR to reach the target

Box 6: Main additional measures to raise employment in specific groups highlighted in Section 3.2

Women: reforms of wage bargaining systems; tax reforms promoting neutral treatment for second earners; affordable child care facilities in preference to higher child benefits; more flexible labour market regulations, especially as regards part-time employment; enhanced enforcement of anti-discrimination legislation; and product market reforms enabling the expansion of sectors where female employment is concentrated.

Older working-age people: removal of early retirement schemes; actuarially reduced benefits or additional contributions in the case of early retirement; proper implementation of eligibility conditions for disability pensions, extended unemployment benefits or unemployment pensions; higher pension accrual rates for people continuing to work beyond a certain age; flexible retirement arrangements such as removing the statutory retirement age and allowing more flexible combinations of part-time work and semiretirement; reforms of wage bargaining, including a more flexible relationship between earnings and seniority; greater participation of older workers in training and lifelong learning.

Young people: reforms of wage bargaining systems; reforms of labour market regulation, in particular redressing the balance between established workers and first-time jobseekers; continued progress on benefit reforms and well-targeted active labour market policies; investments in education and training, especially measures targeted at those with low attainment. Facilitating part-time employment could help more young people to combine education and employment.

Migrants: Measures to ease assimilation, such as language training and validation of existing qualifications, and improved enforcement of anti-discrimination would help. Reforms of labour market institutions as discussed in section 3.1.3 could help in two ways: first, by improving access to low-skilled employment and, secondly, by allowing labour markets the flexibility to attract highly skilled immigrants, particularly in areas of apparent skill shortages.

Broadly speaking, these results are consistent with the promoted in order to favour female employment. An overall findings of Section 2. In particular, the area where there is clear scope for genuine trade-offs is importance of reforms of tax and benefit systems, wage investment in education. It would not be a good idea, for bargaining and early retirement incentives is confirmed. example, to discourage young people from participating Some points emerge more clearly when looking at in further education in order to raise the employment specific groups – two examples are the importance of rate. This might well have long-term consequences for part-time work for the participation of young people and productivity growth, since learning begets further

women, and the need for vigorous implementation of learning. 268 Nor would it be a good idea to promote

anti-discrimination laws in the case of women and inefficient investment in human capital, however, since migrants. the resources could be used more productively

The potential role of education and training also appears elsewhere – in investment in R&D, for example.

in a more positive light. Rising educational attainment

has been an important influence on female employment, 4.2 Priorities for and progress with labour

which raises the question of whether lifelong learning market reforms in the EU-15 could not play a similar role with regard to older Clearly, an improvement in the performance of the EU- working-age people. The role of well-targeted training 15 as a whole depends mainly on an improvement in the measures in facilitating the entry of young people and group of poorly performing countries. But, as stressed in migrants into the labour market is also highlighted. Section 3.1, the precise measures required will vary Increasing the share of young people who successfully from country to country. Thus, it would be difficult to complete upper-secondary education – which is an prescribe a detailed reform package for the EU as a explicit objective of the Lisbon strategy – might be whole.

expected to have a positive long-term employment

impact. However, one should not necessarily expect that Nevertheless, it is possible to identify priorities for the sending more young people into tertiary education, EU as a whole in the following, more restricted, sense. which is where much of the additional investment in First, it is relatively simple to identify which Member human resources seems likely to be concentrated, will States have the greatest potential contribution to make to have a significant impact on aggregate employment, raising the EU-15 employment rate – those with larger though it will raise productivity. working-age populations and/or lower employment

rates. In the light of the results of Section 2.2, one may ask

what the implications of measures to raise employment Secondly, the large body of cross-country evidence among women, older working-age people, young people reviewed in Section 3 and especially in Table 2 is a and migrants might be for productivity growth. There valuable guide to which policy areas are likely to be are clear examples of possible short-run ‘trade-offs’, in most significant in determining employment in a that people with low attachment to the labour market heterogeneous group of countries with different and below-average productivity are likely to be overinstitutional configurations. This literature is less useful represented in net employment growth. Examples may when it comes to designing detailed reforms in include low-skilled migrants, unemployed young people, individual countries. One can always point to women entering the labour market after a long period of imperfections in the data and indicators used, as well as absence and possibly some older workers whose apparent inconsistencies between the results of different productivity may be declining, particularly if they have studies. For example, as regards wage bargaining, it is changed occupations. 267 In any event, Section 2.2 not clear whether union density, coverage of wage suggests that the impact is in general temporary, with no bargains or the degree of coordination and centralisation significant implications for longer-term productivity are the key factors. The answer may be some growth, and so this should not be regarded as a genuine combination of these, together with other factors – such trade-off. as whether industrial relations are more or less

confrontational or constructive – that are very difficult

The question remains whether there are any exceptions to take account of in simple indicators. Nevertheless, to the general rule. Among the measures discussed in despite these imperfections, the empirical literature this section, there are few candidates. One might be if a consistently identifies wage bargaining as a key

particular economic sector with low productivity growth determinant of labour market performance.

potential, for example in the service sector, was In Table 2, incentives in tax and benefit systems are also consistently identified as being a critical influence on

267 On the other hand, there are also highly skilled migrants, employment and unemployment. Incentives to retire

and younger women are now better-educated than men on average, while older workers have, of course, the benefit of

experience. 268 See Heckman (2000).

early are an important special case of this, as stressed in Table 7 summarises the main priorities established since 3.2.2. The evidence is more mixed on other areas such Lisbon, and also progress made against these, as as employment protection legislation and active labour assessed in the Commission’s reports on the

market policies. Some areas of work organisation, such implementation of the BEPGs. as the availability of part-time contracts, are clearly

relevant, but the evidence on other elements – working Graph 6 combines Tables 2 and 7 to provide a broad,

time arrangements, for instance – is lacking. graphical illustration of progress compared to priorities for the EU as a whole. 271 On the horizontal axis,

Some policy areas – such as education and training or Member States are ordered by their approximate labour mobility – are hardly covered in the crosspotential contribution to raising the EU-15 employment country literature on the impact of labour market rate. On the vertical axis, main policy areas are listed institutions on aggregate employment and very tentatively in order of their known potential unemployment. This may be partly due to the lack of contribution to raising aggregate employment over the suitable indicators. In the case of education and training, next decade, as discussed above. Each cell is shaded the expectation that the major impact will be on according to Table 7. Panel A shows priorities. It makes productivity as opposed to aggregate employment may no assertion about precisely what kinds of reforms are also play a role. Nevertheless, lack of evidence should required; it merely identifies the policy areas and not be confused with insignificance; it may be that the countries where appropriate measures would make an importance of these areas simply remains to be proven. important contribution to attaining the Lisbon objectives

Priorities thus identified are broadly consistent with the for the EU as a whole. The clear concentration of four key challenges identified by the European shading towards the upper-left hand corner suggests that

269

Employment Task Force chaired by Wim Kok. These the BEPGs indeed largely focus on areas likely to have

are increasing adaptability of workers and enterprises; the largest impact in raising the EU employment rate.

272

attracting more people to the labour market and making If EU Member States were doing enough, collectively, work a real option for all; investing more and more to hit the Lisbon employment targets, then one would effectively in human capital, for the benefit of expect to see a similar pattern in Panel B, which productivity as well as employment; and ensuring illustrates progress. There is a substantial amount of effective implementation of reforms through better shading, including in some of the key areas, such as tax governance. and benefit systems and early retirement. However,

Thirdly, we may look directly at country-specific many blanks remain, often in country-specific priority priorities for labour market policies, of the kind areas, i.e. the ones shaded in Panel A. Most notably, established in the detailed country-specific analysis almost nothing has been done in the key area of wage underlying the EU’s Broad Economic Policy Guidelines bargaining. Furthermore, a closer reading of the (BEPGs) and European Employment Strategy. The Implementation Reports on the BEPGs reveals that, country-specific recommendations in the BEPGs and the where progress has been made, it has often been of a Employment Recommendations take into account the piecemeal nature, including in the key area of tax and institutional configuration and any national benefit reforms. Thus, while the strategy may be sound, particularities in each Member State. For example, the serious doubts remain over whether its implementation

cross-country evidence suggests that generous is sufficiently timely and comprehensive.

unemployment benefits may be problematic. In Italy, however, coverage of unemployment insurance is very low, and the recommendation is to increase the resources available in order to widen coverage. At the same time, while employment protection legislation (EPL) is not necessarily a problem per se, the rigid systems in Germany and Italy for medium-sized and large enterprises are an issue. On the other hand, EPL is not judged to be particularly problematic for the Netherlands, even though, according to the indicator used in most cross-country studies, EPL is also stringent

270

in the Netherlands.

However, employers do not report particular problems in

obtaining approval where necessary.

269 Jobs, Jobs, Jobs: Creating more employment in Europe’, 271 It is no substitute for a careful reading of the BEPG

Report of the Employment Taskforce chaired by Wim Kok, Implementation Reports when it comes to assessing

November 2003. progress in individual Member States.

270 The OECD indicator of EPL stringency for the Netherlands 272 Of course, this partly reflects the fact that countries with the

is high largely because formal approval by the courts or greatest problems receive the most recommendations, and

public employment services is still required for dismissals. four out of the six problem countries are large.

Table 6: Main priorities and progress on labour market reforms, 2000-03

Priority Policy Areas Progress made (as indicated in (as indicated in BEPGs) Implementation Reports)

-tackle distortions to work incentives in the tax-benefit -changes in housing market and efforts to overcome system linguistic barriers

BE -reduce early withdrawal from labour force -work-related tax credits reducing marginal tax rates at -wage setting to take account of productivity and skill lower levels; streamlining of tax incentives for

differences recruitment of specific groups (female and older workers) -relax restrictions on fixed-term and temporary contracts -steps towards alignment retirement age for men and -address obstacles to labour mobility women -improve efficiency of ALMPs -improved incentives in the benefit systems; steps to -reform of tax system and benefit eligibility; reduce tighten eligibility and reduce marginal tax rates

DK marginal tax rates on low wages -more flexibility of working time -postpone retirement through improved incentives -training measures to address foreseen shortages of skilled

-increase labour supply by integrating immigrants and by workers channelling students faster through education -more effective and efficient ALMP spending

-simplify benefit administration; improve tax-benefit -some progress towards improved incentives in the incentives especially for older workers benefit system; reduced marginal tax rate at lower wages; -improve efficiency and evaluation of ALMPs changes in tax-benefit system to improve mobility -allow for wage differentials to reflect productivity across incentives; unemployment benefits reform regions and skills with Social Partner involvement; -relaxation of social criteria for firing and reduced EPL -tackle excessive rigidity of labour market regulations, for small firms

DE including EPL; - some progress towards more efficient ALMPs (Hartz

-reduce regulatory burden inter alia in view of more reforms) flexible working time -promote childcare availability for more female participation -further reforms in education to improve achievements to address skill shortages -improve work incentives in tax and pension systems - partial but still incomplete implementation of 1998 -reduce non-wage labour costs labour market reform packages; some facilitation of part-reform wage bargaining to allow wage differentials to time work reflect productivity and local conditions -changes in tax-benefit system to improve mobility

EL -improve flexibility, modernise work organisation and incentives

review labour market regulation; relax EPL -some progress on child-care facilities -improve educational and vocational training systems to -steps taken to reform pension system enhance skill levels -take measures to raise the female employment rate -reform wage bargaining through effective -tackled excessive rigidity of labour market regulations decentralisation; wage differentials to reflect that discourage hiring and slow adjustment; increased geographical and productivity differences control of fixed-term contracts

ES -further reforms of EPL to reduce market segmentation

-job search and mobility requirements to unemployment across contract types benefits strengthened – disincentives addressed

-remove fiscal distortions to improve mobility -eased search in the housing market (rental reform) to -improve child care facilities and facilitate part-time work increase mobility -review tax-benefit incentives to promote hiring -orientation for lower-level wage bargaining, but no reforms -tax incentives and child care to increase female participation -fully implement new unemployment insurance system; -improved incentives in the benefit systems; increased tax improve incentives for job-search in tax-benefit system credits; reduced marginal tax rates; changes in tax-benefit -reform EPL system to improve mobility incentives -reform the pension system, adapting it to more flexible -reduced in social security contributions

FR employment and reducing early retirement incentives -35 hour week monitored - ‘closely monitor’ the 35 hour week

• - 

  reduce fiscal pressure on labour -improve efficiency of ALMPs - measures undertaken to reconcile work and family life

IE -ensure that wage bargaining allows wage differentials to through increased parental/ care leave rights reflect productivity and skills; monitor wage -better tax incentives to increase female participation

developments -encourage more decentralised wage-setting mechanism, -tackled excessive rigidity of labour market regulations reflecting differences in productivity and skills that discourage hiring and slow adjustment; efforts to -further reform of EPL to reduce market segmentation achieve more flexible contracts

IT across contract types and firm size -decline in marginal tax rates -lower tax on low wages -expansion of private job placement services

-improve childcare and postpone retirement - partial implementation of tightening of eligibility rules -increase resources and efficiency of unemployment for pension and other benefit schemes benefits -increase flexibility of working time -reduce early retirement incentives; tighten disability -reduction of inflow in disability pensions through

LU pension eligibility rules tightening of eligibility; tax-benefit reform to raise female participation

-improve efficiency of ALMPs -improved incentives in the benefit systems; review of -continue benefit reforms, including disability benefits, unemployment insurance, tightening eligibility

NL especially with regard to eligibility and conditionality requirements; reduction in the disability inflow; taxbenefit reform to raise female participation

-agreements on flexibility of working time -measures undertaken to reconcile work and family life through increased parental/ care leave rights

-improve link between contributions and benefits in -2000 reform package reduces tax burden on labour, pensions increases retirement age, and lowers replacement ratio in -speed up tax-benefit reform for older workers unemployment insurance; reduced marginal tax rate

AT -enhance incentives to work and increase low average envisaged effective retirement age -overhaul of pension system, expected to raise

participation; alignment of the retirement age for men and

women

-reform of severance pay

-allow for wage differentials and encourage wage -new labour code, raising duration of fixed contracts (i.a.) moderation (taking into account productivity and skill -better use of ALMP

PT increases in wage growth) -more cost-effective education spending and progress in -improve training and education systems and reduce early proposed enhancement of productivity and skills

school leaving increase flexibility of working time -reform wage bargaining so that wage differentials reflect -steps to reduce marginal tax rate envisaged

FI productivity -steps towards pension reform taken -improve tax-benefit incentives and reform eligibility -more effective and efficient ALMP spending

criteria to make job search more effective; reduce marginal effective tax rates for low wages -increase efficiency of ALMPs -reform income tax to improve work incentives -agreements on flexibility of working time

SE -make ALMPs more efficient -reforms in the tax-benefit system to improve work (and mobility) incentives

-measures to retain older workers and to promote participation of young and immigrants -more effective and efficient ALMP spending

-improve work incentives for all those who can and want -improved incentives in the benefit systems; tax credits

UK to work by reforming sickness and disability benefits and financial incentives extended -efficient active measures for those at most risk of long-merging employment services with the benefit

term unemployment, particularly in deprived areas administration for those in working age -steps towards alignment of retirement age for men and women

-enhanced ALMPs announced Table 7: Priorities and progress on labour market reforms since 2000

Panel A: Priorities as indicated in the BEPGs

IT DE ES FR UK GR BE PT AT NL IE SE FI LU DK

Tax/benefit systems

Wage bargaining

Early retirement

Employment protection

Active labour market policies

Work organisation

Other female labour supply

Education and training

Labour mobility

Panel B: Progress as assessed in the implementation reports

IT DE ES FR UK GR BE PT AT NL IE SE FI LU DK

Tax/benefit systems

Wage bargaining

Early retirement

Employment protection

Active labour market policies

Work organisation

Other female labour supply

Education and training

Labour mobility

Note: “Tax and benefit system” includes recommendations on labour taxation alone; recommendations on active labour market policies in recent BEPGs mostly concerned improving the efficiency of existing policies. Countries are ordered along the horizontal axis according to their approximate potential contribution to raising the EU-15 employment rate. For present purposes, this is taken to be the number of jobs that would be created if each country equalled the performance of the Member State with the highest employment rate in 2000, which was Denmark with an employment rate of 76.4 per cent.

Source: Commission services.

4.3 Labour markets in the enlarged EU Countries in most EU-15 Member States is low, generally below 0.3 per cent of the total population and

between 0.6 and 0.9 per cent in Luxembourg, Germany

4.3.1. Labour market conditions in new Member and Austria, in 2002. Available projections do not States suggest massive east-west net flows of labour, even if

The labour market situation in new Member States is the movement of workers were completely unrestricted considerably worse than in the EU-15. In the Central after the date of accession. Most studies find that the

and East European countries, the process of transition to bulk of the flows would go to Germany and Austria.

274

the market economy has brought about large structural shifts in the labour market, and this accounts for much 4.3.2. Integrating the new Member States into the of the initial decline in employment and the dramatic Lisbon strategy increase in unemployment. However, the persistence of

labour market problems also reflects deep structural Since most of the new Member States have employment problems. Table 8 provides an overview of labour rates that are below the EU-15 average and, moreover, market conditions in the new Member States. The main on a deteriorating trend, their inclusion into the Lisbon

features can be summarised as follows: strategy will clearly make the employment targets – which apply to the EU as a whole – even harder to

• performance varies among the new Member States achieve. On the other hand, with economic growth as much as it does within the EU-15; potential well in excess of 4 per cent in most cases, these

countries are likely to provide a major stimulus to

• nevertheless, employment rates are generally below attaining the overall Lisbon goals of increased

the EU-15 average of 64 per cent, although higher

than in the worst performing countries of the EU-15. competitiveness and dynamism.

They are particularly low among young and older The broad policy challenges in the field of labour working-age people. Female employment rates, once markets do not differ a great deal between new and old higher than in the EU-15, have fallen considerably EU Member States. Structural problems include a very since the start of transition. In contrast to the EU-15, high tax burden on labour and financial disincentives to employment rates in the new Member States as a work in benefit systems, highly regulated permanent whole have been steadily falling, albeit with employment with relatively loose arrangements for differences across countries. Employment losses in temporary contracts, and undifferentiated national agriculture and industry have not been fully offset by minimum wages that are liable to restrict labour market gains in the expanding services sector; access for new entrants, the low-skilled and those living

in less productive regions. Nonetheless, many of the

• unemployment rates are above the EU-15 average in new Member States do face a somewhat different set of

most of the new Member States, though again these

range widely, from 4 per cent in Cyprus to economic circumstances, which suggests that a slight 19 per cent in Poland. Unemployment tends to be change of emphasis may be warranted. Real

concentrated among certain groups, especially the convergence implies that large structural shifts in the labour market are most likely to continue for many

young, the low-skilled and ethnic minorities.

Moreover, a large share of unemployment is longyears. Significant labour reallocation away from

term in most countries. industry and agriculture towards the service sector and a sectoral composition closer to that in Member States in

In general, differences in regional unemployment the EU-15 is likely to continue. This would underline widened during the 1990s and regional disparities in the need for flexible and adaptable labour markets and some new Member States are serious, close to those of institutions and effective ALMPs to facilitate the existing EU Member States with the largest imbalances. adjustment process, by supporting labour mobility Over that period, changes in the structure of across sectors, regions and occupations, as well as by

employment by sector, occupation and firm ownership enhancing workers’ skills.

were dramatic, yet labour mobility within new Member States has been very low, even declining in some

countries. 273

The large income gap, yet relatively small education and skills differentials, as educational attainment is in some cases much higher than the EU-15 average, would argue for the relocation of labour towards EU-15 Member States that enjoy relatively higher wages. There is a high 274

uncertainty about the potential flows of labour from the Nevertheless, as a result of the accession negotiations,

new Member States after the EU Enlargement. The temporary derogations to the principle of free movement of

share of residents from Central and Eastern European workers will apply for a maximum of seven years. All EU- 15 Member States except Ireland and the UK have

announced that they will maintain restrictions on access to their labour markets for workers from the new Member

273 OECD (2002). States, at least for the first two years following accession.

Table 8: Labour market conditions in the new Member States

Employment rate Unemployment

Older Regional All Female workers All Youth Long term disparities *

Country 2003 2003 2003 2003 2003 2003 2002

CY 69.2 60.4 50.4 4.4 10.6 1.1 :

CZ 64.7 56.3 42.3 7.8 18.6 3.8 0.44

EE 62.9 59.0 52.3 10.1 22.9 4.6 :

HU 57.0 50.9 28.9 5.8 13.1 2.4 0.32

LT 61.1 58.4 44.7 12.7 27.2 6.1 :

LV 61.8 57.9 44.1 10.5 17.6 4.3 :

MT 54.2 33.6 32.5 8.2 19.8 3.5 :

PL 51.2 46.0 26.9 19.2 41.1 10.7 0.17

SI 62.6 57.6 23.5 6.5 15.9 3.4 :

SK 57.7 52.2 24.6 17.1 32.9 11.1 0.23

EU-25 62.9 55.1 40.2 9.1 18.4 4.0 :

EU-15 64.4 56.0 41.7 8.1 15.9 3.3 0.63

• * 

  Coefficient of variation = Standard deviation of NUTS 2 regional unemployment rates / National average unemployment rate

Source: Commission services.

The framework for the conduct of macroeconomic 5. General conclusions

policies will remain different from that facing EU-15

Member States for some time. In particular, the new The Lisbon employment targets look much more Member States will continue to operate national challenging than they did in March 2000. When one monetary and exchange rate policies. Notwithstanding looks at the key demographic groups from which most efforts to support a stable macroeconomic climate, they of the increase in employment must come, it is difficult may be more susceptible to economic shocks compared to see how the overall target of a 70 per cent to the EU-15. Against this background, wages will play employment rate can still be achieved by 2010, even in an important role. It will be necessary to avoid wagethe EU-15 let alone the EU-25.

inflation spirals, and also to ensure that real wage

developments support external competitiveness in light The macroeconomic slowdown has not helped, but of the need to attract foreign direct investment. New cannot shoulder all the blame. Progress on structural Member States will continue to face very tight fiscal reforms has not matched the ambitious targets set at constraints. It is therefore essential that labour market Lisbon and Stockholm. Nevertheless, there is evidence policies are affordable and consistent with achieving that much of the improvement in labour market sound fiscal policies that support a stable performance over the 1990s was structural, and that macroeconomic framework. The challenge is in many significant progress has continued in some areas, respects wider than pure labour market concerns, and including tax and benefit reforms and early retirement. relates to the overall structure of the tax systems as well Also on the positive side, there is no mystery about the as administrative efficiency. main determinants of labour market performance, or New Member States need to combine more jobs with about the kinds of measures Member States could take fast productivity growth to catch-up with the EU-15. in order to permanently raise employment rates. The Productivity gains have been substantial during economic evidence – on the determinants of both overall transition, but may be more difficult to sustain as past labour market performance and employment in specific gains were mainly achieved through labour shedding. demographic groups – suggests that the right strategy With the process of privatisation largely completed in has been set out in the BEPGs, as summarised in most countries, this suggests that future productivity Table 7, and in the European Employment Strategy. increases will rely more heavily on investment and Reform strategies should be country-specific, looking at human capital formation, and underlines the need to the ensemble of labour market and social protection tackle skill shortages and improve the quality of institutions. There is scope for improvements in the education and training systems. design of institutions with a view to improving

incentives to take up employment and eliminating deadweight costs and distortions that benefit vested interests rather than providing genuine social insurance.

Governance, as highlighted in the recent report of the

European Employment Taskforce chaired by Wim Kok, is a key priority.

Reforms aimed at raising the employment rate necessarily imply that productivity growth will be temporarily below full potential, because of the implied increase in the labour intensity of production.

Furthermore, net additions to the labour force are likely to be below the average skill level, at least initially. This negative effect on average productivity is estimated to be fairly small. In any event, it should not be regarded as a genuine trade-off. The higher employment rate represents an unambiguous increase in GDP per capita, since newly employed people clearly contribute more to

GDP than they did before. Moreover, there are no reasons to think that a higher employment rate has any negative implications for longer-term productivity growth.

References

A CEMOGLU , K. D. (1999). “Efficient Unemployment insurance”, Journal of Political Economy, Vol.107, pp.893-928

A CEMOGLU , K. D. (2001). “Good Jobs versus Bad Jobs” Journal of Labor Economics, Vol. 19, pp. 1-22.

A CEMOGLU AND S HIMER (2000). “Productivity gains from unemployment insurance”, European Economic Review, Vol. 44, pp. 1195-1224

A GELL , J (2002) “On the Determinants of labour market institutions: Rent seeking versus social Insurance”, German Economic Review Vol. 3, Issue 2, pp. 107-135.

B ASSANINI , A. (2004), “Improving skills for more and better jobs? The quest for efficient policies to promote adult education and training”, Paper presented at a workshop on Quality and Efficiency in Education, European Commission, May.

B AUER , T.K., L OFSTROM M. AND K. Z IMMERMANN (2000), “Immigration Policy, Assimilation of Immigrants and Natives’ Sentiments towards Immigrants: Evidence from 12 OECD-Countries”, IZA Discussion Paper No 187.

B AUER , T.K., H AISKEN -D E N EW , J.P. AND C. M. S CHMIDT (2003), “International Labour Migration, Economic Growth and Labor Markets: The Current State of Affairs”, mimeo Ruhr-Universität Bochum.

B AUER , T.K. AND K UNZE , A. (2004), “The demand for high-skilled workers and immigration policy”, CEPR Discussion Paper No 4274.

B ELL , B LUNDELL AND V AN R EENEN (1999), “Getting the Unemployed Back to Work: The role of targeted Wage subsidies”, IFS Working Paper No 99/12.

B ELOT , M. AND J.C. V AN O URS (2001), Unemployment and labor market institutions: an empirical analysis, CentER Discussion paper No 50.

B ELOT , M. AND J.C. V AN O URS (2004), “Does the recent success of some OECD countries in lowering their unemployment rates lie in the clever design of their labor market reforms?”, Oxford Economic Papers, forthcoming.

B ERTOLA , G. (2004), “A pure theory of Job Security and Labour income Risk”, Review of Economic Studies, Vol 71, pp 43-61.

B ERTOLA , G. (2004), ”Labour Market Institutions in a Changing World”, forthcoming Moneda y Credito.

B ERTOLA , G. (2004) ”Distribution, efficiency and labour market regulation in theory, in OECD countries, and in Latin America”, http://www.personalweb.unito.it/Giuseppe.Bertola/

B ERTOLA , G., B OERI , T. AND N ICOLETTI , G. (eds.) (2001), “Welfare and Employment in a United Europe”, MIT Press.

B ERTOLA , G. AND W. K OENIGER (2004), ”Consumption smoothing and the structure of labor and credit markets” http://www.personalweb.unito.it/Giuseppe.Bertola/

B ERTOLA , G., F.D. B LAU AND L. M. K AHN (2001), “Comparative analysis of labour market outcomes: lessons for the US from international long-run evidence”, NBER Working Paper No 8526.

B ERTOLA , G., F.D. B LAU AND L. M. K AHN (2002), ”Labour Market Institutions and Demographic Employment Patterns”, CEPR Discussion Paper No. 3448.

B LANCHARD , O. (2004), “The Economic Future of Europe”, NBER Working Paper No 10310.

B LANCHARD , O. AND J. W OLFERS (2000). “The role of shocks and institutions in the rise of European unemployment: the aggregate evidence”, Economic Journal, Vol. 110, pp. 1-33.

B LANCHARD , O. AND D. Q UAH (1989), "The Dynamic Effects of Aggregate Demand ands Supply Disturbances", American Economic Review, Vol. 79 (4), pp. 655-673.

B LANCHFLOWER , D.G., R.B.F REEMAN (2000), “The Declining Economic Status of Young Workers in OECD Countries”, in: David G.Blanchflower and Richard B. Freeman (editors), Youth Employment and Joblessness in Advanced Countries, National Bureau of Economic Research - Comparative Labor Markets Series.

B LANK , R. (2002) “Can equity and efficiency complement each other?” NBER Working Paper No 8820.

B LAU , F.D., AND L. M. K AHN (1996), “International Differences in Male Wage Inequality: Institutions versus Market Forces”, Journal of Political Economy, Vol.104, Issue 4, Suppl. pp.791-837.

B LAU , F.D., AND L. M. K AHN (1999), “Institutions and laws in the labour market”, in O. Ashenfelter and D. Card (eds.), Handbook of Labor Economics, Vol.3A, North-Holland, pp.1399-1461.

B LÖNDAL , S. AND S. S CARPETTA (1998): “The Retirement Decision in OECD countries”, OECD Economics Department Working Paper, No. 98

B LÖNDAL , S. AND S. S CARPETTA (1999): “Early Retirement in OECD countries: the Role of Social Security Systems”, Economic Studies No. 29, OECD

B OERI , T. (2001), ‘Conclusions: Is There a Role for Supranational Institutions?’, in Boeri, G. Bertola and G. Nicoletti (eds.), Welfare and Employment in a United Europe, Cambridge, MA and London: MIT Press, pp. 251-255.

B OERI , T., A. B RUGIAVINI , AND L. C ALMFORS (2001) “The role of Unions in the Twenty-First Century”, Oxford University Press.

B OERI , T., C ONDE -R UIZ , J. I. AND V. G ALASSO (2002), Protecting Against Labour Market Risk: Employment Protection or Unemployment Benefits?, mimeo.

B UDDELMEYER , H., G. M OURRE AND M. W ARD -W ARMEDINGER (2004), ‘Recent developments in part-time work in EU countries: Trends and Policy’, in Labour Supply and Incentives to Work in Europe, edited by R. Gomez Salvador, A. Lamo, B. Petrongolo, M. Ward and E. Wasmer, Edward Elgar (forthcoming) .

B URNIAUX , J. M., R. D UVAL AND F. J AUMOTTE (2003), ‘Coping with ageing: a dynamic approach to quantify the impact of alternative policy options on future labour supply in OECD countries’, OECD Economics Department Working Paper No. 371.

B UTI , M., L.R. P ENCH AND P. S ESTITO (1998), “European unemployment: Contending theories and institutional complexities”, European University Institute Florence, Robert Schuman Centre Policy Papers No 98/1.

C ALMFORS , L (1993). “Centralisation Wage Bargaining and Macroeconomic Performance”, OECD Economic Studies, 21.

C ARD , D. (2004), “Is the New Immigration Really So Bad?” IZA Discussion Paper No 1119.

C ASEY , B. (1998), “Incentives and Disincentives to Early and Late Retirement”, Working Paper AWP3.3, OECD, Paris.

C HISWICK , B. (1978), “The Effect of Americanisation on the Earnings of Foreign Born Men”, Journal of Political Economy, Vol. 86 (5).

C OE , D. AND S NOWER , D. (1997), “Policy Complementarities: The Case for Fundamental Labor Market Reform”, IMF Staff Papers, Vol. 44(1), pp. 1-35.

C ONDE -R UIZ , J.I. AND V. G ALASSO (2002), “Early Retirement”, CEPR Working Paper, May 2002.

C OUNCIL OF THE E UROPEAN U NION (2003), “Joint Report by the Commission and the Council on Adequate and Sustainable Pensions”, 7165/03.

D AVIS AND H ENREKSON (2000), “Wage-Setting Institutions as Industrial Policy”, NBER Working Paper 7502.

D E K ONING , J. AND H. M OSLEY (2001). “Labour market policy and unemployment. Impact and Process evaluation in selected European countries”, Edward Elgar.

D USTMANN , C. AND F. F ABBRI (2000), “Language Proficiency and Labour Market Performance of Immigrants in the UK”, CEPR Discussion Papers No 2487.

ECB (2002), ‘The composition of employment growth in the euro area in recent years’, ECB Monthly Bulletin, November, pp. 67-79.

E LMESKOV , J., J.P. M ARTIN , AND S. S CARPETTA (1998). “Key lessons for Labour Market Reforms: Evidence form OECD Experiences” Swedish Economic Policy Review, Vol.5.

E UROPEAN C OMMISSION (2003) “Gender pay gaps in European labour markets”, Commission Staff working paper, SEC(2003) 937.

E UROPEAN C OMMISSION (2004a), Employment in Europe: Recent Trends and Prospects, Luxembourg: Office for Official Publications of the EU.

E UROPEAN C OMMISSION (2004b), Communication from the Commission on “Increasing the employment rate of older workers and delaying the exit from the labour market”, COM(2004) 146 final i.

F EDERAL R ESERVE B ANK OF B OSTON (2004), “Understanding the “job-less recovery”“, Public Policy brief No. 04-1.

F ITOUSSI , J.P., D. J ESTAZ , D. P HELPS , E.S. Z OEGA (2000). “Roots of the Recent Recoveries: Labor Reforms or Private-Sector Forces?”, Brookings Papers on Economic Activity, Vol. 1, pp. 237-312.

F REDRIKSSON , P. AND B. H OLMLUND (2004). “Optimal Unemployment Insurance Design: Time Limits, Monitoring, or Workfare?”, CESifo Working Paper No. 1019

G ALÍ , J. (1999), “Technology, Employment, and the Business Cycle: Do Technology Schocks Explain Aggregate Fluctuations?”, American Economic Review, Vol. 89, pp. 249-271.

G ARIBALDI P. AND P. M AURO (2002), “Employment growth. Accounting for the facts”, Economic Policy, April.

G AUDE J. (1997), L’insertion des jeunes et les politiques d’emploi-formation, Cahiers de l’emploi et de la formation, no.1, Bureau international du Travail, Genève

G HELLAB , Y. (1998), Minimum Wages and Youth Unemployment, Employment and Training Papers no.26, International Labour Office, Geneva

G ORDON , R.J. (1995): “Is There a Trade-off between Unemployment and Productivity Growth?” CEPR Discussion Paper No. 1159.

G RUBER , D. AND D. W ISE (1999): “Social Security Programmes and Retirement around the World”, University of Chicago Press, Chicago.

G RUBER , J. (1994): “The Consumption smoothing benefit of unemployment insurance” NBER Working Paper 4750.

H ASSLER , J., J.V. R ODRÌGUEZ M ORA , K. S TORESLETTEN AND F. Z ILIBOTTI (2001) “Unemployment, Specialization, and Collective Preferences for Social Insurance” in The New Economics of Rising Inequalities, pp. 331-50, edited by Daniel Cohen, Thomas Piketty and Gilles Saint-Paul, Oxford University Press, 2001

H ATTON , T. J. (2004): ‘Seeking asylum in Europe’, Economic Policy, April, pp. 5-62.

H ATTON , T.J. AND S. W HEATLEY P RICE (1998): ‘Migration, Migrants and Policy in the United Kingdom’, CEPR Discussion Papers No 1960.

H ECKMAN , J. (2000), “Policies to Foster Human Capital”, Research in Economics, Vol. 54, No 1, 2000, pp. 3-56.

I NTERNATIONAL O RGANIZATION FOR M IGRATION (2003): World Migration 2003: Managing migration challenges and responses for people on the move; vol. 2, IOM World Migration Report Series.

J AUMOTTE , F. (2003), “Female labour force participation: past trends and main determinants”, OECD Economics Department Working Paper No. 376.

J ENSEN ,P., M. R OSHOLM , M. S VARER , (2003) ”. ”The Response of Youth Unemployment to Benefits, Incentives, and Sanctions” forthcoming in European Journal of Political Economy.

J IMENO , J.F. AND D. R ODRÍGUEZ -P ALENZUELA (2003), Youth Unemployment in the OECD: Demographic Shifts, Labour Market Institutions and Macroeconomic Shocks, ENEPRI Working Paper no.19.

K AHN , L. (2000), “Wage inequality, collective bargaining and relative employment from 1985 to 1994: evidence from fifteen OECD countries”, Review of Economics and Statistics, Vol. 82(4), pp. 564-579.

K ORENMAN , S. AND D. N EUMARK (1997) Cohort Crowding and Youth Labor Markets: A Cross-National Analysis, NBER Working Paper, no. 6031.

L AYARD , R., N ICKELL , S. AND J ACKMAN , R. (1991), “Unemployment: macroeconomic performance and the labour market”, Oxford University Press.

L INDBECK , A. AND D. S NOWER (1988), The Insider-Outsider Theory of Employment and Unemployment, Cambridge, MA: MIT Press.

M ADSEN , K. (1999), “Denmark: Flexibility, security and labour market success”, ILO Employment and Training Papers No 53.

M ORTENSEN AND P ISSARIDES (1999), “Unemployment response to skillbiased technology shocks: the role of labour market policy”. The Economic Journal, Vol. 109, pp. 242-265

M OSISA , A.T. (2002), “The Role of Foreign-Born Workers in the U.S. Economy”, Monthly Labor Review, Bureau of Labor Statistics, May.

N ICKELL , S. AND L AYARD , R. (1999). “Labour Market Institutions and Economic Performance” in Ashenfelter, O. and Card, D., (eds) Handbook of Labour Economics, Vol. 3, pp. 3029-3084.

N ICKELL , S. N UNZIATA , L., O CHEL , W., Q UINTINI , G. (2003), “The Beveridge Curve, Unemployment and Wages in the OECD from the 1960s to the 1990s”, in: Aghion, P. et al. (eds.), Knowledge, Information and Expectations in Modern Macroeconomics: In Honor of Edmund S. Phelps, Princeton University Press.

N ICKELL , S.J. AND J.C. V AN O URS (2000). “The Netherlands and the United Kingdom: a European unemployment miracle?”, Economic Policy, Vol. 30, pp. 137-175

OECD (1998), “Maintaining Prosperity in an Ageing Society”, Paris.

OECD (2000), “Reforms for an Ageing Society”, Paris.

OECD (2001), Employment Outlook, Paris.

OECD (2002), “Trends in international migration”, SOPEMI Annual Report.

OECD (2003a), “Labour Force Participation of Groups at the Margin of the Labour Market: Past and Future Trends and Policy Challenges”, ECO/CPE/WP1(2003)8

OECD (2003b): ‘The economic impact of migration”, Economic Survey: Spain.

O’H IGGINS , N. (1997), “The challenge of youth unemployment”, Employment and Training Papers No.7. International Labour Office, Geneva.

O LIVETTI C. (2001), “Changes in Women’s Hours of Market Work: The effect of changing returns to experience”, PhD Dissertation, University of Pennsylvania.

P ESTIEAU , P. (2001), “Are We Retiring too Early?”, CESifo Working Paper No. 522.

P ISSARIDES , C HRISTOPHER (1986), “Unemployment and Vacancies in Britain”, in: Economic Policy, No. 3, pp.499- 559.

P ISSARIDES , C., P. G ARIBALDI , C. O LIVETTI , B. P ETRONGOLO , AND E. W ASMER (2003), “Women in the labour force: how well is Europe doing?”, Fondazione Rodolfo De Benedetti.

P ORTUGAL , P. AND J. V AREJÃO (2003). “Why do firms use fixed-term contracts?”, Bank of Portugal Working Paper No 8/03.

R OSEVEARE , D. AND M. J ORGENSEN (2004), “Migration and Integration of Immigrants in Denmark”, Economics Department Working Papers N° 368, April.

S TOCK , J.H. AND W ATSON , M.W. (1988), “Testing for Common Trends”, Journal of the American Statistical Association Vol. 83, pp. 1097-1107.

V AN O URS , J.C. (2003), “Has the Dutch miracle come to an end?”, CentER Discussion Paper No. 32.

V ARIAN , H.R. (1980), “Redistributive taxation as social insurance”, Journal of Public Economics, Vol. 14, pp. 49-68.

V LASBLOM J. AND J. S CHIPPERS (2004), “Increases in female labour force participation in Europe: similarities and differences”, T. C. Koopmans research Institute, Discussion Paper Series 04-12. Utrecht University.

W ALMSLEY , T. AND W INTERS , L. (2003), “Relaxing the Restrictions on the Temporary Movements of Natural Persons: A Simulation Analysis”, CEPR Discussion Paper No. 3719.

• 4. 

  THE LISBON STRATEGY AND THE EU’S

STRUCTURAL PRODUCTIVITY PROBLEM

Summary

The structural nature of the EU’s productivity downturn is confirmed, with the bulk of the deterioration emanating from an outdated and inflexible industrial structure which has been slow to adapt to the intensifying pressures of globalisation and rapid technological change. The EU’s productivity problems are driven by the combined effect of an excessive focus on low and medium-technology industries (with declining productivity growth rates and a globalisation-induced contraction in investment levels); an inability to seriously challenge the USA’s dominance in large areas of the ICT industry, as reflected in the relatively small size of its ICT production sector; and finally, its apparent slowness in reaping the productivity enhancing benefits of ICT in a range of ICT-using industries, although measurement issues severely complicate an assessment of the gains from ICT diffusion.

The post-1995 differences in EU-US productivity patterns are fundamentally driven by the USA’s superiority in terms of its capacity to produce and absorb new technologies, most notably in the case of ICT. Healthy knowledge production and absorption processes are mutually supportive elements of any successful long run productivity strategy. Evidence is presented which suggests that the USA’s overall innovation system is superior to that of the EU’s, both in terms of the quality and funding of its knowledge sector and the more favourable framework conditions prevailing. The repeated ability of the US system to direct resources towards the newer, high technology (and often high productivity growth), industries is a reflection of the quality of the interrelationships between the different actors in its innovation system and of an economic and regulatory framework which has the capacity to transform excellence in knowledge creation into a globally competitive industrial structure.

The systemic inadequacies of the EU’s innovation system are highlighted by the experience of the ICT industry, with the history of this industry suggesting that a “national champions” strategy in high technology industries is highly problematic. A wide range of factors are shown to have contributed to the USA’s global dominance in ICT. These factors include focussed R&D activities; world class research and teaching establishments; defense procurement contracts which nurtured the ICT industry (on the demand side) in its incubation phase in the 1950s and 1960s; and the unique combination of financing mechanisms and a highly competitive domestic marketplace which brought the ICT industry from the knowledge creation phase to the critical diffusion/mass market phase. The history of the ICT industry also suggests that a “national champions” strategy in high technology industries is a recipe for failure, with the chapter highlighting in particular the large price which Europe has paid for its “national champions” policy in this particular industry back in the 1960s and 1970s, which contrasted sharply with the strategies adopted by Japan and the USA.

In terms of policy, the chapter stresses that the EU’s innovation system needs to be fundamentally reformed if the EU is to make a decisive shift towards realising the vision of a successful, innovation-based, economic model, the broad features of which have been laid out in the Lisbon 2010 agenda. The success of such a model will be determined not so much by a massive increase in the amount of financial resources devoted to knowledge production (i.e. increased spending on R&D and higher education) but by an acceptance of the need to improve linkages in the innovation system and to make painful changes in many areas of the EU’s economic and regulatory environment. More specifically the present study stresses the following:

• The systemic nature of the innovation process needs to be fully recognised and the quality of the

interrelationships between the different actors in the EU’s system needs to be dramatically improved.

• The public and private sectors each play important, mutually supportive, roles in determining a country’s

innovation capacity and each must assume its responsibilities if the EU’s knowledge economy objectives

are to be realised.

• Industry-specific framework conditions need to be taken into account by EU policy makers due to the

complicated link between competition and innovation. Product market conditions and the characteristics of specific technologies are what ultimately determines the relationship between market concentration

and R&D intensity.

• Market entry and exit rules, by putting pressure on incumbent firms to innovate and by supporting

market experimentation, are crucial to an effective innovation process in rapidly changing industries.

TABLE OF CONTENTS

• 1. 

  I NTRODUCTION ............................................................................................................................................163

• 2. 

  EU PRODUCTIVITY TRENDS AT THE E CONOMY WIDE LEVEL ........................................................................164

2.1 Overview of main trends .................................................................................................................................... 164

2.2 ICT as an explanatory factor at the total economy level ..................................................................................... 165

2.3 Are low EU productivity growth rates a permanent phenomenon or a temporary blip? ..................................... 165

• 3. 

  T HE STRUCTURAL NATURE OF THE EU’ S PRODUCTIVITY PROBLEM .............................................................168

3.1 A 56 industry breakdown of labour productivity trends ..................................................................................... 168

3.2 Where are the ICT productivity gains coming from? ......................................................................................... 172

• 4. 

  E NHANCING THE EU’ S PRODUCTIVITY PERFORMANCE ................................................................................177

4.1 The knowledge economy must be a central element ........................................................................................... 177

4.2 The USA has a superior innovation model ......................................................................................................... 178

4.3 Reforming the EU’s Innovation Capacity ........................................................................................................... 182

• 5. 

  S UMMARY OF KEY FINDINGS .......................................................................................................................187

R EFERENCES ..........................................................................................................................................................190

ANNEX: P RODUCTIVITY MEASUREMENT ISSUES ...............................................................................................191

THE LISBON STRATEGY AND THE EU’S

STRUCTURAL PRODUCTIVITY PROBLEM

• 1. 

  Introduction Firstly, how does the EU compare with the USA in terms of economy-wide productivity trends and how big

Europe’s growth performance has been the subject of a role has ICT played in explaining the diverging increasing scrutiny over recent years, most notably in patterns? Furthermore, should the post-1995 the context of the Lisbon process and its efforts to deterioration in EU productivity be interpreted as a encourage governments to introduce employment and transitory or a structural phenomenon? productivity enhancing reforms. This reform agenda is Secondly, in explaining recent EU-US divergences in all the more pressing given that the EU’s underlying productivity trends, to what extent is the EU’s relatively growth rate has been trending downwards since the poor performance linked with its particular industrial second half of the 1990s and since the medium to long structure and its difficulty in reorientating its economy term outlook points to a continuation of these trends. towards the newer, higher productivity, growth sectors While many EU countries are understandably such as ICT? In terms of the specific role of ICT, the preoccupied with extricating their economies from the chapter asks whether the contribution of the ICT- relatively prolonged short run downturn, it is widely producing industries to overall productivity patterns has acknowledged that many of the solutions to this slow been underestimated in favour of ICT diffusion growth problem require a longer term policy explanations which stress the crucial role of a small perspective. A sustainable medium-term recovery number of intensive ICT-using industries such as process, according to a wide range of commentators, wholesale and retail trade.

demands action on a Lisbon-inspired structural reform

agenda aimed at effectively addressing the EU’s Finally, in the context of delivering on the EU’s longer fundamental growth challenges, presently posed by the term ambitions of progressively moving towards a more accelerating pace of technological change, globalisation knowledge-based economy, the study focuses on the (most recently in terms of the growing tradeability of specific role to be played by the production and large parts of the service economy) and ageing absorption of new technologies in any overall strategy. populations. While the present chapter fully accepts that the

absorption of innovation from other industries/countries

Whilst accepting the absolute necessity of encouraging a will remain a fundamental element in determining the more labour intensive growth pattern over the medium EU’s future productivity performance, it nevertheless to long term, the present chapter focuses on the argues strongly in favour of a greater recognition productivity part of the Lisbon agenda. It specifically amongst EU policy makers of the importance of a analyses the nature/source of the deterioration in the globally competitive knowledge production system to EU’s productivity performance relative to that in the the realisation of the Lisbon goals. Creating a system USA since the mid-1990s and outlines the approach to capable of delivering on both aspects of the innovation be adopted in order to remedy this situation. Given the process is not simply an issue of more spending on R&D extensive treatment accorded to the productivity theme and 3 rd level education. More importantly it is a question in last year’s Review, the present chapter will build on of better linkages between the different players in the this latter work by focussing on three specific issues innovation system and a recognition of the need for a related to the EU’s recent productivity performance and dynamic, competitive, business environment in of its ambitions to become the most competitive, accelerating the move from the knowledge knowledge based, economy in the world by 2010: creation/absorption phase to the critical commercial

phase.

• 2. 

  EU productivity trends at the Economy over point occurring in the mid-1990s. 276 This recent EU

wide level performance marks a serious downgrading relative to the situation in the early 1990s when annual EU labour

productivity growth was averaging 2½ per cent, 2.1 Overview of main trends compared with 1½ per cent for the USA. Since that time

At the moment, EU living standards (GDP per capita) there has been a dramatic reversal in fortunes, with the are at roughly 70 per cent of US levels, with about 1/3 of EU’s labour productivity growth rate declining by a full the gap due to labour productivity differences, with the 1 per cent point and with the USA’s accelerating by a

remaining 2/3 due to differences in the utilisation of roughly similar amount.

275

labour (measured in hours worked). The EU has also Graph 2: Labour productivity per hour growth trends

experienced some important changes over the course of the 1990s with, on the positive side, the previously

downward movement in total hours worked relative to 6 the USA coming to an end and, on the negative side, the 5

post World War II convergence to US productivity EU15 levels going into reverse. In fact, after having peaked in 4 USA the mid-1990s at around 97 per cent of US levels, EU

labour productivity per hour is projected to deteriorate to p .a

. 3

around 88 per cent in 2005, which is close to its relative % level in the early 1980s (Graph 1). This post 1995 2 deterioration in relative productivity levels reflects a 1

sharp decline in EU productivity growth rates relative to

those of the USA over the period in question. 0

1966 1972 1978 1984 1990 1996 2002

Graph 1: Productivity trends, EU-15 relative to the USA Source: Commission services.

120 Labour productivity per hour

110 GDP per capita From a purely growth accounting perspective, the

Labour input per capita

100 1 percentage point decline in EU labour productivity emanates from 2 sources:

90 Firstly, 50 per cent can be attributed to a reduction in the

80 contribution from capital deepening i.e. relatively low rates of investment per worker.

70

60 Secondly, the remaining 50 per cent appears to emanate from a deterioration in total factor productivity i.e. a

50 1965 1971 1977 1983 1989 1995 2001

Source: Commission services. 276 While overall EU productivity trends have clearly

deteriorated over recent years, it is important to underline the wide range of performances at the individual EU

Graph 2 shows labour productivity per hour trend Member State level, with large numbers of countries

developments in the USA and the EU since the midcomparing favourably with international trends. With regard 1960s. Over most of that time, and indeed for most of to the performance of the EU-15 Member States, there is a the post World War II period up until the mid-1990s, the clear divergence for the euro-area and non-euro area

EU has enjoyed productivity growth rates well in excess countries. The non-euro area Member States have been able to arrest the decline in their 1980s productivity growth rate

of those prevailing in the USA. Given relatively low and stablise it in the 1½ - 2 per cent range over the 1990s. employment rates, the EU was able to use its superior Over the same period the euro-area countries as a group

productivity performance to broadly maintain its living have experienced a decline in their productivity growth rate standards. This is why policy makers need to be from close to 2 per cent to well under 1 per cent. This euroseriously

concerned with the fact that the EU is now, for area pattern is totally dictated by developments in the big the first time in decades, on a trend productivity growth four area countries, namely Germany, France, Italy and

path which is lower than that of the USA, with the cross Spain. The remaining 8 euro-area countries have managed to achieve an acceleration in their productivity growth rates

between the first and second halves of the 1990s similar to that which occurred in the USA. The problem of course is that with the big four countries accounting for nearly 80 per cent of overall euro-area output, the poor performances

275 See Annex 1 for an overview of the ongoing discussion on from all four of these countries ensures that the “area” as a

measurement issues. whole has a clear productivity problem.

decline in the overall efficiency of the production

process. 2.3 Are low EU productivity growth rates a permanent phenomenon or a temporary blip?

2.2 ICT as an explanatory factor at the total Are low EU productivity growth rates likely to be a

economy level permanent phenomenon or a temporary blip linked to

One of the most popular explanations for the diverging labour market reforms? To help answer this question, it productivity fortunes of the EU and the USA has been is helpful to review the basic growth patterns the relative exposure of both areas to ICT. Last year’s (employment and productivity) between the EU and the Review showed that ICT has indeed been an important USA in the 1990s

part of the story, especially in terms of explaining the • The EU’s trend productivity growth rate, as shown

turnaround in the productivity trend of the USA. The earlier, continued to decline throughout the 1990s

overall contribution to labour productivity growth from

ICT investments (i.e. purchases of software, computing and fell below the equivalent, and rapidly and communications equipment) and from technical increasing, US productivity growth rate around the progress in the production of ICT goods and services middle of the decade.

(e.g. the semiconductor and telecommunications • Regarding employment, the decline in employment

industries) accounted for about 60 per cent of US labour rates in the EU came to an end in the early 1990s productivity growth over the second half of the 1990s, and started to trend upwards. In the USA a positive compared with 40 per cent in the four EU countries for trend continued but at a slower pace.

which such a breakdown exists. 277 This translates over • Closely associated with the movement in labour

the second half of the 1990s into an ICT contribution to

labour productivity growth of around 1½ percentage productivity growth, fairly parallel trend

points in the USA and ¾ of a percentage point in the developments for capital-labour substitution were

case of the EU-4. observed, i.e. a further decline in Europe and an increase in the USA.

In terms of the trend acceleration in US labour

productivity growth over the two halves of the 1990s, Graph 4 shows the basic movements for the EU for about half of the 1 percentage point acceleration can be these three variables, with these trends especially directly attributed to ICT. In the case of the EU-4 group striking when contrasted with those of the USA over the of countries, the effects of ICT on both capital same period (Graph 5). This comparison shows in a deepening and TFP over the same period was positive, dramatic way the extent to which the EU economy is although significantly less positive than in the USA. failing to exploit the technological opportunities which Consequently, given that ICT was not responsible for are presently available in the world economy. The USA the deteriorating EU productivity trend, the role of nonin contrast has experienced a marked trend reversal in its ICT determinants such as labour market reforms or the labour productivity performance, with the latter strongly EU’s outdated industrial structure needs to be assessed. linked to its exploitation of the opportunities presented Section 2.3 looks at the role of labour market reforms, by the ICT industry. The trends in these graphs can be with Section 3 asking whether an excessive focus on assessed in alternative ways, with different traditional, low productivity growth, industries could be interpretations having different implications for the long

responsible for the deteriorating EU trend. run outlook for productivity and employment, with our main interest here being productivity.

• A popular interpretation explains the recent

productivity trends as a response of the economy to a positive labour supply shock. The shock to labour supply/wages could be the result of labour market reforms. It could also reflect an increasing

277 These are France, Germany, the Netherlands and the UK.

They constitute the EU-4 in the subsequent analysis.

Graph 3: Breakdown of trend labour productivity into capital deepening and TFP

2.5 4

Total factor productivity EU 15

2 Capital deepening EU 15

3 Total factor productivity USA

Capital deepening USA

1.5 . p .a

.

 p .a 2

% % 1

1 0.5

0 0

1966 1972 1978 1984 1990 1996 2002 1966 1972 1978 1984 1990 1996 2002

Source: Commission services.

Graph 4: EU labour productivity, capital-labour, substitution and employment rate trends

6 0.8

5 0.75

Capital-labour

4 substitution

0.7 Employment rate

.a . Labour productivity

. 3 p

.a

% p % 0.65

2

0.6 1

0 0.55

1966 1972 1978 1984 1990 1996 2002 1966 1972 1978 1984 1990 1996 2002

Source: Commission services.

Graph 5: US labour productivity, capital-labour substitution and employment rate trends

4 0.8

Capital-labour substitution

3 0.75

Labour productivity Employment rate

. 0.7

 p .a 2 .a

.

% p %

0.65

1 0.6

0 0.55

1966 1972 1978 1984 1990 1996 2002 1966 1972 1978 1984 1990 1996 2002

Note: The employment rate is defined relative to the population of working age.

Source: Commission services.

awareness amongst European citizens that pension

income will be more uncertain in the future. This Graph 6: Euro-area employment shock: 1995 Q1 to 2003 Q4

negative income effect could have contributed to an

increase in labour force participation. Under this 6

interpretation, recent developments could be judged 5 Employment as healthy. Slower wage growth could have led to a 4 Productivity

temporary decline in capital-labour substitution. 3

Once full employment is reached, wage and productivity growth could accelerate again and the 2 economy could go back to a higher growth rate of 1 labour productivity at a higher level of employment.

The decline in productivity growth and in capital 0

labour substitution (i.e. capital deepening) could -1

thus be regarded as a temporary phenomenon. -2

• An alternative view regards the labour market story 95Q2 96Q2 97Q2 98Q2 99Q2 00Q2 01Q2 02Q2 03Q2

as incomplete. According to this view, the data can Source: Commission services. be explained correctly only if one assumes a negative shock to productivity, either in the form of

a decline in the growth rate of TFP or in the form of identify the driving forces behind changes in a positive shock to capital productivity, with the employment and productivity and, in addition, to latter shock induced by higher required rates of analyse the temporary versus permanent nature of the

return for investors. At the macro level a trend effects.

decline in TFP could be due to a further increase in The employment chapter in the present EU Review 278 the size of the service sector; a reduction in the provides an in-depth examination of the first question in quality of labour as more low skilled workers are the VAR analysis, namely to what extent the increase in brought into the labour force; and/or a trend decline employment can explain the decline in productivity in technological advances in traditional growth. The VAR analysis identifies a sequence of manufacturing industries. Also with globalisation positive employment shocks in the second half of the and increased international capital mobility, the 1990s which have increased the level of employment by higher returns which can be earned outside Europe about 5 per cent in the euro area. The shock driving may exert pressure on capital productivity. Both employment, however, only had a small effect on developments could explain why capital-labour productivity (Graph 6). According to the estimate, the substitution declined. 5 per cent increase in employment has reduced the level

of productivity by only about 0.75 per cent. This is

Both of the above interpretations would obviously about 10 per cent of the total reduction in productivity provide a different diagnosis for Europe. According to growth experienced since the mid 1990s. Hence, the first view, recent productivity trends are a temporary employment shocks can only marginally explain the

phenomenon and a healthy indication that labour decline in productivity growth. 279

markets in Europe have become more flexible. The

second view is more pessimistic. It regards the The second contribution of the VAR model relates to the productivity slowdown as a continuation of the previous question of the structural versus temporary nature of the adverse productivity trends, with the recent increase in effects. Based on the underlying assumptions on the employment simply having an additional temporary, short, medium and long term impact of the various negative, effect on productivity. The productivity picture shocks, the VAR model attributes most of the decline in is further complicated by a third possible explanatory productivity to a structural trend decline in productivity factor, namely aggregate demand, with domestic growth. As can be seen from Graph 7, the autonomous demand over the most recent period being sluggish, shock to productivity explains a decline in the level of triggering a cyclical impact on measured productivity. productivity of 5 per cent, which would translate into an

To analyse more rigorously whether the productivity pattern is temporary or more structural, we need to be

able to identify the nature of the shocks driving 278 Chapter 3 contains a technical description of the VAR

productivity. We use a VAR methodology to analyse the analysis.

various contributions to the productivity slowdown, 279 It should be noted that this VAR estimate of 10 per cent is

coming from the three shock variables: employment, at the lower end of the estimates obtained using a range of productivity and demand. A VAR analysis is estimation methods. For example, results from the particularly suited for this purpose since it allows us to Commission’s QUEST model suggest that about 30 per cent

of the reduction in productivity growth could be explained by the employment shock. Also results from growth regressions suggest that about 25 per cent of the

productivity decline is due to the increase in employment.

Graph 7: Euro-area productivity shock: 1995 Q1 to 2003 Q4 outdated industrial structure which has failed to fully exploit the direct and indirect productivity 1 benefits from new, leading edge, technologies such as ICT.

0 • Secondly, whilst not questioning the overall

-1 contribution of ICT to labour productivity trends, Section 3.2 adds to the ongoing debate regarding

.a . -2 the relative importance of the different channels (i.e. % p production, investment and spillover effects) via

-3 Employment which ICT impacts on the respective economies. It Productivity is contended that a large proportion of the recent

-4 literature may be underestimating the direct gains from the production of ICT goods and services in

-5 favour of the view that most of the gains emanate 95Q2 96Q2 97Q2 98Q2 99Q2 00Q2 01Q2 02Q2 03Q2 from the use of ICT. This debate on the respective

Source: Commission services. contributions of the different ICT transmission channels is important to the policy debate in the

annual average productivity growth rate effect of the final section of this chapter when we discuss a order of 0.6 per cent. This is fully consistent with the productivity agenda for the EU and the importance growth accounting result given earlier of a decline in to be attributed to the production and absorption of

TFP of the order of ½ a percentage point, with TFP new technologies.

considered to be a reflection of the structural component

of the productivity trend. Graph 7 also indicates that the 3.1 A 56 industry breakdown of labour

autonomous productivity shock is unable to explain the

increase in employment. Therefore, an interpretation of productivity trends

both shocks is necessary in order to give a complete The basis for this industry level analysis is a 56 industry picture of the employment and productivity breakdown of the EU and US economies, which enables developments. 280 However, concerning productivity, the us to show the contribution of each of the individual overall conclusion from the analysis suggests that the industries to overall labour productivity growth in both decline in productivity growth is to a large extent areas (i.e. the combined effect of productivity growth in structural in nature. the specific industry and of its share in overall output). 281 This breakdown is shown in Graph 8 and

visualises the productivity dilemma facing the EU by

• 3. 

  The structural nature of the EU’s giving a panoramic overview of the contribution of the

productivity problem 56 industries. For ease of exposition the industries are shown as part of the manufacturing, private services and

rest of economy (primary industries and public services) The present section extends the analysis from Section 2,

in particular its suggestion that the EU has a structural sectors.

productivity problem, by taking a closer look at

sectoral/industry level productivity developments. Two 3.1.1 Overview of all 56 industries (1996-2000)

specific issues are examined: Graph 8 shows that the EU has been doing reasonably

• Firstly, an attempt is made in 3.1 to isolate the well compared with the USA in a wide range of source of the EU’s productivity problems at the manufacturing and service industries over the second sectoral level: are those difficulties confined to the half of the 1990s. However, the problem is that most of manufacturing, private services or rest of the these industries are not making big contributions to economy sectors or linked to particularly dynamic overall productivity growth, with the graph indicating a specific industries within these broad categories? In contribution of much less than 0.1 per cent for most of addition, by categorising the different industries on the industries concerned. For example, while Graph 8 the basis of their ICT content into ICT producing, shows that the EU’s chemical industry contributed more

intensive ICT-using and less-intensive ICT-using industries, the section gives a more detailed insight into the role of ICT in shaping overall EU and US productivity trends. The key question is to what

extent Europe’s problems reflect an inflexible and 281 For the analysis in this section we use an internationally

comparable dataset from the Groningen Growth and Development Centre (GGDC). This dataset has a 56

280 See the employment chapter for a more detailed discussion industry breakdown for all 15 of the old Member States and

of the trade-offs between productivity and employment for the USA and is essentially an expanded version of the

shocks. OECD’s STAN database.

S o

ur

ce -0

• : 

  C .2

  -0 0.

  % .1 % 0%

  • 0. 

    1% 0. 2% 0. 3% 0. 4% 0. 5% 0. 6%

o m

m iss Semiconductors

io n

 s Office Machinery er

vic Clothing es Chemicals a

n d Rubber & Plastics

 G

G Textiles D

C Motor Vehicles .

 Furniture

Basic Metals G

 Air & Spacecraft ra

Pulp & Paper M p h 8 Metal Products a nu : C

 Instruments fa o n

 Railroad Equipment ctu tr ib

Insulated Wire P rin u g tio

 Ships & Boats n s Leather o f t

Telecom Equipment h e

Mineral Products 5 6 in

Oil & Nuclear Fuel d

 Printing & Publishing u st

Radio and TV U rie S E

A U s Wood & Wood Products 1 5 to

 o

 Electrical Machinery ve ra

9 b Food, Drinks & Tobacco o u

 Wholesale Trade r p ro

 Retail Trade d

Real Estate Activities u c tiv

 Aux.Financial Services it y

 Financial Services g ro

Communications w

 Renting of Machinery P th in

Air Transport riv th

 Insurance & Pensions ate e U

Aux.Transport Activities s e S +

 Research & Development rv ic E U

Land Transport es 1 5 Water Transport ( 19

 Legal & Advertising 9 6 -2

Computer Services 0 0

Electricity, Gas & Water 0 )

Other Business Activities

Motor Sales & Repairs

Construction

Hotels & Restaurants

Agriculture R

Public Admin & Defence es t o

Fishing f e

Forestry co

Mining & Quarrying no

Social & Personal Services m y

Health & Social Work

Education

than twice as much to the EU’s overall productivity 2000 divergences in EU-US productivity growth rates growth rate as did the equivalent US industry, it (see also Graph 9). In fact, these two groups of nevertheless still contributed only 0.07 percentage points industries were responsible for virtually all of the to the EU’s overall total. This is only 1/8 of the acceleration in US productivity over the second half of contribution of the semiconductor industry to overall US the 1990s. It is precisely in these two areas of the productivity growth. This latter industry in fact economy that the EU fares badly relative to the USA contributed nearly a quarter of all US productivity either in terms of the size of the respective industries growth over the period 1996-2000. This basic story is (i.e. small shares of overall EU output) or by having replicated right across the 56 industries. In the 37, relatively low productivity growth rates. In addition, as

mainly traditional and medium tech, industries where shown in the 2003 Review, in terms of explaining EU-

the EU has equalled or outperformed the USA over the US productivity growth differentials over the second

282

second half of the 1990s, apart from communications,, half of the 1990s, it turns out that out of the total of 56 all of the remainder are either low productivity growth industries, just five (semiconductors; communications; industries or do not have a large enough share of EU wholesale trade; retail trade; and financial services) output to alter the EU’s overall productivity dominate the overall labour productivity growth patterns

performance. In addition, for most of these industries and all five are located in the ICT-producing and ICT-

not only are productivity growth rates low but they have using categories. These 5 specific industries contributed

been declining over the course of the 1990s. 283 80 per cent of the US total productivity growth rate over

the 1996-2000 period, compared with a contribution of 3.1.2 Breakdown of 56 industries based on their ICT only 40 per cent in the case of the EU.

concent Regarding the less-intensive ICT-using part of the

Another way of highlighting the EU’s underlying respective economies, the slowdown in the EU’s productivity problem is to classify the 56 industries productivity growth rate in both the “rest of according to their ICT content into ICT producing, manufacturing”, “rest of services” and “rest of intensive ICT-using and less intensive ICT-using economy” categories shown in Table 1 is marked over industries. This has the advantage of firstly isolating the the most recent period. These more traditional industries importance of ICT in driving overall productivity collectively still account for over 70 per cent of EU growth and secondly this three-way ICT breakdown can GDP. The USA has also experienced a slowdown in also be used as a rough proxy for high, medium and low productivity growth in their “rest of manufacturing” productivity industries in the EU and USA as a whole. category, whilst showing only marginal changes in the This breakdown is given in Table 1 which indicates that “rest of services” and the “rest of economy” categories.

the ICT producing manufacturing and intensive ICT- In the case of the USA, however, the globalisationusing private services categories are driving the 1996- related downturn in their more traditional manufacturing industries and the relatively poor contribution from a

282 range of its low to medium-tech service industries was Within the ICT-producing sector, communications is an

industry where the EU has an undoubted advantage over the offset by strong performances elsewhere in the

USA, is characterised by high productivity growth and has a economy. In particular the USA has had good large share of EU output. performances in the newer, more knowledge intensive,

283 An analysis of longer term trends 1980-2000 in productivity manufacturing industries such as semiconductors and in

in the EU and USA is also revealing. At the level of the 3 a number of its intensive ICT-using service industries.

broad sectors of manufacturing, private services and the rest The problem for the EU is that its pattern of

of the economy, the key developments are the following: 1. declining/expanding industries is very different to that in For manufacturing, the EU is on a long run downward trend the USA, with the EU’s trend productivity growth rate

due to its dependence on a range of low to medium being pushed downwards by: technology industries which are increasingly exposed to the

competitive pressures of globalisation. The USA on the Firstly, having a greater share of its production

other hand appears to have arrested its 1980s decline and concentrated in traditional manufacturing sectors where has managed to put itself on a slightly rising trend since the the EU has in the past been strong in global terms but

mid-1990s, driven in large part by its global dominance in where competitive conditions are now becoming more

high technology industries such as semiconductors and difficult due to globalisation.

office machinery. 2. More impressive still has been the

USA’s relative performance in the private services Secondly, the EU is experiencing a further increase in its

industries. Nearly 2/3 of the USA’s overall productivity share of private services, with below average growth growth rate now emanates from services, compared with as rates of labour productivity (at least historically), and

little as 1/6 at the beginning of the 1980s. Over the same with an additional downward shift in productivity in period the EU’s private services industries have been these industries over the 1990s due to the labour market

contributing less and less in absolute terms to overall EU

labour productivity growth. 3. Finally, the EU is doing reforms discussed earlier. It is in these traditional service

better than the USA in the “rest of the economy” sector (i.e. industries, such as hotels and restaurants, transport etc,

primary industries/public services) but even here the trend where the productivity reducing effects of these reforms

is downwards and the contribution of the sector to overall have been felt most.

productivity growth is small.

Table 1: Breakdown of total economy into 3 categories – 2 ICT categories (ICT producing + Intensive ICT- Using) and 1 category of Less Intensive ICT using (i.e. more traditional) Industries

Hourly Labour Productivity Value Added Share Contribution to Total Change in (Average % Change) Hourly Labour Productivity

1991-1995 1996-2000 1991-1995 1996-2000 1991-1995 1996-2000

Total economy (1+2+3) EU 2.3 1.6 1 1 2.3 1.6 USA 1.1 2.3 1 1 1.1 2.3

• 1. 

  Manufacturing sector

EU 3.7 2.6 0.23 0.21 0.9 0.5 USA 3.6 4.6 0.19 0.18 0.7 0.8 1(a) ICT-producing manufacturing industries

EU 9.6 17.1 0.02 0.01 0.2 0.2 USA 16.4 26.0 0.03 0.03 0.4 0.7 1(b) Intensive ICT-using manufacturing industries

EU 2.6 2.0 0.07 0.06 0.2 0.1

USA -0.6 1.4 0.06 0.05 0.0 0.1

1(c) Rest of manufacturing (Less-intensive ICT using) EU 3.6 1.6 0.14 0.13 0.5 0.2 USA 2.6 0.6 0.10 0.11 0.3 0.1

• 2. 

  Private services sector EU 1.9 1.4 0.52 0.54 1.0 0.7 USA 1.0 2.7 0.53 0.54 0.5 1.5

2(a) ICT-producing service industries

EU 4.8 6.8 0.03 0.03 0.2 0.2

USA 2.4 0.8 0.03 0.04 0.1 0.0

2(b) Intensive ICT-using service industries EU 1.8 2.1 0.20 0.21 0.4 0.4 USA 1.6 5.3 0.23 0.25 0.4 1.3

2(c) Rest of Services (Less-intensive ICT using)

EU 1.7 0.2 0.29 0.30 0.5 0.1

USA 0.2 0.3 0.27 0.26 0.1 0.1

• 3. 

  Rest of Economy (Primary industries + public services, less intensive ICT-using)

EU 2.0 1.1 0.25 0.25 0.5 0.3

USA -0.3 -0.1 0.28 0.27 -0.1 0.0

Source: Commission services and GGDC.

Graph 9: Contribution to the total change in trend labour productivity per hour from the ICT-producing manufacturing and intensive ICT-using private services

0.7 1.4 ICT producing

manufacturing EU

0.6 ICT producing 1.2

Intensive ICT using manufacturing USA

0.5 1 private services EU

Intensive ICT using

0.4 private services USA t t 0.8

 p p

% 0.3 % 0.6

0.2 0.4

0.1 0.2

0 0

1980 1984 1988 1992 1996 2000 1980 1984 1988 1992 1996 2000

Source: Commission services and GGDC.

Thirdly, unlike in the case of the USA, the productivity from settled with a large degree of controversy still contributions from the EU’s ICT manufacturing and surrounding the size of the productivity contribution intensive ICT-using service industries cannot make up coming from these specific ICT using industries, with for the losses in its more traditional manufacturing and Gordon (2003) remaining sceptical whilst Stiroh (2002) private services sectors. and O’Mahony et al. (2003) are more optimistic.

Attempts to disentangle ICT production, ICT investment

3.2 Where are the ICT productivity gains coming and ICT spillover effects on labour productivity growth, from? using different methodologies, different levels of

aggregation and different datasets arrive at rather

The analysis in Section 3.1 showed that it was the heterogeneous conclusions. This makes it difficult not superior performance of the USA in ICT-producing only to locate the precise source of the current manufacturing and in ICT-using service industries, such productivity divergence between the USA and Europe as wholesale and retail trade, which was the source of but it also complicates projections on future productivity the diverging EU-US productivity trends since the midgrowth and policy recommendations. This section 1990s. While this is the generally accepted view of reviews the alternative approaches and tries to trace the developments, a number of commentators have been source of the productivity gains in specific ICT using surprised by the fact that the large productivity service industries, such as wholesale and retail trade, at a enhancing effects of ICT have tended to appear in the higher level of disaggregation. However, it also points to hard-to-measure service industries and not in other well a more fundamental problem, namely how to measure measured areas of the economy such as manufacturing. productivity in those service industries which are heavy While the present analysis is not an attempt to rewrite users of ICT technologies.

the conclusions from Section 3.1, it nevertheless tries to

rebalance the messages coming out from this work in Results from international/regional comparison order to impress on policy makers that the EU’s approach . If one looks at international/regional cross productivity problems emanate both from the ICT section data, then ICT production rather than ICT-use, production side as well as from the ICT adoption/using appears to be the dominant source of productivity side. Whilst accepting that these measurement issues in growth. As can be seen from Graphs 10 and 11, there is the services sector are unlikely to be resolved in the near a correlation between productivity growth and ICT future, what must be avoided at all costs is that production in the 1990s but there is little correlation erroneous policy conclusions are drawn given the between productivity growth and ICT investment. uncertainties involved. Consistent with the international data, Daveri and

Mascotto (2002) present evidence across US states

Productivity developments and the difficulties in which suggest that the productivity acceleration mostly disentangling the respective contributions from ICT occurred in those States specialised in the production of production and diffusion (capital deepening and IT goods and services. Based on cross-State econometric spillovers). A primary source of the acceleration in US regressions over the period 1987-2000 they conclude

productivity growth in the 1990s has been the increasing that “… when States where IT production and non IT share of ICT production in the overall output of the US durable manufacturing which are mostly localized are economy allied to the extraordinary TFP gains in this excluded, the remaining States do not exhibit any specific industry. A second channel through which ICT significant acceleration in productivity. In particular, the has impacted on productivity has been through capital association between productivity gains and IT use is deepening, with the falling prices for ICT equipment weak”.

leading to sharp increases in ICT investment rates (i.e. diffusion in the narrow sense of the term). While the economy-wide productivity gains from these two ICT transmission channels are both impressive, what has been missing up until now has been evidence that these large ICT investments have been generating productivity gains in those industries actually using this equipment

(i.e diffusion in the wider sense of the term). Given the

“general purpose technology” characteristics of ICT, one would expect to be witnessing these productivity

“spillover” effects from using the technology, with these

TFP gains representing a third channel via which ICT can impact on aggregate productivity.

From the analysis in 3.1 it would appear that the experience of a small number of intensive-ICT using industries in the USA has provided some evidence that these elusive “spillover” effects are finally emerging.

However, as this section will show, the debate is far

Graph 10: Hourly labour productivity growth and the ICT production share (1995-2000)

6

IE

5

• 0) 

  00 -2 95 4

 ( 19

ity

ct iv AT

du 3 FI

ro

 p PT

ur bo

 la 2 AU JP

SE

rly US

ou CA UK

 h

 in DE NL w th 1

FR IT

ro DK

 G %

0

ES

-1

0 2 4 6 8 10 12 14 16 18 ICT share in value added (1995-2000)

Source: Commission services.

Graph 11: Hourly labour productivity growth and the ICT investment share (1995-2000)

6

IE

5

• 0) 

  00 -2

95 4 ( 19

ity

ct iv AT

du 3

FI

ro

 p PT BE

ur bo SE la

rly 2

AU JP US

CA

 h ou UK

 in DE NL

w th ro 1

FR DK

IT G

%

0

ES

-1

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

ICT investment share (% ) (1995-2000)

Source: Commission services.

Table 2: Growth accounting estimates

Productivity growth differentials (% Points) (1979-1995 versus 1995-2000)

Labour productivity (1+2+3+4+5) USA EU-4 1.25 -0.27

Contributions to Labour Productivity Growth Differential

• 1. 

  Labour quality -0.07 -0.09

• 2. 

  Sectoral employment reallocation effect 0.05 -0.06

• 3. 

  ICT capital deepening : Total economy 0.40 0.19 3a: ICT producing industries 0.04 0.03 3b: ICT using industries 0.29 0.14 -ICT using manufacturing 0.01 0.01 -Wholesale trade 0.05 0.05 -Retail trade 0.01 0.01 -Financial services 0.17 0.02 -Business services 0.05 0.05 3c: Less Intensive ICT using industries 0.07 0.03 4. Non ICT capital deepening :Total economy 0.08 -0.45 5. TFP : Total economy 0.79 0.13 5a: ICT producing industries 0.36 0.24 5b: ICT using industries 0.83 0.02 -ICT using manufacturing 0.06 0.00 -Wholesale trade 0.31 -0.02 -Retail trade 0.28 -0.03 -Financial services 0.27 0.06 -Business services -0.10 0.01 5c: Less Intensive ICT using industries -0.40 -0.13

Note: In terms of non-ICT capital deepening in WT and RT, only small differences exist between the US and the EU-4 (D, F, NL, UK).

Source: Inklaar et al (2003).

Table 3 : Productivity in the US wholesale trade (WT) industry

1988-1995 1995-2000 Difference in

Productivity

Growth Share of Total Contrib. to Productivity Contrib. to

Contribution

Rate Output in WT Prod’y Growth Rate

Share of Total Prod’y (1995-2000 vs.

(%) Growth Rate (%)

Output in WT Growth Rate 1988-1995)

Total

Wholesale 3.04 1.00 3.04 4.03 1.00 4.03 0.99

Durables 4.84 0.46 2.24 5.94 0.48 2.86 0.62

Non Durables 0.32 0.42 0.13 0.75 0.38 0.29 0.15

Electronic

Markets 4.70 0.12 0.57 6.17 0.14 0.84 0.27

Durables decomposition

Motor Vehicles 2.21 0.11 0.23 3.92 0.11 0.44 0.21

Furniture 3.33 0.02 0.05 1.72 0.02 0.03 -0.03

Construction -2.11 0.02 -0.05 -0.36 0.02 -0.01 0.04

Equipment 13.08 0.07 0.97 13.78 0.08 1.13 0.15

Metals -0.31 0.05 -0.02 -0.41 0.04 -0.02 0.00

Electrical &

Electronic 8.81 0.06 0.50 12.98 0.07 0.93 0.43

Hardware 2.70 0.02 0.04 2.50 0.02 0.04 0.00

Machinery 2.75 0.08 0.22 2.97 0.08 0.24 0.02

Miscellaneous 2.43 0.04 0.09 3.24 0.04 0.13 0.04

Note: The productivity measure is real output divided by total hours.

Source: BLS.

Results from the growth accounting approach . There are also two counter arguments to the US Growth accounting exercises on the other hand attribute spillover thesis that should be taken into account:

a sizeable fraction (i.e. about half) of the productivity • Gordon (2003), argues that the revival in RT and

acceleration to the use of ICT. Recent exercises 284 for WT productivity is due to organisational changes

the USA estimate that ICT investment has contributed unrelated to the use of ICT. 286 Microeconomic

about 0.5 percentage points to US productivity growth,

with ICT production contributing another 0.5 percentage evidence provided by Foster et al. (2002) shows that productivity growth is strongly linked with new

points. Other studies, such as Inklaar et al. (2003), establishments, whilst existing establishments do suggest a 0.24 percentage point contribution from ICT not experience a productivity gain. This is despite

production and a 0.19 percentage point contribution

from ICT investment for an EU4 aggregate (see Table the massive investment in ICT which presumably went into both old and new establishments (e.g. bar

2), 285 with higher contributions for the USA code readers have become universal in all retail

(0.40 percentage points for ICT investment and stores). Gordon consequently speculates that 0.36 percentage points for ICT production). productivity gains in the newly built “big box”

As the Inklaar results in Table 2 show, the absolute stores may reflect far more than just the use of difference between the USA and the EU in terms of ICT computers, such as for example size effects, better

capital deepening is largely due to one specific industry, unloading systems, improved storage facilities etc.

namely financial services (FS). If one excludes the FS • A second line of argumentation stresses statistical

industry, it is striking that the contribution of ICT capital

deepening to the change in labour productivity growth problems with measuring productivity in WT and (i.e. ICT diffusion in a narrow sense) has been RT. Volume measures for WT and RT are remarkably similar on both sides of the Atlantic. These calculated using the deflators of the products sold by

figures suggest that the EU is catching up with the USA the WT and RT industries.

287

  This practice could

in terms of the usage/diffusion of ICT in the narrow imply that countries where the share of ICT goods sense of the term (i.e. in terms of the actual purchases of sold to firms and private households is large and ICT investment goods and services by the different where quality improvements are fully taken into

industries). account in the price measures, may have a larger increase in WT and RT productivity simply because

However, what Table 2 also shows is that there are big prices in the basket of goods sold are falling more

differences between the EU and the USA in terms of the strongly. 288 If there has been a genuine productivity

spillover effects from these investments. For example, acceleration, because of the higher use of ICT in while the EU and the USA wholesale trade (WT) and WT and RT, one would expect the productivity retail trade (RT) industries have both made similar gains gains to be evenly distributed across different WT in terms of ICT capital deepening, the USA appears to and RT sub-sectors. If the productivity acceleration have reaped substantially more from the use of this can be traced to specific sub-sectors within the RT capital in the form of much higher TFP gains (i.e. ICT and WT industries with a relatively large exposure diffusion in the broader sense). It must be emphasised to ICT, there is a higher likelihood that the however that these TFP gains occur in a very narrow acceleration could be predominantly due to

segment of the services sector where productivity is measurement issues related to ICT.

difficult to measure. In other better measured areas such • For WT one observes (Table 3) that the productivity

as ICT using manufacturing, Table 2 shows that the

relative TFP gains in the USA are significantly smaller. increases are concentrated in the durables sector,

One can argue that the above analysis provides evidence 286

of positive spillover effects in the USA, i.e. ICT In the case of financial intermediation, it has been argued by

investment is enabling organisational changes in ICT Stiroh (2002), for example, that one should be careful in assuming spillover effects since the productivity effects in

using industries. The fact that the TFP accelerations in financial services have probably been strongly influenced ICT using industries are not observed in the EU could be by the stock market bubble.

due either to adjustment costs (EU is in an earlier stage 287 See Ahmad et al. (2004), Triplett and Bosworth (2000).

of the transition) or it could be the result of 288

institutional/regulatory constraints in specific industries For example, this means that since the volume of computer sales has increased due to improvements in speed and

(e.g. land use regulations/opening hours in WT and RT; capacity, a store which sells the same number of computers

less entry of new establishments/insufficient competition at the end and at the beginning of the 1990s would record a etc) which prevents firms from reaping the full benefits higher productivity growth rate without any change in the

of the new technology in EU countries. organisation (number of employees and hours worked) of

the store. As noted by Triplett and Bosworth, a volume measure for the goods sold by a certain retail sector which

284 Gordon (2003) quoting an unpublished update to Oliner and combines the increase in quality with the growth in the Sichel (2002). number of goods, as is the case with hedonically deflated

285 The countries are France, Germany, the Netherlands and the goods, bears little relationship to the actual activities of the

UK. store, even though it is the appropriate output measure for that specific good.

and within durables in sub-sectors with a high ICT Regarding the results from growth accounting studies concentration such as commercial equipment and and in particular the gains from ICT diffusion, a closer electrical and electronic goods. This desegregation look at the productivity growth acceleration in the WT suggests therefore that the productivity acceleration and RT industries in the USA, which is commonly used is closely linked to the evolution of IT prices, with as evidence in favour of positive TFP effects, also casts this evidence underlining the need for an extremely doubt on the robustness of the ICT careful interpretation of growth accounting studies investment/productivity link. One should bear in mind using data on WT services. that the recent growth accounting studies do not attribute

• In RT (Table 4), two sub-sectors, electronics and the productivity growth acceleration in the USA relative appliance stores and non-store retailers (with a large to the EU to different speeds of ICT investment (i.e. ICT share of ICT equipment dealers) show high diffusion in a narrow sense) but to an increase in TFP in productivity growth rates. However, compared with these industries (ICT diffusion in a broad sense). In WT, WT, the productivity acceleration in RT is more the measurement effect could explain a substantial part

widespread across the sub-sectors. It is difficult of the TFP acceleration in this industry.

290

nevertheless to assess the extent to which the In RT, while the effects are more dispersed, there are productivity acceleration in RT can be traced to the other factors beyond ICT which could account for the use of ICT. A recent paper by Sieling et al. (2001) TFP acceleration in the USA, such as larger store size, traces the productivity improvements in various the increased share of self service markets and the retail sectors to two developments, increased greater entry of new establishments. It is difficult concentration in the industry and ICT investment. In however to establish a clear causal ordering amongst

1987, the 50 largest retail firms accounted for 20 per these latter factors. cent of all sales but by 1997 that proportion had

grown to 26 per cent. In the case of department Finally, the above discussion on ICT diffusion effects stores, labour productivity growth partly reflects must not be seen as contradicting the correct belief that shifts away from conventional stores to discount or ICT spillover effects are making a positive contribution mass merchandising department stores. In 1987 the to labour productivity growth, or that these gains may be latter had a market share of 43 per cent which larger in the USA than in the EU because of increased significantly to 63 per cent in 1997. With institutional/regulatory constraints in a number of the such a shift, productivity gains arise naturally EU’s Member States. It simply suggests that the US because the latter are to a large extent self-service benefits are presently not as high as some commentators

stores. 289 estimate when one correctly accounts for non-ICT drivers of productivity change and measurement issues.

Overall assessment of ICT’s contribution to productivity ICT diffusion in the narrow sense of ICT capital growth. Regarding the international/US regional deepening is clearly contributing strongly to comparison approach, the results on ICT productivity growth, whereas the evidence for ICT production/diffusion effects suggest that ICT production diffusion in the broader sense of large TFP gains in rather than ICT-use is the dominant source of specific ICT-using industries is still open to some productivity growth and that the evidence of debate. While it is undeniable that given the extraordinary spillover effects associated with ICT pervasiveness of ICT in developed economies that there investment is still somewhat questionable. Given the are TFP gains related to the use of this general purpose EU’s relatively small ICT producing sector, especially technology, the present section has simply questioned on the manufacturing side, this raises important the spectacular nature of those gains in a small number

questions as to why the EU has failed to allocate of US service industries. sufficient resources to ICT production.

289 The composition effect is especially visible with food

stores. Grocery stores are by far the largest group within the 290

food stores area. Here superstores and hypermarkets are It is important to keep in mind that the mis-measurement of productivity in the US WT and RT industries is not

replacing conventional grocery stores. In 1988 conventional translated onto the aggregate level. The combined grocery stores accounted for 43 per cent of all consumer productivity improvement in ICT production plus ICT expenditures for food at home; by 1998, that proportion had diffusion is measured correctly, it is only the distribution of fallen to 13 per cent. The productivity improvements with the productivity gains across production and diffusion car dealers can probably be traced to the increased use of which is questionable. In this context, the present analysis computer diagnostic equipment. Productivity growth suggests that the contribution from the ICT production side amongst the non-store retailers (catalogue and mail order to the acceleration in US labour productivity growth has houses) was increased by online sales. Based on annual been underestimated in a lot of the most recent growth

retail trade data by the Census Bureau, E-commerce sales accounting studies. In other words while the size of the EU- accounted for 0.5 per cent of total retail sales in 1999, with US productivity growth gap is still the same, less of this gap 77 per cent of these sales occurring in the non-store retailer is due to the US’s performance in ICT-using industries such industry group. as WT and RT and more of it is located in semiconductors

and other ICT-producing industries.

Table 4 : Productivity in the US retail trade (RT) Industry

1988-1995 1995-2000 Difference in

Productivity Productivity Contribution

Growth Share of

Contrib. to

Prod’y Growth Share of

Contrib. to (1995-2000

Rate Total Output Growth Rate Rate Total Output

Prod’y vs. 1988-

(%) in RT of RT (%) in RT

Growth

Rate of RT 1995)

Total Retail 1.91 1.00 1.91 3.79 1.00 3.79 1.88

Motor

Vehicles 1.15 0.25 0.28 1.74 0.26 0.46 0.18

Furniture 3.06 0.03 0.10 3.84 0.03 0.11 0.01

Electronics &

Appliances 10.81 0.02 0.25 15.46 0.03 0.44 0.19

Building

Materials 1.98 0.09 0.18 3.45 0.09 0.30 0.12

Food &

Beverages -0.86 0.20 -0.17 1.56 0.17 0.26 0.44

Health / Personal 0.74 0.04 0.03 3.61 0.05 0.16 0.13 Care

Petrol 2.16 0.08 0.17 2.76 0.08 0.22 0.05

Clothing 4.21 0.06 0.27 4.62 0.06 0.26 0.00

Sports /

Hobbies 2.80 0.02 0.07 5.66 0.03 0.15 0.08

General

Merchandise 3.18 0.12 0.39 4.96 0.13 0.65 0.26

Misc

Retailers 3.62 0.03 0.10 3.26 0.03 0.11 0.01

Non-Store

Retailers 6.52 0.04 0.27 10.26 0.05 0.51 0.24

Note: The productivity measure is real output divided by total hours.

Source: BLS.

• 4. 

  Enhancing the EU’s productivity The empirical growth literature emphasises knowledge

performance and the creation of knowledge via the investment activities of firms, households and the government in

The analysis in Section 3 on the contribution to overall both R&D and education as significant drivers for productivity growth from ICT production/ICT use has enhancing the level of technology (total factor indicated a more general theme, namely the importance productivity). Last years EU E CONOMY R EVIEW to the EU’s future productivity performance of an indicated the relative potency of knowledge investments ongoing process of structural change aimed at boosting (R&D and education) in determining long run the production and absorption of new, more knowledge productivity growth rates, with a simulation indicating intensive, technologies. The relative success of this that a combination of regulatory reform and a substantial whole process hinges ultimately on the extent to which increase in EU knowledge production could boost EU the Lisbon Strategy’s objective of creating a more potential growth rates by between ½ to ¾ of a knowledge-driven economic model is realised over the percentage point annually over a 5-10 year horizon. coming years and decades and in particular on the ability Investment in education, training and lifelong learning of governments to create an environment in which the are thus essential to the Union’s international EU’s innovation infrastructure can generate the new competitiveness in knowledge-intensive, innovation skills, ideas and products needed to compete sectors, and to sustainable growth and employment.

successfully in the global marketplace. Regarding the USA, the knowledge based economy would appear to be more fully entrenched, with

4.1 The knowledge economy must be a central Graph 12 suggesting that investments in R&D and element education can explain nearly 75 per cent of the US

productivity growth rate over the period 1950-2003 and

With the striking impact of ICT, there has been with the more recent decades accelerating its considerable interest in analysing the effects of dependence on more knowledge intensive forms of investments in knowledge and human capital formation. investment, such as ICT. According to Jones (2002), the

US’s average labour productivity growth rate of 2- industries such as semiconductors now have overall 2¼ per cent over this period could only have been knowledge investment budgets which are equal in size generated via a permanent shift of resources into to their spending on physical investments such as plant knowledge production activities and that without such and machinery and buildings. Furthermore ICT investments US labour productivity growth would have investment itself has not only a larger than average averaged only 1/3 of a percentage point over this period. “knowledge” content, in the form of the software and In other words, over the longer run, these knowledge R&D spending needed to generate it, but has an investments are the key drivers of productivity growth in additional knowledge element in that it is also

advanced economies and our future standards of living complementary to skilled labour.

depend crucially on them. Given ICT’s status as a high productivity growth

Graph 12: Determinants of US labour productivity growth industry and at the same time its potential as a “general

(1950-2003) purpose technology”, inciting productivity growth in

ICT-using industries, it should be a concern to policy R&D Education makers that the USA has established, and is retaining, a

Population Growth Physical Investments large global advantage in this pivotal industry. How

have the Americans achieved such a dominant position and why have other industrialized countries failed so far to catch up to the technology frontier? With the USA continuing to reap enormous gains from its dominance of the global ICT industry, Europe should be looking at those factors which have allowed this industry to flourish in the USA. Box 1 explores the mix between knowledge investment, government support and market structure that lay behind the USA’s success in the ICT area and some of the historical reasons why Europe

stayed behind.

Source: Commission services and Jones (2002). 291 4.2 The USA has a superior innovation model

An important question arising from the analysis in Box 1

ICT is a striking example of the importance of is the extent to which the example of ICT will be knowledge investments. As shown in Section 3, specific replicated in future high tech industries. If this is a knowledge intensive sectors such as ICT are now crucial credible risk then the key issue is whether the EU has to the overall productivity performances of individual specific problems in relation to its innovation countries. ICT in fact is a very good example of the infrastructure (i.e. in terms of the resources devoted to growth in importance of more knowledge intensive innovation, the linkages between the various actors in forms of investment, with its share of total investment the system etc) and whether the USA has specific growing steadily over the last 15-20 years, having now features/framework conditions which make it more reached 1/3 of overall non-residential gross fixed capital likely to be the location of any future breakthroughs in formation in the USA. Within the ICT sector, specific technology. This is a pertinent question if one accepts

the contention of Gordon (2004), amongst others, that the USA’s lead in ICT is not an isolated case. The USA

291 The contributions to productivity growth in the Jones holds a comparative or absolute advantage not only in

analysis are calculated by multiplying historical changes computer hardware, but more broadly in software and in (from 1950 to 2003) of R&D, education and capital shares other general purpose technologies, like its initial with their respective output elasticities. The relatively small leadership in the electricity industry and in its

contribution of physical capital to growth is due to the fact exploitation of the internal combustion engine (Gordon that unlike the shares of R&D and educational attainment, 2004). While some comfort can be taken from the EU’s

the share of physical capital has not changed much over the ability in the past to catch up with the USA in the latter last 50 years. This is typical of a country (such as the USA) technologies, this did not occur without a large

at the technology frontier, with steady state physical

investment levels. For countries however in the “catchingrestructuring and refocusing of EU industry. In addition,

up” phase of their economic development the productivity the wider issue is why is it that the USA seems to be contribution attributed to physical investments would be systematically better in creating and exploiting new

substantially larger. The contribution of population to (general purpose) technologies? This requires productivity growth in the Jones analysis comes from an broadening the discussion beyond ICT to consider why

increasing returns to scale effect in production. The basic the USA seems to have a better innovation capacity than

point of the graph is that since the EU-15 is now close to the EU.

the technology frontier, any additional productivity gains

over the coming decades are more likely to be generated National innovation systems and national innovation from a boost to knowledge investments rather than from capacity. While traditional growth theories explain

changes to our present physical investment to GDP ratio.

differences in growth by the expansion in inputs, such as • Human resources. The USA invests a far larger

capital and labour, and by the catching-up of countries share of GDP in higher education than the EU with lower productivity, modern theories emphasize (2.7 per cent compared to 1.1 per cent, with a large research inputs and human capital as the key drivers for proportion of the increase accounted by the private long-run growth. They stress not only the importance of sector). The USA employs nearly 300000 more “own” innovation but also the capacity to imitate and to researchers compared with the EU, with the vast absorb externally available know-how. Institutional majority of the overall total (over 80 per cent) factors and framework conditions are seen as an employed in the business sector, compared with less

important part of the “innovative system” in which than 50 per cent in the EU.

innovative firms operate. • Basic R&D expenditure differences at the economy

Using the macroeconomic insights from neo-classical wide level. A persistent and growing differential and endogeneous growth theory, as well as the ideas exists in the amount of resources devoted to R&D in from the literature on “National Innovation Systems”, the EU and the USA both in terms of the overall applied economic theorists have synthesized what research intensity of the respective economies determines an economy’s “national innovation (1.9 per cent versus 2.8 per cent of GDP) and in capacity” defined as the ability of a nation to not only absolute amounts. To put the respective research produce new ideas but also to commercialize a flow of efforts into context, the absolute gap in the volume

innovative technologies over the longer term. 292 From of research is roughly $110 billion. If one widens

this perspective a range of factors are deemed to be the definition of the knowledge economy to also important for an effective innovation effort: include expenditure on the higher education sector,

• Overall innovation infrastructure. A sufficiently the USA is investing well over $200 billion more developed ‘supply’ side of R&D (as reflected in the annually on its knowledge economy compared with

amount of R&D carried out or the number of skilled the EU. researchers) is a necessary but insufficient condition • Basic R&D expenditure differences at the sectoral

for successful innovation.

• Essential framework conditions/flanking policies. level. Compared with the EU, a much larger share of US R&D is carried out by the business as opposed

Broader framework conditions are important as to the government sector. Within the business sector well, including a sufficient ‘demand’ for innovation the USA spends substantially more on services to reward successful innovators. This requires compared with the EU, especially in the “Computer sophisticated lead users willing to pay for and Related Activities” area. Of the total US R&D innovations, effective intellectual property rights effort, roughly 1/3 is devoted to services and 2/3 to (IPR) schemes, a favourable macro-economic manufacturing. Regarding the EU, its research environment and effective competition in output efforts continue to be overwhelmingly focussed on

markets.

• Interconnectedness of the overall innovation system. the manufacturing sector which presently accounts for around 85 per cent of its overall business sector

Perhaps the most critical element in the framework R&D spending.

is the interconnectedness of the agents in the

system, linking the common innovation • Basic R&D expenditure differences at the

infrastructure to specific technology clusters. industry level (technology specific R&D). Since Through networking amongst firms, researchers and no reliable comparative figures exist for a governments, the supply of new ideas diffuses breakdown of service sector R&D activities, the throughout the economy. This requires good industry level comparison is restricted to the industry-science links and well functioning capital manufacturing sector. Of the 27 industries which and labour markets, such that the human and make up the manufacturing sector in the present financial capital inputs get allocated to their most study (see Graph 8), only 8 can be regarded as efficient applications. having an above average R&D intensity and

therefore classified as high technology industries. The details regarding these 8 industries and their

4.2.1 Overview of innovation infrastructures in aggregation into the two categories of ICT and nonthe

EU and the USA ICT is given in Tables 5 and 6 below, with some supplementary information given in Graph 13.

The basic differences in the overall innovation infrastructures of the EU and the USA can be summarised as follows.

292 See, for example, Furman, Porter and Stern (2000).

Table 5: Shares of some specific R&D intensive manufacturing industries in the total R&D spending of the

manufacturing sector (period average 1996-1999)

EU USA

% Share of total Actual Expenditure % Share of total Actual Expenditure Manufacturing R&D (Current PPP$) Manufacturing R&D (Current PPP$)

• 1. 

  Chemicals 17.4 13583 9.2 11307

• 2. 

  Mechanical engineering 6.8 5314 2.8 3441

• 3. 

  Office machinery 2.4 1858 6.3 7739

• 4. 

  Electrical Machinery 3.0 2338 1.9 2384

• 5. 

  Semiconductors / 10.6 8307 9.0 11067 Communications

• 6. 

  Instruments 4.3 3331 6.9 8502

• 7. 

  Motor vehicles 12.9 10039 7.3 8985

• 8. 

  Aircraft and spacecraft 6.7 5225 7.0 8646

Total High-Technology 64.1 49995 50.6 62072 Industries

Total Manufacturing 100 78048 100 122717

Source: OECD Anberd Databank.

Table 6: Splitting high-technology manufacturing R&D spending into ICT and non-ICT

EU USA Specialisation Gap Indicator for Actual Expenditure Actual Expenditure R&D Spending

(Current PPP$) (Current PPP$) EU-USA* 1991-95 1996-99 1991-95 1996-99 1991-95 1996-99

Total High

Technology 44488 49995 52033 62072 0.855 0.805

ICT 11849 13496 20125 27308 0.589 0.494

Non- ICT 32639 36500 31918 34764 1.023 1.050

Note: * This gap is calculated by dividing the EU figure with that of the USA, with a value of less than 1 indicating that the USA’s R&D expenditure is relatively more concentrated / specialised in a particular sector or industry, with a value in excess of 1 showing the same for the EU. A value of around 1 suggests broad balance.

Source: Commission services.

The key points from Tables 5, 6 and Graph 13 are as of the overall growth rate effect, that the ICT follows: manufacturing industry is remarkable and possibly

Firstly, in terms of overall manufacturing R&D, it is unique. From this perspective, it is disturbing to note striking how concentrated the EU’s efforts are, with 2/3 from Table 6 that the USA totally dominates the EU in of its overall R&D spending focussed on the high terms of its research efforts in this area and that this technology sector as a whole compared with 50 per cent dominance has continued to grow over time.

in the USA. In terms of absolute expenditures, however, Thirdly, given that the productivity enhancing the USA still retains a sizeable advantage over the EU in characteristics of ICT were already known in the first terms of overall R&D spending on high technology half of the 1990s, what is striking from Table 6 is the industries (i.e. of the order of $ 45 billion annually). fact that the USA’s dominance in ICT manufacturing

Secondly, while there is no doubt that the EU is was not seriously challenged over the second half of the focussed on high technology industries (i.e. those 1990s. In fact the USA increased its advantage industries with above average R&D intensities), there is significantly over this period, with an absolute increase a serious doubt that they are focussed on the best in its ICT R&D investments which was roughly 4½ industries from a high productivity growth rate times greater than the equivalent increase for the EU. perspective. The ideal combination would appear to be Over this period, the EU instead extended its dominance industries, such as ICT, which combine both high in the relatively low productivity growth, non-ICT, technology and high productivity growth rate manufacturing industries which contributed only 1/20

th

characteristics. While one cannot exclude the possibility of the productivity gains achieved by the USA from that there have been other similar “dual” technologies in ICT. This is an important point to bear in mind in the the past, it is fairly safe to conclude, in terms of the size context of the Lisbon strategy’s objective of an increase

in the EU’s R&D intensity from 2 per cent to 3 per cent

over the coming years. On the basis of the above Graph 13: R&D expenditures by the top 300 international firms by sector: EU 15- USA (2002)

analysis, if this target had been set in 1990 for attainment in 2000, without any specific sectoral focus, Software + and if the EU continued to invest heavily in the more Computer Services traditional high technology industries such as cars and

chemicals (which it actually did do), it would have Telecommunications GAP

gained relatively little in terms of closing the Aerospace + USA productivity gap with the USA. A sizeable productivity Defence EU effect from the additional expenditure would have

necessitated a shift in focus to the newer, high Chemicals technology, industries such as ICT. IT Hardware +

Finally, Graph 13 confirms the broad trends from Electronic

Tables 5 and 6, indicating on the basis of the R&D Pharma + Biotech

expenditures of the top 300 international firms in each

individual sector that while the EU may be dominant in Cars

low productivity growth, high technology, industries

such as cars and chemicals, the USA is dominant in the -20 -10 0 10 20 30 40 50

high productivity/high technology areas of IT hardware and electronics. This US dominance is already Source: Commission services. worryingly been extended to software and computer services. Graph 13 further allows one to single out pharmaceuticals and biotechnology from the rest of the

chemicals sector. It is again a source of concern to see in 4.2.2 Can the superiority of the US innovation the pharma/biotech part that the USA is leading in terms infrastructure explain EU-US productivity of R&D expenditures. Within the chemicals sector, an growth differentials in the manufacturing area of traditional strength for the EU, pharma/biotech is sector?

arguably the key productivity growth component for the

future. Beyond the higher expenditures on manufacturing R&D,

The most significant issue posed by the above analysis is particularly in the high productivity growth ICT not so much the differences in the amounts of resources industries, various links can be made at the industry devoted to the knowledge production sector, but the level which can contribute towards explaining the EU’s systemic failure to refocus its R&D activities over productivity gap through R&D expenditures. In this the 1990s, firstly on established high productivity context, the two key issues are:

growth industries such as ICT, and secondly on • Firstly, has the US economy specialized more in potentially high productivity growth industries in the specific high technology industries which are also pharma/biotech area and in a number of service the high-productivity growth areas – in other words industries (software and computer related services). are the EU-US productivity growth differentials

Building on the analysis of differences in terms of the linked to industry specialisation?

innovation infrastructure, an attempt is made in 4.2.2 to

assess the extent to which differences in EU-US • Secondly, within each industry, beyond the effect of

productivity growth rates can be linked to differences in differences in spending levels, is the USA getting a the innovation capacities of both areas. While this is not higher rate of productivity growth from its R&D possible at the total economy level, a tentative spending i.e. a better leverage out of its R&D into assessment can be made for the manufacturing sector on productivity growth which can roughly be equated the basis of the analysis in the present section of R&D with a higher rate of return on its R&D spending in this sector. Since this analysis underlined expenditures.

the dominance of the USA’s innovation model, it constitutes a prime candidate for explaining EU-US While the ECFIN analysis of these two issues, and of differences in the productivity growth performances of their role in explaining the link between R&D spending their respective manufacturing industries. and EU-US productivity growth differentials, is still in

its infancy, a number of interesting conclusions are

already emerging:

Table 7: Contribution to growth in productivity from high technology manufacturing industries (% Points)*

EU USA Specialisation Gap Indicator *

1991-1995 1996-2000 1991-1995 1996-2000 1991-95 1996-00

Total High Technology

Manufacturing Sector 0.399 0.352 0.830 0.849 0.48 0.41

• - 

  ICT 0.137 0.219 0.603 0.802 0.23 0.27

• - 

  Non-ICT 0.262 0.132 0.227 0.047 1.15 2.81

Note: * Calculated by dividing the EU figure by the US figure for the respective periods, with a value of less than 1 indicating that the US’s is relatively more specialised in a particular sector or industry, with a value in excess of 1 showing the same for the EU. A value of around 1 suggests broad balance.

Source: Commission services.

Table 8: Comparison of EU-US differences in R&D rates of return

EU-US Gap in R&D Spending EU-US Gap in Productivity Growth Rates

1991-1995 1996-1999 1991-1995 1996-2000 Total High Technology

Manufacturing Sector 0.85 0.81 0.48 0.41

• - 

  ICT 0.59 0.49 0.23 0.27 - Non-ICT 1.02 1.05 1.15 2.81

Source: Commission services.

Firstly, as shown earlier in Section 3, there are large EU- This is entirely due to the ICT high-technology US differences in terms of specialisation (i.e. differences industries in the size of specific industries as a share of total

output). Within the overall high technology sector, the Overall assessment of the EU’s innovation USA is especially concentrated/specialized in ICT infrastructure in the high technology manufacturing manufacturing, with nearly 15 per cent of total US sector. Taking all the caveats in mind of the basic manufacturing output coming from these ICT industries analysis presented, the evidence for the manufacturing compared with only 6 per cent for the EU. In the nonsector supports the importance of differences in the ICT area, there are no differences between the EU and innovation system in explaining diverging EU-US the USA, with the high technology industries productivity growth rates. Within high-technology representing around 34 per cent of the overall industries, the specific role of ICT cannot be ignored.

manufacturing output of both areas. ICT producing industries have the highest productivity growth rates in all of manufacturing (in fact in the total

Secondly, since amongst the high technology industries economy). The USA is more specialized in these ICT as a whole, it is ICT which has been shown to have the industries as compared to other high-tech sectors; it has highest opportunity for productivity growth, it is not a higher productivity growth rate in these sectors; surprising to find in Table 6 that the US’s specialization spends more in total on R&D; and gets a higher rate of in these industries, and their realization of a high return out of its R&D investments. For the non-ICT productivity performance, is a key factor in explaining high-tech industries, the picture is less devastating for overall EU-US productivity growth differences. In fact the EU, particularly in the second part of the nineties. the ICT industry totally explains the better performance There is no difference in specialization in these of the US’s manufacturing sector over the 1990s industries, nor a productivity disadvantage. The gap in compared with that of the EU’s and it contributes nearly total expenditures on R&D is also minimal. four times more to the US’s, economy-wide, Unfortunately, however, these industries have far less productivity growth rate compared with the equivalent scope for productivity growth than ICT. sector in the EU. Not only is the productivity gap

substantial, there is no evidence of significant catching 4.3 Reforming the EU’s Innovation Capacity

up after 1995. In terms of policy prescriptions, what do the results in

Thirdly, Table 7 presents evidence of EU-US 4.2 suggest for the innovation capacity of the EU differences regarding “rates of return” from R&D relative to the USA. While differences in the amount of investments. For specific high-tech manufacturing resources committed are large, it is abundantly clear that industries, the gap in productivity growth is in addition to much larger investments in R&D both by considerably higher than the gap in R&D spending, the public and the private sector (i.e. the basic implying a lower rate of return from R&D spending in innovation infrastructure), there are also other high-tech industries in the EU compared with the USA. characteristics of the US innovation system which

explain its ability to focus on the high productivity and with it a large proportion of the EU’s future

growth areas and to gain a higher rate of return from its productivity potential.

knowledge investments. It is these latter features which

determine its superior overall innovation capacity and Given the above worrying EU trends, the fact that

which need to be taken into account in assessing the the USA’s comparative advantage in this area of relative effectiveness of both systems. These features knowledge production appears to be becoming more relate to the USA’s established capacity to link its entrenched, and that the new knowledge industries are common innovation infrastructure to technology specific increasingly driving economy-wide productivity trends, know-how and the generally more favourable the calls for reform at the EU level are becoming more environment for innovation in the USA compared to the urgent. What can policy makers do to address the EU’s EU. Gordon (2004) identifies a better connectedness of innovation weaknesses? The most important point to science and industry with an openly competitive system stress is that R&D spending is only one of the key of private and public universities and government elements of a country’s knowledge production system – subsidies to universities through peer-reviewed research the present analysis has underlined that it is the overall grants, which result in a higher quality of the research package of elements which matters. Reforms are base. Other important framework conditions present in particularly needed in terms of entry and exit rules (to the USA are the advantage of a large, unified market allow, for example, new innovative firms to come unencumbered by differences in language, customs and through the system and challenge the incumbents) and in standards; a clearer and stronger US Intellectual the overall business environment (to improve the “rates Property Rights system; more flexible financial markets, of return” of any additional R&D investments which making available venture capital finance to innovating may be linked to the Lisbon 3 per cent target). This will firms; and more flexible labour markets, affecting both require getting the framework conditions right; internal migration and the international immigration of improving the overall interconnectedness of the

highly skilled people. innovation system; and ensuring that the common innovation inputs are better aligned on specific

The importance of the above features to an effective technology clusters, where the EU’s production structure

innovation process may help in explaining a number of displays a specialisation.

specific worrying trends which have emerged in the EU

over the 1990s which are suggestive of the need for a

radical overhaul of it’s knowledge creation system.

These include the failure in the 1990s, as stressed

earlier, to re-orientate its R&D activities towards the

new, high technology, ICT industries; the increasing

proportion of R&D by EU firms which is being done

outside the EU (over 40 per cent); the large and growing

brain drain from the EU to the USA on the research side

(at present, twice as many EU researchers move to work

in the USA compared with the opposite inward flows

and, in stock terms, DG Research estimates that roughly

40 per cent of US R&D is carried out by EU-trained

scientists); and finally the USA’s rapidly expanding

share of internationally mobile R&D expenditures. This

latter point is an important new risk factor given the

evidence that such flows are increasing rapidly and that

the relative quality of third level education systems is a

key locational determinant for such mobile R&D

flows. 293 In overall terms, it is reasonable to conclude

that without EU reforms to its innovation system, the present haemorrhaging of R&D spending and of research talent from the Member States is set to continue

293 In this context, it is no accident that the top US regions in

terms of knowledge production owe much of their success to the presence of world class educational establishments such as San Francisco’s Stanford / Berkeley and Boston’s Harvard / MIT. These latter universities have been the key driving forces which have propelled the San Francisco and Boston metropolitan areas respectively into the top two positions in most global knowledge competitiveness, benchmarking, exercises.

Box 1: History of ICT

The invention of the point contact transistor at AT&T’s Bell Laboratories in the USA in 1947 marked the beginning of a new technology, semiconductors, which has paved the way for the information age. An important feature of this technology is the speed with which microprocessors become more powerful, leading to a widening in the number of ICT applications. Semiconductors were initially used for hearing aids and in mainframe computers. In the 1950s computers largely replaced mechanical calculators. They were seen as being good in performing complicated and lengthy sets of arithmetical operations. The first leading edge applications were military. The Korean War won IBM the first contract to deliver a computer. In the late 1950s and early 1960s computers began to be used for simple calculations in civilian government agencies (e.g. the Census Bureau) and by the human resource departments of large corporations. The next generation of computers was used for storing and releasing data in real time. This was important for airline reservation processing, insurance companies and for inventory control. The computer surpassed the stage where it was only used as a calculator and became an organizing device. In a further step, the invention of the spreadsheet computerized white collar work in the 1980s. But the domain of computer usage has been widening further. Robots in manufacturing and scanner based retailing are transforming production and distribution processes. The internet, which connects computers all over the world, further transforms the way business is conducted.In terms of pure numbers, nominal spending on ICT in the USA rose from about 1 per cent of GDP in the 1960s to about 2 per cent in 1980. The share increased further to 3 per cent in 1990 and has reached about 5 per cent to 6 per cent of GDP in the year 2000. Given the other distinguishing feature of this technology, namely the rapid speed of technical progress in the production of semiconductors, this sector now shows up in the aggregate productivity statistics of those countries which have managed to have a sizeable ICT production sector.The semiconductor and computer industry has been a USA dominated industry since the end of the second World War and the USA has not given away the lead to other countries though they have faced severe challenges, especially from Japan. The ICT industry has some special features which poses specific challenges for government-industry interactions. The most important characteristics are:

• - 

  The semiconductor/computer industry is the high tech industry par excellence. It undertakes large amounts of knowledge investment, with R&D shares exceeding hugely the average shares of most other manufacturing industries.

• - 

  The sunk cost nature of R&D requires careful thinking about competition and industrial policy regarding the best strategy of combining large R&D efforts with a competitive environment.

• - 

  The industry also relies on a stream of well educated scientists and engineers as well as on the basic research undertaken in national research labs and universities.

• - 

  Since ICT has become a general purpose technology (GPT), with the ability to influence the productivity growth rates of ICT- using industries, it is also therefore of strategic interest since the products sold by this industry shape process innovations in other manufacturing and service sectors.

How have the Americans achieved such a dominant position and why have other industrialized countries failed to catch up to the technology frontier? What was the mix between knowledge investment, government support and market structure that created the success in the USA and what were the reasons why Europe stayed behind. The history of ICT in the USA, Japan and Europe will at least provide some tentative answers to these questions. In the USA, early computer technology had a distinctly military focus (Brock, 2003). Japan and Europe in contrast tried to reduce the substantial lead of US companies in commercial markets. However both regions pursued rather different strategies. Japanese technology policy was based on a system of cooperation and competition amongst diverse groups of firms. In Europe, all bets were usually placed on a single ‘national champion’, the beneficiary of a steady diet of financial subsidies and preferential procurement policies’ (Flamm, 1987). An important factor in the development of the ICT industry has been the level of knowledge investments. The size of R&D spending and government funding in IT shows marked differences between countries. In the early 1970s, total R&D spending in the USA’s computer industry was about 5 to 6 times larger than the combined efforts in Japan, France and the UK (Flamm, 1987). In the 1960s and early 1970s about 1/3 of all R&D spending in the USA was publicly financed, while the French and UK share ranged between 10 per cent and 15 per cent. The Japanese share of public funding was in between. Thus in contrast to the popular view which saw the USA as the least interventionist amongst the major industrial countries, it must be acknowledged that the USA was strongly supporting industrial investment in technology directly in the formative years of the ICT industry.

Table 9: Determinants of ICT

USA Japan EU

R&D It is undeniable that in the 1950s, when commercial There were four major players in Japan: 1) Ministry of trade and The European countries responded to the applications of the new technology were hardly imaginable, industry (MITI) plus its technical arm, the Electrotechnical increasing commercial success of US computer that the US government was supporting all major computer Laboratory (ETL); 2) Nippon Telephone and Telegraph (NTT); 3) companies in the 1960s (mainly IBM) with the technology projects. Military projects (such as Whirlwind Ministry of Education (national universities); 4) Industry. The creation and support of national champions. and SAGE) filled assembly lines and helped to train division of labour between these four players and how it has Small firms were encouraged to merge in order electronic engineers. Purchases of the US government (not changed over time can briefly be described as follows: to exploit scale economies in research and all was military, the Census Bureau also ordered production. The national champions in the computers) amounted to more than 50 per cent of total sales -1950 S : Research is carried out in 1) to 3). Industry adopted the 1960s were ICL in the UK, CII in France and in the 1950s. If one includes defense contractors, about designs which had been developed in the various laboratories. Siemens in Germany. Research subsidies 70 per cent of the computer bill was paid directly or -1960 S : Erection of trade barriers and the price for foreign started relatively late in the 1960s and were not indirectly by the taxpayer. Due to the strong increase in the admission was access to important technology. The first research very generous. Especially striking is the UK use of computers for administrative purposes in large cooperation amongst Japanese manufacturers (NEC, Hitachi, example. In 1950, the UK’s computer corporations, the government’s share fell rapidly in the Fujitsu) was started and a period of joint government–industry technology matched or even surpassed that of 1960s. However, government purchases remained the cooperation during all phases of research and early development the USA. However, within a decade the lack of largest single factor in sales of new, leading-edge, began. financial and technical resources led to a machines (about 40 per cent until the mid 1980s). The US -1970 S : Dual crisis: 1) IBM system 370 forces other US companies decline. The government tried to stop this government’s share of funding for computer related R&D (which were cooperating with Japanese companies) to exit the decline by creating ICL via mergers of smaller was 75 per cent in the 1950s, 50 per cent in the mid 1960s, market. 2) Commitment to open Japanese computer market. MITI producers.

and 15 per cent in the late 1970s. With the military build up prescriptions: 1) Increase research funding: + 60 per cent in 1973. in the 1980s the share increased again to more than 20 per 2) Consolidate research amongst private firms and promote the In the 1980s larger efforts were undertaken to cent survival of the largest and technologically fittest firms. coordinate research at a European level. The The USA spends vastly more on ICT research than any -1980 S : Steep increase in private R&D funding: Like in the USA in Esprit program amounted to about 1.5 Bio $ other country in the world and it started much earlier. In the the 1960s, with the large expansion in the commercial sales of over a five year period starting in 1984 (50 per 1950s, many of the American computers built were Japanese computers, the role of private R&D increased (while cent was paid for by the European Community government financed machines later adapted to the public funding stayed at a constant nominal level). and the remainder was paid by participating commercial market. The Air force’s SAGE project alone firms). Numerous European firms were accounted for billions of dollars in development funds cooperating in what was labeled precompared to only the tens of millions spent by governments competitive research. A similar program in in other countries. telecom research was RACE.

Procurement In the 1950s purchases of computers by the US government In the 1970s, nearly 100 per cent of all the computer purchases of Like in other countries, government Policies and defense contractors amounted to about 70 per cent of the government were Japanese. procurement was used to provide markets for

total sales. In the 1970s the government’s share declined to national champions. However the share of only about 5 per cent. national producers never reached the same levels as in the USA and Japan. This is likely to have been the result of dismal technical

performance and not of policy.

USA Japan EU Market Structure Vigorous antitrust suits instituted during the 1950s played Joint research (in order to avoid duplication and increase Because the national champions model was

some role in the rapid diffusion of semiconductor and productivity) but competition in downstream applications and adopted early on, antitrust was not an important

computer technology from Bell Telephone Laboratories commercialization e.g. in the 1970s the three groups of Japanese issue in the computer industry.

and IBM. Entry and exit into the market played a big role computer producers (Fujitsu-Hitachi, Mitsubishi-Oki and NEC-

in adapting technologies for commercial use. In the early Toshiba) shared development costs but remained in direct

1970s, the top 5 companies were IBM, Texas Instruments, competition. MITI coordinated the research cooperation. The

Motorola, Western Electric and Fairchild. None of them government also carefully controlled access to the Japanese market,

was a leader in the 1950s. Top ranked companies in the with MITI attempting to induce US producers to transfer computer

1990s such as Microsoft and Intel were not around in the technology to Japanese manufacturers.

1970s. However, the US government has also responded to

foreign competition by allowing for more cooperation

amongst US companies in R&D. In 1984 the Joint research

and Development Act was passed by Congress, which

encouraged firms to undertake cooperative research (this

was likely to have been provoked by Japan’s 5 th Generation

Computer Research Program in 1981). Furthermore, in response to a loss in the USA’s market share in the semiconductor industry (The Japanese market share exceeded that of the USA for the first time in 1986), SEMATECH, a joint research effort of the semiconductor industry, was initiated and supported by the US government. The defense department contributed about 100 Mio $ per year, about 50 per cent of the total budget (the project was expected to end in 1997). Reflecting concerns about the national security implications of dependence on foreign sources for the supply of semiconductors, its goal was mainly to improve US semiconductor production technology. That the US government regarded SEMATECH as a national project can also be seen by the total entry restrictions for the US affiliates of foreign companies. The USA also emphasizes industry-university research centers. They have been initiated by the government but also by private companies. (Flamm 1987)

• 5. 

  Summary of key findings factors and framework conditions which determine a

country’s overall innovation system and ultimately its

Stuctural nature of the EU’s productivity downturn is success in producing and absorbing the latest, leading confirmed. The overriding conclusion from the analysis edge, technologies. The chapter argues strongly that in this chapter is that the former EU-15 group of healthy knowledge production and absorption processes countries have a structural productivity problem, with are mutually supportive elements of any successful long this problem mainly located in the four large euro-area run productivity strategy. Evidence is presented which Member States which presently account for close to suggests that the USA’s overall innovation system is 80 per cent of the euro area’s overall output (2/3 of EU- superior to that of the EU’s, both in terms of the quality 15). This interpretation of recent productivity trends and funding of its knowledge sector and the more differs from that of respected commentators such as favourable framework conditions prevailing. This Olivier Blanchard and the IMF which suggest that the system has facilitated a substantial re-structuring of the present productivity downturn is temporary, linked to US economy since the early 1990s towards a range of the substantial labour market reforms enacted in many knowledge intensive, high productivity, growth of the EU’s Member States throughout the 1990s. In our industries which have compensated for the relatively view these reforms can only explain a small proportion poor productivity performance of its more traditional of the deterioration in EU productivity since 1995, with industries. The inadequacies of the EU’s overall the bulk of the decline due to the EU’s outdated and innovation system have, in contrast, been cruelly inflexible industrial structure which has been slow to exposed over the same period. Despite the growing adapt to the intensifying pressures of globalisation and evidence of the importance of high productivity growth rapid technological change. The EU’s productivity industries such as ICT, the EU continued to focus its problems reflect the combined effect of an excessive R&D investments throughout the 1990s on relatively focus on low and medium-technology industries (with low productivity growth areas such as cars and declining productivity growth rates and a globalisationchemicals. The repeated ability of the US innovation induced contraction in investment levels); an inability to system to direct resources towards the newer, high seriously challenge the USA’s dominance in large areas technology (and often high productivity growth), of the ICT industry, as reflected in the relatively small industries is a reflection of the quality of the size of its ICT production industry; and finally, its interrelationships between the different actors in its apparent slowness in reaping the productivity enhancing innovation system and of an economic and regulatory benefits of ICT in a range of ICT-using industries, framework which has the capacity to transform although measurement issues severely complicate an excellence in knowledge creation into a globally

assessment of the gains from ICT diffusion. 294 The competitive industrial structure. chapter also points to the worrying evidence that the The systemic inadequacies of the EU’s innovation USA is extending its dominance in ICT production to a system are highlighted by the experience of the ICT range of new, high technology, areas in pharmaceuticals, industry, with the history of this industry suggesting that biotechnology and computer-related services. a “national champions” strategy in high technology The Post-1995 Differences in EU-US productivity industries is highly problematic. The systemic nature of patterns are fundamentally driven by the USA’s the EU’s productivity problems is highlighted by an superiority in terms of its capacity to produce and analysis of the ICT industry, where a wide range of absorb new technologies, most notably in the case of factors are shown to have contributed to the USA’s ICT. The contrasting productivity experiences of the EU global dominance. These factors include focussed R&D and the USA over the post 1995 period have their origin activities; world class research and teaching in the knowledge production sectors of the EU and US establishments; defense procurement contracts which economies and in a complex range of institutional nurtured the ICT industry (on the demand side) in its

incubation phase in the 1950s and 1960s; and the unique combination of financing mechanisms and a highly

294 competitive domestic marketplace which brought the The paper stresses the need for a critical assessment of the

respective roles of ICT production and ICT diffusion in ICT industry from the knowledge creation phase to the explaining EU-USA productivity growth differentials. It critical diffusion/mass market phase. The history of the

suggests that due to measurement issues a higher proportion ICT industry also suggests that a “national champions” of the post-1995 acceleration in US productivity should be strategy in high technology industries is doomed to

linked with the production of ICT than is commonly failure, with a number of interesting questions emerging

assumed. In terms of diffusion, it stresses that ICT capital from the analysis as to the type of optimal competition deepening (diffusion in the narrow sense) is contributing policy which should be pursued for high technology strongly to US productivity growth but that the evidence for industries. The chapter highlights in particular the large

large TFP gains in specific ICT-using industries (diffusion price which Europe has paid for its “national

in the broader sense), such as wholesale and retail trade,

was still questionable. These latter gains are perhaps more champions” policy in the ICT industry back in the 1960s modest when proper account is taken of measurement issues and 1970s, which contrasted sharply with the strategies

and of the role of a number of important non-ICT adopted by Japan and the USA. In addition, if one looks

productivity drivers in these specific industries. to the future, and given the changes which have

occurred over recent decades, it is safe to conclude that in the form of financial support for human capital

the case for such a “national champions” policy is development 295 and for the public innovation system

becoming more and more tenuous as the new industries and, more importantly, indirectly in terms of shaping the of the future will increasingly need to draw on an EU, or macroeconomic fundamentals (low and stable inflation; even a global, knowledge/talent pool. moderate tax burdens on labour and capital; trade

Policy conclusions openness) and providing adequate framework conditions for the private sector to enhance productivity via well

In terms of policy, the chapter stresses that the EU’s functioning product, labour and capital markets.

296

  The

innovation system needs to be fundamentally reformed private sector for its part is the ultimate source of if the EU is to make a decisive shift towards realising productivity growth in an economy, with its overall the vision of a successful innovation-based, economic performance determined by the success of the public model, the broad features of which have been laid out in sector’s policies in creating a competitive, dynamic, the Lisbon 2010 agenda. Lisbon is in effect a business environment and by its own ability to use its recognition of the importance of such a model to the labour and capital resources to create an industrial

EU’s long run economic prospects and of the key role structure capable of competing successfully in both the which it must play in responding to the challenges of domestic and global marketplaces. globalisation and ageing. Creating a successful Industry-specific framework conditions need to be taken knowledge-based economy involves both enhancing the into account due to the complicated relationship

EU’s capacity to produce and commercialize a flow of between competition and innovation. While competition world class innovative technologies and creating an is a crucial determinant of productivity growth, acting as environment conducive to the imitation and absorption a powerful incentive for firms to continuously enhance of externally available know-how. The success of such a their underlying performance via process or product model will be determined not so much by a massive innovations, there is nevertheless a need to recognise the increase in the amount of financial resources devoted to complicated non-linear relationship between innovation knowledge production (i.e. increased spending on R&D and competition. This relationship may in certain cases and higher education) but by an acceptance of the need favour oligopolistic competition between a few large to improve linkages in the innovation system and to firms in some industries or stronger competition among make painful changes in many areas of the EU’s many small players in others, as the optimal market economic and regulatory environment. More specifically structure for boosting the innovation process in the the present study stresses the following: respective industries. Due to this non-linear relationship,

The systemic nature of the innovation process needs to it is incumbent on policy makers to take industrybe recognised and the quality of the interrelationships specific circumstances into account when assessing the between the different actors in the system needs to be precise link between competition and productivity. dramatically improved. Policy makers need to recognise Product market conditions (eg possibilities for product

that the different players in the innovation system,

public research institutes; third level education 295 Investment in education, training and lifelong learning are

establishments; SME’s and large firms are not isolated essential to the Union’s international competitiveness in players but are part of a complex system, with its overall knowledge-intensive, innovation sectors, and to sustainable

strength driven by the relative efficiency of its different growth and employment. Investment in education in its components. While a large number of specific problems widest definition, including non-formal and tacit learning,

can be highlighted in relation to the specific players, the is, together with technological advance, the driving force of most serious issue is the poor quality of the linkages dynamic, innovative growth, restructuring towards more

within the overall system In addressing this issue of high value-added production, and the generation of a

linkages and of the wider problem of an general culture of entrepreneurship. The USA invests a far

underperforming EU research sector, some politically larger share of GDP in higher education than the EU (2.7 per cent compared to 1.1 per cent, with a large proportion of

sensitive areas will need to be examined at the national the increase accounted by the private sector) and has

and EU levels i.e. the principle of an developed a far more effective system of linkages with the

excellence/meritocratic based system for awarding world of innovation.

research funds; greater university autonomy, in financial 296 The “European Research Area” launched in 2000 and the

as well as academic terms; a change of culture towards “Investing in Research Action Plan (2003)” address some of the commercialisation of research via closer the wider framework conditions which impact on private

university/business sector links; and the need to develop R&D investment, e.g. intellectual property rights; science and nurture centres of excellence and leading edge and technology human resources; access to venture capital

technology clusters. markets; product market regulations; “technological platforms”; with a view to both initiating a process of

The public and private sectors each play important, structural change towards high tech sectors, and supporting mutually supportive, roles in determining a country’s the internal specialization of traditional industries towards

innovation capacity and each must assume its higher R&D intensity and higher quality products. These responsibilities. Governments have crucial direct and measures all aim at influencing the specialization of

indirect roles to play in the innovation process, directly European industries towards high tech sectors and products. See European Commission (2003).

differentiation) and the characteristics of specific the USA’s global dominance in the ICT industry being a technologies (eg is it a radical or incremental good example of the latter. While world trade volumes innovation; are there network externalities; are there have been rising steadily since the 1950s, what has economies of scale in R&D) is what ultimately changed recently is the nature and scale of the determines the industry specific relationship between globalisation phenomenon, with an increasing focus on market concentration (i.e. the degree of competition) and trade in services and on capital movements in the form

R&D intensity. of FDI, with for example the stock of FDI as a per cent

Market entry and exit rules are crucial to an effective of world GDP tripling since the mid-1980s.

innovation process in rapidly changing industries. The This dramatic intensification of the globalisation process example of the ICT industry highlights the need for is already transforming the economic structures of the policy makers to promote entrepreneurship and a healthy developed and developing worlds, with India emerging process of “creative destruction”. Entry and exit rules as a global power in services, China consolidating its play an important role in boosting productivity by position in manufacturing and with the developed world putting pressure on incumbent firms to innovate and by as a whole searching for an appropriate response. Many supporting market experimentation. This countries in the developed world have recognised the experimentation role is particularly important in seismic nature of the change and are responding industries where the general purpose technologies being positively by embracing an open-economy, innovationused are changing rapidly such as in the production and based, model which emphasises the importance of world use of ICT. In these industries the evidence is clear that class educational establishments; higher levels of, product market regulations that facilitate the easy entry excellence driven and better targeted, R&D; more and exit of firms have contributed enormously to the market based financing systems; and more flexible diffusion of innovations in these industries. regulatory and institutional frameworks delivering a

Concluding remarks more dynamic and competitive business environment. Others are responding in an inappropriate manner by

The present analysis has highlighted the need for the EU attempting to cling to the belief that our present to shift the emphasis in its present economic model more economic problems are temporary and that the towards innovation. This shift in our view is necessitated magnitude of the changes wrought by globalisation will by the increasing competitive pressures of globalisation, avoid the need for fundamental reforms. In this context, by the future challenges of ageing populations and by the collective challenge for EU governments is to the fact that many of the EU’s Member States are close embrace the reality of a rapidly changing global to the technology frontier. Of these factors, the one of marketplace and of the structural changes which it most immediate concern to productivity patterns is inevitably provokes. While Lisbon is a manifestation of undoubtedly globalisation, with the growing the collective desire for change, implementation of the interconnectedness of the world’s economy already needed reforms will be the litmus test of whether the driving up the pace of technological progress, future will witness a substantial recovery in the EU’s intensifying competitive pressures and magnifying the productivity fortunes or will confirm the EU’s ongoing gains from excellence, with the gains being reaped by decline as a global economic power.

References

A HMAD , N., F. L EQUILLER , P. M ARIANNA , D. P ILAT , P. S CHREYER AND A. W ÖLFL (2004), “Comparing Labour Productivity Growth in the OECD Area: the Role of Measurement”, Paper presented at the Banque de France,

CEPII and IFO Workshop, Paris.

B ROCK , G. W. (2003), “The Second Information Revolution”,. Harvard University Press.

D AVERI , F. AND A. M ASCOTTO , (2002), “The IT revolution across the US States”, IGIER Working Paper, No 226.

E UROPEAN C OMMISSION (2003), “Investing in research: an action plan for Europe”.

F LAMM K. (1987), “Targeting the Computer, Government Support and International Competition”, The Brookings Institution, Washington D. C.

F OSTER , L., J. H ALTIWANGER AND C. J. K RIZAN (2002), “The Link between aggregate and micro Productivity Growth: Evidence from Retail Trade”, NBER Working Paper No 9120.

F URMAN , J., M. P ORTER , S. S TERN (2002), “The determinants of national innovation capacity”, Research Policy, 899- 934.

G ORDON , R. (2003), “Hi-tech innovation and productivity growth: does supply create its own demand?”, NBER Working Paper No 9437.

G ORDON , R. (2004), “Why was Europe left at the station when America’s productivity locomotive departed?”, CEPR Discussion Paper No 4416.

I NKLAAR , R., M. O’M AHONY AND M. T IMMER (2003). “ICT and Europe’s Productivity Performance: Industry level Growth Account Comparisons with the United States”. Groningen Growth and development Centre, Research

Memorandum GD-68.

J ONES , C. I. (2002), “Sources of US economic growth in a world of ideas”, American Economic Review, Vol. 92, pp. 220-239.

OECD (2003), “The Sources of Economic Growth in OECD countries”.

OECD (2004), “Understanding Economic Growth”, Paris.

O’ M AHONY , M. AND B. VAN A RK (2003), “EU productivity and competitiveness: an industry perspective”, Study prepared for the Enterprise Directorate General.

O LINER AND S ICHEL (2002),”Information Technology and Productivity: Where are we now and where are we

rd

going?”, Federal Reserve Bank of Atlanta Economic Review, 3 Quarter.

S IELING , M., B. F RIEDMAN AND M. D UMAS (2001), “Labour Productivity in the Retail Trade Industry, 1987-99”, Monthly Labor Review, pp. 3-14.

S TERN , S., P ORTER , M. E. AND J. F URMAN (2000), “The determinants of national innovative capacity”, NBER Working Paper, No 7876.

S TIROH , K. (2002), “Are ICT spillovers driving the new economy”, Review of Income and Wealth 48, No 1.

T RIPLETT , J., AND B. B OSWORTH (2000), “Productivity in the Services Sector in the International Economy”, R. M. Stern (ed.), University of Michigan Press.

ANNEX: P RODUCTIVITY MEASUREMENT ISSUES

The IMF in its recently released Article IV report on euro-area policies maintained that the euro area did not have a productivity problem since, according to their estimates, hourly productivity was higher than in the USA and all of the differences in per capita incomes were due solely to the lower number of hours worked by euro-area workers. On the basis of Eurostat’s structural indicators, however, a completely different picture emerges, with Eurostat’s productivity measure suggesting that 45 per cent of the gap in living standards between the euro area and the USA is due to lower labour productivity per hour, with the remaining 55 per cent an hours worked issue. To complicate matters even more, the equivalent OECD estimates suggest a position which roughly lies between that of Eurostat and the IMF, namely that the EU has a productivity problem but that it is not as severe as suggested by Eurostat. The OECD figures roughly coincide with those of ECFIN’s own analysis. This ongoing issue of conflicting measurements of the EU-US productivity gap (or the lack of it) is fundamental to the present policy debate and the

relative emphasis to be placed on the employment or productivity aspects of the Lisbon strategy.

Overview of current situation and the extent of the problems. The graph shows the extent of the problems to be resolved on the basis of the Eurostat and OECD estimates (comparable IMF data is not available). While there are some differences in terms of GDP per capita and the per person employed productivity measures, it is very clear that the real source of the differences lies in the hours worked calculations and the associated hourly labour productivity

figures.

Short and Long Run Solutions. There Comparison of GDP per capita, employment rates, hours worked per worker and

seems to be only one long term labour productivity: Eurostat versus OECD

durable solution to these ongoing

productivity measurement problems 120 Eurostat

and that is for all of the interested OECD

parties, most notably, Eurostat, the 100

OECD and the IMF to discuss the

different methodologies which they 80

employ for calculating the various

input series. This in fact is what is 60

happening at the moment, with

Eurostat and the OECD actively 40

discussing the various issues. It is

hoped that a final long run resolution 20

to these problems can be

forthcoming before the end of 2004. 0 GDP per capita in Employment Rate (1) Hours Worked per Hourly Labour Labour Productivity –

Regarding possible short run PPS Worker (2) Productivity GDP in PPS per

solutions, ECFIN decided that given person employed

the uncertainties involved, it was

very important to have its own Note: (1) Calculated- employment rate+100* (GDP per capita/labour productivity per person employed). (2) Calculated- hours worked per worker=100* (labour productivity per person

internally produced productivity employed hourly labour productivity). EU 15 performance 2002 (US=100).

series for analytical purposes. Source: Commission services.

Following an assessment of the strengths and weaknesses of the various statistical sources, the most internally consistent source was deemed to be the OECD’s labour force statistics databank, which has internationally comparable figures for hours worked and employment for both the EU’s Member States as well as for the USA. ECFIN is satisfied that the derived series constitute an acceptable interim solution, with the absolute levels for the different series and their evolution over time indicating a plausible pattern. These derived series, not surprisingly, are very similar to those of the OECD. Once convergence has been reached in the discussions between Eurostat and the OECD regarding the best input series to be utilised, ECFIN will adjust its own methodology to make it fully consistent. According to the ECFIN staff who are participating as observers to the Eurostat/OECD discussions, it appears that the final solution will not differ dramatically from the present ECFIN estimates. In fact, in terms of

productivity levels, the EU-15 estimate is likely to lie between the present OECD and Eurostat estimates. 297 Since the

mid-point estimate looks likely to be the basis of a final consensus, the IMF’s viewpoint that the EU-15 does not have a productivity problem (and consequently that it should focus its Lisbon agenda solely on the employment

297 It has been tentatively agreed that the OECD, for the purpose of productivity measurement, will move from the labour force

framework to the national accounts. Eurostat will align their basic numbers for the USA with those from the OECD, with the OECD in turn converging towards Eurostat’s figures for the individual EU Member States. Eurostat, in a provisional assessment following the discussions with the OECD, now estimates that the EU-15’s hourly labour productivity relative to the

USA’s was 88.5 in 2002 (compared with Eurostat’s previous estimate of 86.7 and the OECD’s estimate of 91).

front) will not be supported by the underlying data. This IMF position, it should be stressed, only applies to productivity levels, with the IMF also accepting that there has been a significant deterioration in the EU’s relative

position over recent years due to much lower EU productivity growth rates compared with those of the USA. 5. THE LINK BETWEEN PRODUCT MARKET

REFORMS AND PRODUCTIVITY: DIRECT

AND INDIRECT IMPACTS

Summary

Does the lack of product market reforms have some connection with Europe’s poor productivity record? This question is relevant given the central role played by these reforms in the Lisbon strategy. In order to try to answer this question, this chapter examines to what extent and through which channels product market reforms can have an impact on productivity developments. Thereafter, it identifies the areas of product market reforms where the EU significantly lags behind the USA. This analysis allows drawing some conclusions on the most pressing product market reforms needed to improve the EU productivity performance.

Product market reforms play a central role in the Lisbon strategy as they improve the framework conditions in which business operates. In the year 2000, the EU decided in Lisbon on the strategic economic goal of becoming the most competitive and dynamic knowledge-based economy with sustainable economic growth and greater social cohesion by 2010. To achieve this goal, the EU set a strategy for delivering stronger growth, known as the Lisbon agenda. A major aim of the Lisbon agenda is to increase GDP per capita and to boost productivity. Well-designed product market reforms can contribute to this objective. The way product markets are regulated has an important impact on the degree of competition in the market, the scope and the size of the market and therewith on the size and the structure of economic activity in goods and services. The existing regulations are often a heritage of the past.

Therefore, the framework conditions in which business operates may not always be optimal in terms of today’s circumstances and challenges. As a consequence, they may act as a brake rather than as a spur on economic activity and stifle initiative rather than encourage it. That is why product market reforms, by improving these framework conditions, can help to unleash a hidden potential in the economy thereby spurring productivity growth and increasing welfare.

Product markets reforms have direct and indirect impacts on productivity and the later operate through three main channels. The direct impacts occur through the decrease in costs of doing business and through the removal of barriers to penetrate new markets. However, the change in the framework conditions which improves the functioning of product markets has also indirect effects on productivity. These indirect effects operate through three main channels, namely a reduction in mark-ups and a reallocation of scarce resources (allocative efficiency); an improvement in the utilisation of the production factors by firms (productive efficiency); and an incentive for firms to innovate and to move to the modern technology frontier (dynamic efficiency).

Gains through allocative and productive efficiency represent one-off changes to the level of productivity and output and accrue relatively rapidly but product market reforms may also result in dynamic efficiency. The effects from this third channel tend to accrue over a longer period of time. But improvements in such dynamic efficiency gains potentially have a much larger impact on productivity. Successful innovations should eventually raise the level and growth rate of total factor productivity in the long term but this may take some time to accrue.

Empirical evidence shows that a large part of the impact of product market reforms on productivity is through indirect effects. Empirical studies have shown that the net effects of the direct impacts on productivity tend to be small. However, a number of studies have estimated much higher impacts on productivity as a result of product market reforms. This suggests that a large part of the impact on productivity is through indirect effects.

Product market reforms reduce the economic rents in the economy and promote business dynamism. Studies confirm that product market reforms that ease entry, reduce trade barriers, remove price controls and reduce public involvement in production, negatively affect the average level of economic rents in the economy. Product market regulations also stand out as having a substantial impact on the levels of new entry by businesses. Studies found that overly complicated license and permit system or badly designed tax systems discourage the creation of new enterprises. The direct effect of free entry and exit can be decomposed in two effects: First, internal restructuring

(also called “within effect”) refers to productivity growth of individual firms in the industry via factors internal to the firm such as organisational change, new technologies, or reallocation of inputs. Second, external restructuring represents a reallocation of resources among firms via a process of exit of least efficient firms and/or via a shift in market shares towards most efficient firms. Most studies points to large within effects. However, in high–tech sectors where productivity gains are the most important, it is the new firms that make the most significant contribution to productivity growth.

Product market reforms can stimulate innovation but their effects take longer to materialise. The relationship between product market reforms and dynamic efficiency is more complex as there is increasing evidence of an inverted U-shaped relationship between innovation and competition. The direct link between competition and dynamic efficiency – measured by productivity growth - seems to be clearer with several studies showing a significant positive effect, especially thanks to the process of entry and exit of firms. However, competition seems to deliver its full effects on dynamic efficiency with long lags and the literature underlines differential effects of innovation on productivity growth depending on the distance to the technological frontier: the closer to the technological frontier, the more positive the effects.

Medium to long-term gains in productivity due to product market reforms could be substantial Several studies that have analysed the potential gains of adopting US-level product market reforms show substantial potential GDP and/or productivity increase in the long-term. In the short-term, the effect of increased competition is a boost in employment, which puts less productive workers into jobs. Therefore, the immediate impact on productivity level and growth is at best small because the direct effect of product market reforms (an increase in productivity thanks to a decrease in costs) is partly cancelled out by the integration of less productive workers into the job market. However, although this eventually decreases average labour productivity, it still remains that the impact on standards of living is unambiguously positive. There is also evidence that long-term gains could be large. Some studies show that reforms facilitating market entry and raising the level of competition on goods and services markets could result in productivity gains of between 2 and 4 per cent.

The EU has already undertaken profound product market reforms with the ‘Internal Market strategy’. The next question is which product market reforms are the most pressing to improve the productivity performance of the

European Union. While further work is necessary to investigate the relative importance of the three channels and to analyse – possibly at sectoral level – the links between the gap productivity and the gap in product market reforms, there are already interesting conclusions which emerge from a comparison between the European Union and the

United States. The European Union has already initiated profound reforms, in particular with the Internal Market. The European Union is open to international competition and its network industries are liberalised to a degree that equals if not exceeds the United States. The ‘Strategy for Europe’s Internal Market’, launched in the fall of 1999, should be seen as a deepening of the Single Market Programme, which aimed at the elimination of all barriers to the free circulation of goods, services, capital and persons by the end of 1992. This new Internal Market strategy should contribute to deepen market integration and improve the regulatory environment for business.

Further product market reforms should mainly aim at promoting business dynamism and pursuing integration as these are the areas where Europe lags behind the USA. Europe’s backwardness in product market reforms seems to be concentrated in measures that promote entry and exit of firms and in a lower degree of intra-state trade integration. Also, evidence of backwardness in product market reforms in the EU is the apparently higher costs of complying with regulation than in the USA, although European companies do not perceive regulations as more timeconsuming than US companies do. State involvement in the economy is higher in Europe but the consequences of this are debatable. Furthermore, a lack of flexibility in labour markets and to some extent more regulations on credit – two issues not reviewed here – may also explain a sizeable share of the USA-EU gap in productivity.

In the context of the mid-term review of the Lisbon strategy, this analysis can contribute to the choice of priorities for reforms in the area of product markets. We conclude that reforms to ease entry and exit are important. These should go beyond measures to reduce time and cost to start up a company and should include reforms promoting Europe as an attractive and easy place to do business. Similarly, making sure that the internal market is working at full capacity should be a clear objective for the Union.

TABLE OF CONTENTS

• 1. 

  I NTRODUCTION .............................................................................................................................................197

• 2. 

  D EFINITION AND MEASUREMENT OF PRODUCT MARKET REFORMS ...............................................................198

• 3. 

  T HE THREE TRANSMISSION CHANNELS .........................................................................................................198

3.1 Allocative efficiency............................................................................................................................................ 198

3.2 Productive efficiency ........................................................................................................................................... 199

3.3 Dynamic efficiency.............................................................................................................................................. 201

• 4. 

  E MPIRICAL EVIDENCE ON THE INDIRECT LINKS BETWEEN PRODUCT MARKET REFORMS AND PRODUCTIVITY 202

4.1 Product market reforms and allocative efficiency................................................................................................ 202

4.2 Product market reforms and productive efficiency .............................................................................................. 203

4.3 Product market reforms and dynamic efficiency ................................................................................................. 204

4.4 Direct and indirect effects of product market reforms. ........................................................................................ 206

• 5. 

  I S THE EU LAGGING SIGNIFICANTLY BEHIND THE USA IN TERMS OF PRODUCT MARKET REFORMS ? ............207

5.1 Measurement of product market reforms in the EU and the USA ....................................................................... 208

5.2 Taking stock of product market reforms in the USA and the EU ........................................................................ 209

• 6. 

  P OLICY IMPLICATIONS ..................................................................................................................................215

R EFERENCES ...........................................................................................................................................................216

ANNEX: A COMPARISON OF THE DEGREE OF LIBERALISATION OF NETWORK INDUSTRIES IN THE

EU AND THE USA...................................................................................................................................220

THE LINK BETWEEN PRODUCT MARKET

REFORMS AND PRODUCTIVITY: DIRECT

AND INDIRECT IMPACT

• 1. 

  Introduction impact on productivity performance. 299 Indeed, product market reforms have direct effects on productivity

In the year 2000, the EU decided in Lisbon on the developments because they decrease the costs of doing strategic economic goal of becoming the most business and remove barriers to penetrate new markets. competitive and dynamic knowledge-based economy However, the change in the framework conditions that with sustainable economic growth, more and better jobs improves the functioning of product markets has also and greater social cohesion by 2010. To achieve this indirect effects on productivity. These indirect effects goal, the EU set a strategy for delivering stronger operate through three main transmission channels, growth, known as the Lisbon agenda. A major aim of namely a reduction in mark-ups and a reallocation of the Lisbon agenda is to increase GDP per capita and to scarce resources (allocative efficiency); an improvement boost productivity. Product market reforms play a in the utilisation of the production factors by firms central role in this agenda because they are expected to (productive efficiency); and an incentive for firms to improve the framework conditions for business. innovate and to move to the modern technology frontier

298 (dynamic efficiency).

Studies by the European Commission and the IMF

(2003) indicate that reforms facilitating market entry and One of the main objectives of this chapter is to discuss raising the level of competition on goods and services these indirect links between product market reforms and markets could result in productivity gains of between 2 productivity. To that end, Section 2 starts by explaining and 4 per cent. The way in which product markets are how to define product market reforms and to measure regulated has an important impact on the degree of product market reforms. Section 3 provides the competition in the market, the scope and the size of the theoretical framework for our analysis and describes the market and therewith on the size and the structure of three transmission channels through which product economic activity in goods and services. Regulations market reforms can influence productivity. Section 4 may act as a brake rather than as a spur on economic reviews the empirical literature. This empirical evidence activity and stifle initiative rather than encourage it. That shows that these indirect effects are potentially larger

is why product market reforms, by improving the than the direct effects.

framework conditions, can help to unleash the potential in the economy thereby spurring productivity growth and increasing welfare.

This chapter aims to analyse to what extent and through 299 At the same time, interactions between reforms in the

which channels product market reforms could have an product market and in the labour market are likely to impact on employment and need to be taken into account. In addition, investment in human capital is also potentially important for productivity. Those interactions are not

298 See chapter 2 in the EU ECONOMY 2002 REVIEW and considered here as this chapter focuses on productivity

chapter 2 in the EU ECONOMY 2003 REVIEW . issues.

The next question relates to which product market Europe’s Internal Market’, launched in the fall of 1999, reforms are the most pressing to improve the should be seen as a deepening of the Single Market productivity performance of the European Union. While Programme. This Internal Market strategy should further work is necessary to investigate the relative contribute to furthering market integration and

importance of the three channels and to analyse – improving the regulatory environment for business.

possibly at sectoral level 300 – the links between the gap

productivity and the gap in product market reforms, Major reforms that have affected European product interesting conclusions already emerge from a markets include measures taken within this new Internal comparison of the degree of product market rigidities Market strategy, such as removal of remaining barriers between the European Union and the United States. This to integration of goods and services, liberalisation and comparison is made in Section 5. Section 6 concludes by regulatory reform in network industries, reduction in

drawing some policy implications. state aids, reforms of competition policy, simplification of regulation to set up a company, etc. However, there is

relatively little direct information available on the scale

• 2. 

  Definition and measurement of and scope of these reforms across countries. Consequently, the computation of a reliable summary

product market reforms indicator of product market reforms in the EU is difficult

and only economy-wide indicators of specific product Product market reforms are microeconomic reforms that market reforms or indicators of reforms implemented in aim to improve the framework conditions in which specific industries are available. business operate. Four main types of measures may be

undertaken to reform product markets: 3. The three transmission channels

• First, measures to open up markets (goods and

services) that were previously sheltered from Product market reforms have direct and indirect effects competition from abroad by tariff barriers (trade on productivity. This chapter does not discuss the direct

openness) or legal barriers (liberalisation). impacts that occur through the decrease in costs of doing business and through the removal of barriers to penetrate

• Second, measures to open up markets that were new markets. Product market reforms have also indirect

previously sheltered from competition from effects on productivity. Theoretical models suggest that newcomers – whatever their origin – because of reforms that liberalise or improve the functioning of stringent regulations on entry, such as permits and markets can positively affect productivity through three licences, or non-tariff barriers, such as specific different channels that this section discusses: a national regulations (deregulation). reallocation of scarce resources (allocative efficiency),

• Third, measures to create a more business-friendly an improvement in the utilisation of the production factors by firms (productive efficiency) and an incentive

environment, such as the reduction of time and for firms to innovate to move to the modern technology costs to set up a new company or appropriate levels frontier (dynamic efficiency). This typology however and systems of taxation gives a simplified picture of the reality because it • Fourth, measures that seek to reduce the State’ ignores the possible interactions between the three involvement in the economy, since this is likely to channels. For example, entries can have a negative disturb the well functioning of markets (ad hoc effect on mark-ups and a positive effect on innovation. State aid, subsidies and State-owned firms Likewise, innovation can increase the pressures of

competing with private firms). competition in a market.

The implementation of the Single Market Programme 3.1 Allocative efficiency

intended to abolish all barriers to the free movement of

goods, services, persons and capital within the European Product market reforms increase the number of Union by the end of 1992. This programme constitutes competitors or the threat of entry of new competitors,

the most recent comprehensive exercise of product leading to more competitive markets. By increasing the contestability of markets and by reducing incumbents’

market reform. The 1988 Cecchini report considered

that the economic gains from the completion of the market power, this induces firms to set prices closer to internal market would stem from more intense marginal costs. As a consequence, mark-ups tend to

competition and economies of scale. The ‘Strategy for decrease while the allocation of both inputs (labour and capital) and goods is more efficient, i.e. the allocation of

resources is made so that consumer wants and needs are

300 met in a better way than they were in the previous Because of data availability constraints, this chapter only

looks at economy-wide regulations. Further studies should period. More product market competition can also lead look at whether these regulations could – maybe through to increased allocative efficiency as less productive

interactions with other variables – have more influence on firms exit and market share moves from less productive

some specific sectors and therefore explain sectoral to more productive firms.

differences in productivity gap between the EU and the USA.

Theoretical models that focus on the reallocation effects 3.2 Productive efficiency

of liberalisation generally consider that the latter has a

positive impact on economic performance. For instance, Productive efficiency is the capacity for any given firm Melitz (2003) specified a model with imperfect to allocate its resources in such a way that makes it competition and heterogeneous firms in which opening possible to reduce or eliminate the under-utilisation of

to trade leads to a reallocation of resources towards its production factors, i.e. capital and labour.

302

more productive firms within industries. Low Productive efficiency and productivity are not identical

productivity firms exit, high productivity firms expand concepts but they are interrelated.

303

  A decrease in

in the domestic market and some enter the export productive inefficiency could be associated with an

market. This leads to an increase in aggregate increase in productivity.

304 Productive or technical

productivity, even when there is no productivity growth efficiency gains come from the introduction of new or

within the firms. better production methods within the firm, and this could lead to increased productivity.

However, it is not always the case that a rise in

competition would lead to increased allocative The main impact of higher product market competition efficiency. For example, Vickers (1995) points out that on productive efficiency that the literature emphasises is an increase in competition through more aggressive the incentive effect on managers and workers to reduce interactions between firms could increase industry slack, trim fat and structure the workplace more

concentration in the medium term since more aggressive efficiently.

305

  Principal-agent models under information

firm’ behaviour first reallocates profits from inefficient asymmetry generally assume that managers and workers firms to more efficient ones (reallocation effect) and can partially capture monopoly rents to a monopolistic subsequently drives out inefficient firms (selection firm in the form of managerial slack, i.e. lack of efforts. effect). This thereby raises industry concentration and By giving more incentives to increase managerial efforts mark ups, but this model implies that market entry is not and improve efficiency, competitive pressures may possible for new competitors, and so does not present reduce slack and discipline firms into efficient

the whole picture. operation.

306

 Incentives to improve productive efficiency

could arise through different channels. 307

Blanchard and Giavazzi (2001) show the importance of

market entry - or more precisely the importance of • Competition reduces information asymmetry and

market contestability - as a stimulus for competitive creates greater opportunities to compare pressures and economic performance. In a model in performance. This makes it easier for the which firm and employee productivity is fixed, and in shareholders to monitor managers and hence

which labour is the only factor of production, they reduces slack; consider the impact of product market regulations. In • In highly competitive markets where price elasticity their model, deregulation of product markets can take of demand is high, cost-reducing productivity the form of either increased substitutability between improvements are likely to generate large increases goods or of a reduction in entry costs. In the short run, in market shares and profit;

increased substitutability between goods leads to lower

mark ups, reduced unemployment and higher real • The probability of bankruptcy is likely to be higher

wages. In the long term, the same results occur only if in a more competitive environment. Consequently, deregulation leads to a reduction in barriers to entry. If managers have an incentive to step up their efforts

this is not the case, then firms exit because of lower to avoid such a failure.

level of rents and, as a result, mark ups, unemployment • Competition may also influence the effort of

and real wages return towards their original levels. In

this framework, liberalisation through the ease of workers, as they are likely to capture a part of firm/market entry is thus a major determinant of the product market rents in the form of slack or higher effectiveness of product market reforms aimed at wages. Therefore, there is a direct link between the

stimulating competition. degree of competition and the level of worker’s effort. 308

However, the welfare gains achieved by increasing allocative efficiency are not by themselves likely to be

very large. 301 Allocative efficiency gains mainly impact

indirectly on economic performance by inciting firms to

improve their productive efficiency and to enhance 302 See Pilat (1996).

efforts to innovate and speed up diffusion of innovation, 303

but this is still a debated issue in the literature. See Sharpe (1995). 304 See Pilat (1996).

305 See Griffith and Harrison (2004).

306

  See Ahn (2001).

307 See Winston (1993), Meyer and Vickers (1997), Nickel et

301 See Harberger (1954), Leibenstein (1966), Scherer and Ross al. (1997), Aghion and Howitt (1998).

(1990). 308 See Nickell (1996).

Theoretical framework describing the indirect effects of product market reforms on productivity.

Product market reforms

Deregulation, liberalisation, trade

openness

Competition

Entry and threat of entry

Opening up of markets

ALLOCATIVE EFFICIENCY PRODUCTIVE EFFICIENCY DYNAMIC EFFICIENCY

Downward pressure on prices and mark Incentive to reduce both slack and costs Incentive to innovate to escape

ups competition at low levels of

Industry restructuring competition

Inverted U-curve relationship

Increase in technical efficiency Increase in

(K and L productivity) Total Factor Productivity

3.3 Dynamic efficiency competition increases each firm’s incentive to

Gains through allocative and productive efficiency reduce its production costs through the acquisition represent one-off changes to the level of productivity of a technological lead over its rivals. Competition and output and accrue relatively rapidly, i.e. in the short may also increase the incentive of each firm to run. However, an increase in competition may also act innovate to escape competition, for instance if as a stimulus for firms to develop product and process innovation translates into more sophisticated and

innovations and hence to speed up the move to the differentiated products.

modern technology frontier. Improvements in such In these models, the link between competition and dynamic efficiency gains potentially have a much larger innovation does not remain unambiguous as this link is impact on productivity but are also likely to take much prone to be positive in ‘neck-and-neck’ industries

longer to accrue, i.e. successful innovations will whereas it is prone to be negative in less ‘neck-andeventually

raise the level and growth rate of total factor neck’ -or more ‘product-differentiated’- industries. In

productivity in the long run. 309 However, the link the latter type of industries, more competition may between competition and innovation is a debated issue reduce innovation as more competition reduces the in the theoretical literature. monopoly rents that reward successful innovators

On the one hand, in line with the Schumpeterian view of (Schumpeterian effect). Moreover, by increasing market power and innovation, the early endogenous innovation incentives in ‘neck-and-neck’ industries growth and industrial organisation literatures 310 suggest more than in ‘product-differentiated’ industries, this will that increased product market competition leads to tend to reduce the fraction of ‘neck-and-neck’ industries reduced innovative activity, as more competition in the economy in equilibrium. This effect reinforces the reduces the monopoly rents that reward successful Schumpeterian effect in inducing a negative correlation innovators. However, this literature was based on the between product market competition and aggregate

assumption that innovation was made by outsiders or by productivity growth or aggregate rate of innovations.

313

new entrants competing against incumbents with New endogenous growth models allow incumbent firms conventional technology. They also assured that the to innovate and assume that innovation incentives payoff of innovation was equal to the post-innovation depend on the difference between post-innovation and rent (while the pre-innovation rent was zero). pre-innovation rents. Aghion et al. (2003a) show that,

311

On the other hand, new endogenous growth models when entry is introduced into such models, the effect of extend the basic Schumpeterian models by allowing an increase in competition through (the treat of) entry incumbents firms to innovate and by assuming that depends on the country, industry or firm’s distance to innovation incentives mainly depend on the difference the world technological frontier. In countries that are between the post-innovation and the pre-innovation close to the world technological frontier, fostering entry rents. These models predict that more product market or competition will increase incumbents’ incentives to competition could foster innovation. Aghion and Howitt innovate in order to escape potential entrants or new (1998) offer two theoretical cases where competition is competitors. However, in countries or industries lagging

indeed conducive to innovation: far behind the world technological frontier, higher entry or higher competition tends to discourage incumbents

• Intensified product market competition could force from innovating. This model thus suggests that the managers to speed up the adoption of new overall impact of trade liberalisation will depend on the technologies to avoid loss of control and/or current state of technology in the country or the bankruptcy. Indeed, if successful innovators that industry. However, in the long run, trade liberalisation introduce new technology lead to a gain in market will increase the overall average growth rate because in shares because of more efficient production equilibrium there will be more industries where the

processes, they would be able to replace the firms affect is positive.

that produce with old technology. The latter are thus Finally, work by Aghion et al. (2002) suggests that the

forced to innovate themselves in order to survive. 312 relationship between competition and innovation may be

• In ‘neck-and-neck’ industries, i.e. in industries in of a non-linear nature, with both very high and very low which oligopolistic firms face similar production levels of product market competition providing lower costs, product market competition could create a incentives to innovation. Using a Schumpeterian growth large incentive to innovate. This is because intense model in which firms innovate step by step (i.e. a

laggard firm must first innovate to catch up with the technological leader before becoming itself a leader in

309 See Ahn (2002), Griffith and Harrison (2004). the future), where both technological leaders and their 310 Dasgupta and Stiglitz (1980), Romer (1990) or Aghion and followers engage in R&D activities, and where

Howitt (1992). competition may increase the incremental profit from

311 Such as the ones developed by Aghion, Harris and Vickers innovating while reducing innovation incentives for

(1997), Aghion and Howitt (1998) or Aghion, Harris, Howitt and Vickers (2001).

312 See Ahn (2002). 313 See Aghion et al. (2002).

laggards, these authors indeed predict that the confirming the theoretical analysis. 315 Jacquemin and

relationship between competition and innovation is an Sapir (1991), for instance, showed that the European inverted U-shape, i.e. the escape competition effect national industrial sectors that were protected from intradominates for low initial levels of competition, whereas EU competition by important non-tariff barriers the Schumpeterian effect dominates at higher levels of benefited in the early 80’s from abnormally high competition. profitability levels. More recently, Oliveira Martins et

To sum up, the new endogenous growth models predict al. (1996) did not find any correlation between the that the link between competition and innovation may be degree of market concentration and the level of markpositive or negative depending on the initial state of ups but identified a significant negative correlation competition (neck-and neck industries versus ‘productbetween the latter and the entry rates in a market. The differentiated’ industries and more generally low level European Commission (1996) also showed that the of competition versus high degree of competition) and implementation of the SMP led to an increase in on the country, industry or firm’s initial distance to the competitive pressures in the manufacturing industry,

world technological frontier. resulting, in particular, in reductions in the price–cost margins. Griffith and Harrison (2004) estimated the

relationship between product market reforms and the

• 4. 

  Empirical evidence on the indirect level of economic rents. They found that reforms that ease entry, barriers to trade, remove price controls and

links between product market reduce public involvement in production negatively reforms and productivity affect the average level of economic rents in the

economy.

This section presents a survey of recent empirical work However, if high profit levels can be interpreted as a analysing the indirect links between product market consequence of low competitive pressures, in contrast

reforms and productivity. Drawing on the theoretical

framework presented in Section 2, this section makes a they can also result from efficient behaviour of firms (Ahn, 2002). 316 As the effects of competition on the

distinction between the three transmission channels of mark-up ratio may be ambiguous, its evolution over time the effects of product market reforms, namely allocative has to be analysed simultaneously with the evolution of

efficiency, productive efficiency and dynamic its two components, namely prices and unit costs. 317 efficiency. Applying this methodology for analysing what impact 4.1 Product market reforms and allocative the implementation of the Single Market Programme has

efficiency had on mark ups in the European manufacturing industry, Sauner-Leroy (2003) finds evidence that profit

By increasing competition, product market reforms have margins of EU firms in the early 1990’s declined in line two main effects on allocative efficiency. First, they with a decrease in real prices and that this phenomenon weaken the market power of firms, leading to a could be at least partly attributable to increased reduction in monopoly rents or mark-ups. Second, they competition stemming from intra-EU imports, thus facilitate market entry of new firms and this may lead to indicating the realisation of allocative efficiency gains

some restructuring process, with market exit by the least induced by the intra-EU trade liberalisation. efficient firms. As some incumbent firms tend to be less productive than their more profitable new competitors, their exit from the market raises the average productivity

in the sector. 315

See Schmalensee (1989).

316

  For instance, Demsetz (1974) considers that high profit

4.1.1 Product market reforms and mark-ups levels within an industry can be explained by good

The mark-up is defined as the price over marginal cost performances of firms, i.e. their ability - mainly for largest

ratio. A mark-up ratio exceeding unity is an indication firms - to produce at low costs. More generally, an increase in competition may have effects on both prices and costs,

of the existence of market power enabling firms to set and therefore the mark-up may remain stable although prices above marginal costs and thereby to achieve prices may fall. Geroski et al. (1996) show in particular that

monopoly rents. 314 this is likely to occur because incumbent firms have excess

Most empirical studies that aimed to test the links costs – such as managerial slack or rent sharing with the workforce- that can be reduced to compensate for lower

between the degree of market opening and/or the degree prices. This latter argument could be a good explanation for of competition, on the one hand, and the profitability the paradox mentioned by Konings et al. (2001) concerning

level of firms, on the other hand, have found - since the the levels of mark-ups in the Dutch manufacturing sectors pioneering works of Bain (1951 and 1956) - a negative facing high import rates. The authors found mark-ups to be

relation between these two variables, therefore higher in these sectors than in sectors where import rates

were low. Indeed, if the intensity of competition results in a fall in costs larger than the fall in prices, then the profit

margin increases.

314 See Oliveira Martins et al. (1996), Konings et al. (2001). 317 See Bils (1987), Machin and Van Reenen (1993).

4.1.2 Product market reforms and market entry decompose the direct impact of free entry and exit in

and exit several effects. First, internal restructuring (“within effect”) refers to productivity growth of individual firms

Several factors explain entry and exit of firms. Among in the industry via factors internal to the firm such as those, product market regulations stand out as having organisational change, new technologies, or reallocation substantial impact on entry rate of businesses. Brandt of inputs. Second, external restructuring represents a (2004) found that overly complicated license and permit reallocation of resources among firms via a process of system discourages the creation of new enterprises. creative destruction with exit of least efficient firms or

Among product market reforms, differences in corporate via a shift in market shares towards most efficient firms. tax systems across Europe can act as a key obstacle to Barnes, Haskell and Maliranta (2001) found substantial cross-border activities in Europe. Studies suggest a high within effects for the OECD. Baily, Hulten and compliance cost related to the lack of co-ordination of Campbell (1992) found similar results for the US tax and accounting systems in Europe. 318 Besides these manufacturing firms between 1972 and 1988, and so do direct compliance costs, tax systems are known for Griliches and Regev (1995) for the Israeli industry over

having a large impact on entrepreneurship and on 1979-1988.

innovation activities, either through the general tax Nicoletti and Scarpetta (2003) also demonstrate

framework or through targeted tax policies. 319 The effect significant links between product market policies and

of taxation on entry is tricky to apprehend because productivity performance, with entry liberalisation entrepreneurs have the possibility to be self-employed or leading to productivity gains in all of the countries

to incorporate, 320 therefore involving decisions based on considered regardless their position in terms of

both the personal and corporate income tax systems. In technology adoption. They also found evidence of a addition, not only the level but also the progressivity of twofold effect of entry liberalisation that release their some tax systems and the relative difficulties to carryeffects over a ten years time horizon. First, entry

over losses across tax periods 321 imply that decisions liberalisation in the services industries is estimated to

related to entrepreneurship also depend on the forecast boost annual multi-factor productivity growth in the level and distribution of earnings. Other types of overall business sector. Second, an indirect (and taxation also matter such as capital gain taxation that has positive) effect of the removal of trade and an effect on the level of venture capital supplied to administrative barriers to entry was found. The intensity

entrepreneurs. 322 Finally, targeted tax measures – be of the effect depends on the technology gap that some

they in terms of tax credits, specific depreciation rules, countries accumulated in heavily regulated

reduced taxation or preferential exemptions – have a manufacturing industries.

large impact on entrepreneurial activities. 323 In conclusion, studies point a significant effect of market

Generally, studies analysing the impact of product entry on allocative efficiency. market reforms on economic performance through the

process of entry and exit focus for a large part on the 4.2 Product market reforms and productive

link with productivity. Empirical studies generally find a efficiency

positive link between the two indicators. One can Studies analysing the links between product market reforms and productive efficiency are also relatively

318 scarce, maybe because productive efficiency is difficult See European Commission (2001b) and European

Commission (2004b). to measure as it depends on various factors, some of

319 them not being observable such as, for instance, See Gentry and Hubbard, (2004).

organisational changes within companies. 324

320 In some cases, incorporation can be made so that

entrepreneurs still face personal income taxes. The design Empirical works already been done on this issue mostly

of tax rules will also have an impact on productivity include studies focusing on the relationship between

through its incentives. This is for example the case for the competition and productive efficiency, and not taxation of performance-related pay systems, such as stock specifically on the relationship between product market options or bonuses. reforms and productive efficiency. Nevertheless, one

321 Leading in some cases to “success taxes” because losses do can argue that as product market reforms tend to

not lead to negative taxes. Therefore, successful companies increase competition, the conclusions drawn from

usually face a higher effective taxation than unsuccessful empirical work linking competition and productive ones. efficiency are also valid for the analysis of the links

322 A sizeable collateral damage of a lower supply of venture between product market reforms and productive capital is the decrease in managerial advice that usually efficiency. Caves and Barton (1990), Caves et al. (1992)

accompanies capital invested in risky activities. See

Keuschnigg and Nielsen (2000). or Green and Mayes (1991) used frontier production

323 function techniques to compute efficiency indices and to See Cullen and Gordon (2002). In addition, the effect of tax

systems on foreign direct investment is well-established, relate them to competition variables. They found that,

adding foreign competitors to the level of entry (see for

example Clark, 2002). 324 See Sauner-Leroy (2003).

above a certain threshold, increases in market and Audretsch (1987) found that different types of concentration (i.e. decreases in competition) tend to be industries would produce innovative advantage for associated with reductions in technical efficiency. different sizes of industries. Small companies have

These results are consistent with the ones found in innovative advantages in highly-innovative and skilled studies focusing on product market reforms. Griffith intensive sectors whereas large companies enjoy this (2001) for instance shows that the increase in product advantage in more concentrated and capital-intensive market competition brought about by the industries. Using firm-level UK data, Blundell, Griffith implementation of the SMP led to an increase in overall and Van Reenen (1999) found that firms with higher levels of efficiency, but that these efficiency gains market share innovated more but that at the industry occurred more particularly in firms where management level, more competitive industries were more and ownership were separated (principal-agents type of innovative. Therefore, aggregate competition leads to firms), suggesting then that product market competition more innovation but within the competitive industries

325

can play an important role in reducing agency costs. dominant firms innovate more often. One difficulty is Sauner-Leroy (2003) also shows that the rise in the possible endogeneity of market structure as it may competition induced by the implementation of the Single itself be the result of innovation. In addition, although Market Programme led EU manufacturing firms to affecting R&D investment, regulations do not seem to increase their productive efficiency to compensate for be its main driver – some forms of protection could even

lower prices and profit margins. be beneficial for risky R&D activities – and market size and education appear to be more pronounced

To summarise, there is evidence that competition and determinants. 327

product market reform act as a stimulus for firms to

increase productive efficiency. However, the empirical There is however increasing evidence of an inverted U-

work on this specific issue remains relatively scarce. shaped relationship between competition and R&D or innovation, as predicted by most recent models (Aghion

4.3 Product market reforms and dynamic et al. 2002). Griffith and Harrison (2004) looked at the

efficiency link between microeconomic reforms in product markets on macroeconomic performance through their effects on

The empirical literature on the link between product mark-ups. The authors use a two-step approach to link market competition and innovation has so far been product market reforms and macroeconomic relatively sparse and inconclusive. The reasons lay in the performance. They first identify the link between poor availability of comprehensive time series of indicators of product market reforms and economic rents product market indicators, in a “still-in-progress” measured by mark-ups. In a second step, they use the theoretical framework, and in the difficulties of predicted mark-up to assess the effect on measuring dynamic efficiency given that it takes time to macroeconomic variables. The authors relate R&D deliver its full effects and that innovation is difficult to expenditures with the predicted mark-up from the first measure. regression (indirect effect), its squared value, and policy

As surveyed by Ahn (2002, p.15), studies on the indicators (direct effect). Their results suggest a nonrelationship between market power and innovation lead linear relationship between competition and the levels of to mixed results. For example, some studies show that R&D expenditure. However, they find an inverted-U- companies’ size has no significant effects on innovation shape relationship between mark-ups and R&D that only whilst other studies point to either a positive relationship turns downwards at high levels of regulations. Direct between concentration and innovation, or an inverted U- effects of regulation appear to be stronger – although Shaped relationship, or simply no effects when with a negative sign.

controlling for industry differences. Apparently, measurement and modelling issues blur empirical results as good proxies for innovation are difficult to find and regression methods fail to take into account “bounds”

effects between R&D intensity and concentration. 326 Acs

with the identification of the intensity of price competition and the level of endogenous sunk costs as the key

325 Agency costs may be defined as costs induced by decisions determinants. For example, R&D can allow firms to

taken by managers with the view to increasing their differentiate their products and therefore more R&D can personal gratification or to reaching their own personal lead to less concentration.

objectives (use of “free cash-flow”) instead of being taken 327 See EU ECONOMY 2003 REVIEW , chapter 2. One important in order to maximise the net present value of the firm. determinant of innovation is skills and education as Agency costs may also stem from the existence of suggested by Rao et al. (2002) and Aghion and Cohen

managerial or organisational slack translating into a misuse (2004). Griffith and Simpson (2003) found that foreignof

human resources. owned manufacturing firms in Britain have higher levels of

326 The “bounds approach” has been developed by John Sutton labour productivity and investment per employee. Their

(2002). Sutton looks at the determinants of market results suggest that the higher proportion of skilled workers concentration and finds lower bounds for concentration. His in foreign-owned industries matches differences in labour

work attempted to connect the analysis of concentration productivity.

Table 1: Summary of the main empirical results

Channel Main empirical results

• Product market reforms usually reduce economic rents (mark-ups). • Product market reforms have substantial impact on entry.

Allocative efficiency • Productivity gains are mainly due to reorganisation within the firm,

except in high-tech industries where new firms contribute the most to productivity gains.

• Increase in competition is associated with increase in technical

Productive efficiency efficiency.

• Product market competition reduces agency costs. • Evidence of an inverted U-shaped relationship between innovation

and competition.

•

Dynamic efficiency More competition usually leads to TFP growth but with long lags. • Creative destruction accounts for most of the increase in TFP growth

rates.

• Distance to technological frontier matters.

The direct link between competition and dynamic manufacturing firms in Taiwan, Wang and Tsai (2003) efficiency as measured by productivity growth rates found that whilst average output elasticity stood at .18, seem to be clearer. Nickell (1996) found a positive this effect was larger in high-tech firms. In addition, the impact of competition on firm-level TFP growth and effects of R&D on TFP performance may appear with Disney et al. (2000) found that competition is an long lags and investment in ICT could even be important determinant of internal restructuring, which in associated with lower TFP in the short-run as resources turn has an impact on TFP growth. In terms of relative and energy are diverted to reorganisation and learning as

importance, the authors distinguish between 'internal' suggested by Basu et al. (2003). restructuring (i.e. new technology and organizational

change) and 'external' restructuring (i.e. entry of Recently, the literature has underlined differential efficient firms and exit of least efficient ones) and find effects of innovation on productivity growth depending that 'external restructuring' accounts for 90 per cent of on the distance to the technological frontier. As TFP growth. Griffith and Harrison (2004) also find a mentioned earlier, Nicoletti and Scarpetta (2003) use an non-linear relationship between competition and the endogenous growth model to integrate productivity growth rate of labour productivity or total factor growth and catch-up so that product market regulations productivity. When looking at the evolution of will both directly and indirectly (through interacting competition and the indicators of macroeconomic them with the distance to the technological frontier) performance within countries, the authors find a impact TFP growth. They found a positive impact of

negative relationship (i.e. more competition decreases entry on TFP growth, especially in services.

performance). However, this finding has to be balanced These gains seem to be larger the further an economy by possible measurement errors and lag effects. Indeed, lies away from the technological frontier. Aghion et al. when comparing across countries the authors find the (2003b) looked at the effects of reforms on trade in India expected positive relationship (i.e. countries with more and found the opposite effect. The authors found that

competition have better performance). 328 Several studies Indian States that were close to the technological

also found a significant elasticity of R&D to tax credit, frontier and had liberalised labour markets enjoyed a

329

even more so in the long term. positive impact of trade liberalisation on growth, whilst

A positive effect of innovation on dynamic efficiency the opposite holds for Indian States that lay far from this

finds additional empirical support. Rao et al. (2001) find frontier.

a strong correlation between innovation and TFP growth Finally, R&D expenditures have an impact on in Canadian manufacturing industries. The R&D output productivity through two channels: innovation and

elasticity could however depend on sectors. Looking at imitation of other’s discoveries. 330 . The diversity of

these findings suggests that, although positive, the

328 impact of competition on innovation and productivity Technically, the difference between the two techniques

relates to the presence or not of ‘country fixed effects’ takes complex forms.

variables that are there to capture country-specific features which are not observable but may explain better performance. The authors do not just look across countries because there is always a theoretical risk that countryspecific non-observable features other than the level of competition may impact on macroeconomic performance and that this impact could be wrongly attributed to the level of competition.

329 See European Commission (2002a) for a discussion. 330 See Griffith, Redding and Van Reenen (2000).

4.4 Direct and indirect effects of product In particular, the chapter acknowledges additional

market reforms. dynamic effects of product market reforms and stresses

in particular the role of knowledge production in the

The EU Economy 2003 review, (chapter 2) identified EU’s structural productivity problems. Other exercises the regulatory environment as a key determinant of which have attempted to take these indirect effects into

capital deepening 331 , indicating an impact on account find indeed large indirect effects of product

productivity levels. It was estimated that the implied market reforms. Some of these results are presented change of moving to US levels of regulation would hereafter.

suggest a long-run productivity effect of about 5 per cent

(i.e. 0.15p.p. on the long run growth rate of In 2002, the EU ECONOMY 2002 REVIEW (chapter 2) productivity). These gains were mostly static and presented a scenario with labour and product market occurred through increased investment. This year, the reforms that included a reduction in the NAIRU by 1.5 chapter on “The Lisbon strategy and the EU’s structural per cent, a reduction in price mark-ups by 0.5 percentage productivity problem” shows that the direct effects of points (p.p.) and an increase in total dactor productivity deregulation on productivity are relatively weak. The level of 1 per cent. This “big bang” scenario is in line results show that deregulation could lead to a meagre with the observed effects of recent labour and product 0.15 percent increase in the rate of productivity growth. market reforms. That experiment led to an increase in This should be compared with a 0.05 percentage point GDP of about 4 per cent and an acceleration of output effect of a 1 per cent increase in investment, a 0.60 growth by about 0.5 percentage points annually over 7-8 percentage point effect of a permanent 1 per cent years. Using the same framework but restricting it to increase in R&D spending, and a 0.45 percentage point product market reforms, Dierx, Pichelmann and Röger effect of a permanent one year increase in average (2004) found a medium-term increase in GDP relative to education levels of the labour force. However, by its baseline level of about 2 per cent and an acceleration looking at the effect of R&D on productivity, the of output growth by almost a quarter of a percentage

analysis in some way shed light on the possible indirect point annually over a period of seven to eight years.

effects which arise from product market reforms on total

factor productivity. In its report “World Economic Outlook 2003”, the IMF (2003) simulated the impact of closing the gap with the

USA in terms of labour and product market reforms, which points to a potential 10 per cent GDP increase.

331 Needless to say, it also influences technical progress.

Table 2: Estimates of the quantitative effect of product market reforms

Initial channels Size Timing Additional GDP

TFP increase + 0.5% level

Price mark-up

decrease with SMP -0.9 p.p. for economy as

EU a whole 10 years of SMP

Combined effect of SMP is 1.8%

Price mark-up

decrease in -0.5 p.p. for economy as

Network Industries a whole

TFP increase + 1% level

Dierx, Pichelmann Price mark-up -0.5 p.p. for economy as Seven to eight and Röger decrease in a whole year

+2%

Network Industries

Price mark-up Long-term

IMF decrease -10 p.p. for economy + 4.3% in long-run (not specified)

Product market Euro mark-up goes from Long-term

Bayoumi et al. reforms in the form 1.35 to the US level of +8.6% in long-run of lower mark-up 1.23 (steady state)

Source: Commission services (2002b), Dierx et al. (2004), IMF (2003), Bayoumi et al. (2004).

increase in entry rate, an improvement in the efficiency with which firms are managed and a stimulus for firms to innovate. These indirect effects appear to be potentially much higher than the direct effects. The next question is which product market reforms are the most pressing to improve the productivity performance of the

If we focus on product market reforms only, the European Union, in particular in relation with the USA. message of the IMF is that the long-run effects of more To answer this question, we need to investigate the competition-friendly policies on product markets in the relative importance of the three transmission channel euro area would be substantial with GDP increasing by and to identify the areas where the EU is significantly 4.3 per cent in the long run (along with a consumption lagging behind the USA. While further work, i.e. the increase of 3.4 per cent and an increase in investment by construction of a model putting together the three 12.1 per cent). channels and analysing their interactions, is necessary to

Using a similar methodology, Bayoumi, Laxton and address the first issue, some interesting conclusions can Pesenti (2004) find an even larger value of an increase already be drawn regarding the second issue.

in GDP of 8.6 per cent if product market reforms lead to To that end, we have compared indicators of product a mark-up similar to the US level. The gain goes up to market reforms in the EU and the USA so as to assess 12.4 per cent of GDP if labour market reforms are whether differences in performance originate from added. As a comparison, the European Commission differences in the economic framework. It is an (2002c) finds a combined effect of the Single Market established fact that the EU is lagging the USA in terms Program of 1.8 per cent of GDP after 10 years. Although of GDP per capita. A common perception is one of the comparing the results is made difficult because of United States enjoying large economic freedom whilst differences in methodologies, the outcome suggests the EU is stuck in its red tape and heavy regulations. substantial benefits. This argument fails to take into account the fact that the

These results suggest that while deregulation directly European Union has initiated profound reforms (see affects productivity, its real potential effects may lie in Section 1) and that the United States still have several indirect effects via the three channels that we have segments of its economy that are regulated and sheltered

identified. from competition. The first step is to have summary indicators of product market reforms and the second is

• 5. 

  Is the EU lagging significantly behind to analyse to what extent they differ in the EU-15 332 and

  the USA in terms of product market

reforms? 332 There are various reasons to focus on the EU-15 and not on new Member States. For example, most empirical results in

The indirect effects of product market reforms on the literature focus on the EU-15. The same goes for the

productivity operate through a reduction in mark-up, an estimates, which may depict the relationship between product market reforms and productivity in an institutional

the USA. Then, one should look at the potential gains of reforms in Europe.

5.1 Measurement of product market reforms in the EU and the USA

Summary indicators of product market reforms are

provided by three main sources: 333

• First, the OECD database on regulatory reforms

contains indicators of economy-wide regulation (eg. State control), of industry-level regulation (eg. Barriers to trade in manufacturing) and of regulatory reforms (eg. Trade liberalisation). These indicators date from around 1998 and the OECD launched a project aimed at updating them by October 2004.

• Second, composite indicators are also available

from the Fraser Institute. They refer to business regulations and identify the extent to which regulatory restraints and bureaucratic procedures limit competition and the operation of markets. In addition, Fraser provides indicators on the freedom to trade with foreigners and on the State’s involvement in the Economy. The latest available report contains data for 2002.

• Finally, the World Bank’s database “Doing

Business” provides indicators on the cost of doing business by identifying specific regulations that enhance or constraint business investment, productivity, and growth.

On the basis of information gathered by these three sources, four main categories of indicators can be used to measure and compare the intensity of product market reforms across countries: ease of starting a new business, trade openness, state involvement in the economy and administrative burden on business.

and policy framework that is different from those of the new Member States.

333 The structural indicators produced by the European

Commission do not provide per se regulatory indicators but rather performance indicators such as prices or market share of incumbents.

Table 3: Components of the economic freedom indicator

Of which Of which Of which Of which General I. II. III. IV.

2002 economic Legal structure freedom Size of gov. and security of Access to sound Freedom of exchange with property rights money foreigners

USA 8.2 7.4 8.2 9.8 7.8

Av. EU-15 7.3 4.6 8.1 9.6 8.2

Source: Fraser Institute. The indicators range from 0 to 10, a higher value indicating a better score. EU-15 is the GDP-weighted value.

5.2 Taking stock of product market reforms Although a comparison in time is difficult because of in the USA and the EU changes in definitions and in the number of subindicators

 used to create the composite index, globally,

To compare the United States and the European Union, the main result is that economic freedom constantly

it is convenient to start with the Fraser indicator on

334 appears to be higher in the United States than in the EU-

general economic freedom. This indicator indeed 15 whatever the period considered.

gathers information about five major areas, combining

regulations in product, labour and capital markets, legal Interestingly, when looking into more details at the structures and security of property rights, the indicators, the difference between the EU and the USA involvement of the government in the economy, the is strongest in terms of the degree of involvement of the access to sound money and the freedom of trade. State in the economy, with the EU also trailing the USA

in terms of regulations of credit, labour and business. By Graph 1: General economic freedom (2002) contrast, the EU is close to the USA in the field of

access to sound money (which mainly include indicators

USA 8.2

EU-15 7.3 of financial stability) and legal issues, and slightly leads

the USA when it comes to freedom of international

FR 6.8

EL 6.9 exchanges. Since 1990, the main evolution has been a

IT 7 decrease of the indicator of the size of the government

ES 7.1

PT 7.2 for the EU-15 compensated by increased freedom of

SE 7.3

DE 7.3 exchange with foreigners. The general index of

BE 7.4 regulation has slightly improved on both side of the

AT 7.5

DK 7.6 Atlantic, although by less in Europe (see Table 3).

NL 7.7

FI 7.7 Among regulations, differences in labour markets

LU 7.8

IE 7.8 flexibility are the most glaring, with all EU economies

UK 8.2 lagging behind the USA. In particular, the low German

0 1 2 3 4 5 6 7 8 9 score burdens the EU weighted average. Product market

Fraser index 0-10 regulations on businesses also appear to be higher on the

Source: Fraser Institute. old continent, as regulations on businesses are lower in the United States than in any European Member States,

bar Finland and Sweden.

335

In 2002 – the most recent available data – the USA Looking in more depth at countries, we have plotted the was above all European Member States with a value of labour market regulation index and the business 8.2 – at par with the United Kingdom. Our computed regulations index for 2002 (see Graph 2). The GDP-weighted average for the EU-15 stood at 7.3 (see presumption is that one would see two types of

336

Graph 1). countries: low regulated ones and heavily regulated ones

in both labour and business regulations. The plot reveals a more complicated story because the levels of labour

334 The OECD indicator probably encompasses much more and business regulations come out as not correlated.

337 .

aspects of regulation but it dates from 1998 and is often a

‘one-shot’ measure. Member States in total EU GDP. The same goes for most 335 Worldwide, the US came third behind Hong Kong and indicators.

Singapore. It is at par with New Zealand, Switzerland and 337 Both the Spearman and Pearson tests fail to reject the null the UK. hypothesis of no correlation. However, when one takes

336 Although new Member States generally displayed lower Austria, Germany, Finland and Sweden out of the sample,

scores, the EU-25 GDP-weighted average came close old the correlation appears highly positive and strongly Member States’ value because of the low share of new significant. The same result is valid when taking the OECD The first interesting, albeit unsurprising, finding is that Finally, Germany stands out as outliners with relatively the USA, unlike most EU countries achieve high good performance in business regulations but has too performance in both indicators. To regroup countries, inflexible labour markets. Given its weight in EU GDP,

we carried out a hierarchical cluster analysis. 338 Being its unsatisfactory performance bears on the overall EU

the closest to the USA, Luxembourg, and the UK index. succeed to achieve relatively high performance on both

indicators. At the other extremity, the cluster composed To investigate further those differences, we have

339

of Greece, Italy and, to some extent, Belgium and regrouped indicators within the four categories of

Portugal shows poor performance in both labour and product market reforms identified in Section 1.

business regulations. In the middle range, two groups Table 4: Regulation indices

emerge.

2002 Regulation Regulation Regulation

Graph 2: Labour and business regulations 2002 of credit of labour of business

USA 9.2 7.3 6.7

8 Av. EU-15 8.3 4.4 5.8

FI

7 Source: Fraser Institute. The indicators range from 0 to 10, a higher

SE US

LU value indicating lower regulations. EU-15 is the GDP-weighted

AT DK

DE UK value.

6 NL IE

ES

ns BE PT FR

la tio 5 IT 5.2.1 Trade openness and legal barriers EL

re gu s Based on this indicator, the EU appears to be a slightly

in es 4 more opened economy than the USA. Most of the

B us difference seems however to come from differences in

3 taxes on trade and hidden import barriers, whilst the cost

2 of importing is slightly lower in the United States.

2 3 4 5 6 7 8 Indicators of freedom of exchange have remained fairly

Labour regulations stable since 1990, with most pronounced increase for the new Member States. In addition, most EU Member

Source: Fraser Institute. States show very close values when it comes to the subindicators.

 However, the Fraser values reflect

The first one is composed of France, Ireland, The international trade and tell nothing about trade Netherlands, and Spain. It achieves good results in integration within the EU and the USA. Trade labour regulations and average performance in business integration among Member States was one of the first regulations. 340 aims of the European Union with the creation of a

Custom Union and, subsequently, the creation of a

The second group achieves good if not superior Single Market and a Common Currency Area. Between performance in business regulations but generally shows 1999 and 2002, intra-EU export trade in products for the average to poor performance in labour regulations. It is 15 Member States varied between 15.7 per cent and 17.3 composed of Austria, Denmark, Finland, and Sweden. per cent of GDP. Similar data for the intra-state trade in

the United States are not directly available. However,

indicator (for 1998). Further investigations should be done using recent 2001 data for exports of manufactured with regards to differences in the labour market index. The goods from manufacturing establishments per State. 341

OECD index seems to focus more on hiring and firing (US Department of Commerce and US Census Bureau,

issues, whilst the Fraser index looks in addition at minimum 2004), we subtract the value of direct manufactured

wages, collective bargaining, and unemployment benefits exports from the value of all manufacturer’s shipments issues. to get intra-state US exports of manufacturing product

338 Based on Ward’s minimum variance method. We arbitrarily instead. We then divide this value by US-GDP.

stopped the clustering procedure at five clusters, whose regrouping still explains 80.6 per cent of the original variance. Technically, the distance between two clusters is the ANOVA sum of squares between the two clusters added up over all the variables. At each generation, the within 341

cluster sum of squares is minimised over all partitions Note that in this case, the definition of "manufacturing obtainable by merging two clusters from the previous shipments" includes what the Census bureau calls intergeneration. plant transfers. This is the value of goods shipped to another

339 establishment owned by the same company even if no The relatively poor performance of Belgium in the index of

explicit sale occurs due to the common ownership. The

business regulation mainly stems from a poor performance Census bureau asks respondents specifically to include an in the ease of starting a new business. estimate of the market value of such shipments in their

340 France seems to obtain a good score in labour market value of total shipments. Note also that the value at the

regulations thanks to a relatively decentralised wage plant does not include transportation or trade margins

bargaining system. (purchasers’ prices less producer prices).

The aggregate value for the USA stands at 34.0 per cent. operate a limited liability company (see Table 7). It For comparison purpose, we computed the 2001 EU-15 indicates that, although the number of procedures may intra-EU exports of manufactured products as have declined, their duration and cost remain percentage of GDP. Its value stood at 16.6 per cent in substantially larger than in the United States for most

2001. Globally and with all the necessary caution 342 , this EU Member States.

result suggests that, even when correcting for

measurement errors, the USA appears to be a more In addition, the World Bank provides indications on the

integrated trade area than the EU. time and cost necessary to close a business (see Table 8). Swift and inexpensive death of inefficient

businesses is also important to increase overall 5.2.2 Regulations on entry productivity. Despite shorter procedures, the cost of

Free entry and exit of firms is a key element for the closing a business in Europe remains higher than in the process of enabling the reallocation of resources towards USA. In addition, with the exception of Finland, the the most productive sectors and firms, forcing Netherlands, Belgium and Ireland, the European companies to reach more efficient ways of doing insolvency systems lack efficiency in terms of cost, business and giving incentives to firms to innovate. time, priority of claims, and outcome achieved. The Tight regulations to create a business, numerous and Fraser indicators on obstacles to new businesses are lengthy procedures to set up a company, or high costs to difficult to compare over time because their definition start an economic activity are deterrents to has changed, leading to jumps in the data. Looking at entrepreneurship and end up to be de facto regulatory data for 2002, we link the indicators of “ease of starting protections for incumbent companies. Obstacles to the a new business” and “administrative obstacles to start a

set up of new businesses appear to be relatively high in new business”.

343

the EU, despite the fact that Denmark, Finland, Ireland, Unsurprisingly, both indicators are highly correlated and Luxembourg Sweden and the UK have indices that contain similar information (see Graph 4). When

come close to or are higher than the US value. This proceeding with the clustering procedure, 344 one can

result is extremely important because the economic identify Finland as a front-runner that displays obstacles literature suggests that potential competition is a to entrepreneurship that are as low if not lower than the

necessary condition for the channels between product USA. market reforms and productivity to work.

Graph 3: Intra trade in manufactured products (intra Graph 4: Obstacles to entrepreneurship 2002

exports a % of GDP 2001)

40 7

35 ew n 6 FI 30 rt

 a ta

to s 5 SE AT

25 le s ss DK LU

US

D P st ac 4 IE

f G 20 ob si

ne

DE

 o e bu UK % 15

ra tiv 3 PT ES NL

10 io

ni st IT

2 EL

A dm FR BE

5 1

0 3 4 5 6 7 8 9 EU15 United States Ease of starting a new business

Source: Commission services. Source: Fraser Institute.

The Fraser indicator assesses the situation in 2002. In the course of the Lisbon strategy, many if not all Member States have initiated important reforms to promote entrepreneurship. The World Bank provides

more recent and complementary data in this respect. The 343 Both indicators are taken from the Global Competitiveness indicator lists all procedures legally required to legally Report. They represent the business’ assessment on whether

starting a new business is generally easy and whether administrative procedures are an important obstacle to

342 Notably, the level of desegregation can bias the results, the starting a new business, respectively.

data sources and methodologies differ and this analysis is 344 We arbitrarily identify five clusters, whose regrouping still

only made for 2001. explains 84.2 per cent of the original variance.

Table 5: Components of the freedom of exchange with foreigners indicator

Of which Of which Of which Regulatory Regulatory I. II. III. barriers of barriers of which: which: III.b.

III.a. Freedom of

exchange Restrictions Taxes on Regulatory Hidden Cost of importing 2002 with on capital int'l trade trade import

foreigners markets barriers barriers US 7.8 8.4 8.1 8.2 6.8 9.6

Av. EU-15 8.2 8.7 9.0 8.6 7.8 9.3

Best EU-15 9.0 (IE) 9.6 (LU) 9.1 (EL, LU) 9.5 (FI) 9.3 (FI) 9.9 (LU)

Worst EU- 7.4 (EL) 7.2 (PT) 9.0 (all 7.8 (IT) 6.7 (IT) 9.0 (IT) 15 others)

Source: Fraser institute. Range: 0 to 10 from most restrictive to less restrictive. Hidden import barriers originate from the Global Competitiveness Report published by the World Economic Forum. The cost of importing is defined as the combined effect of import tariffs, licence fees, bank fees, and the time required for administrative red-tape raises costs of importing equipment and share the same source. EU-15 is the GDP-weighted value.

Table 6: Obstacles to entrepreneurship Table 7: Starting a business

2002 Regulation Administrative Ease of Starting a on business obstacles for starting a business Num. of Duration Min.

new businesses new (January procedures (days)

Cost * Capital * business 2004)

USA 6.7 4.0 8.0 USA 5 5 0.6 0

Av. EU-15 6.0 3.0 5.3 Av. EU-14 7 31 10 39

Best EU-15 7.5 (FI) 6.2 (FI) 7.7 (FI) Best EU-14 3 (FI) 4 (DK) 0 0 (DK) (IE,UK)

Worst EU-

15 4.9 (EL, IT) 1.8 (BE, FR) 3.7 (FR) Worst EU-

14 13 (EL) 108 (ES)

23.3 135.2 (IT) (EL)

Note: * EU-24 (minus CY). EU-15 and EU-25 are GDP-weighted

values. Note: * % GNI per capita.

Source: Fraser institute. Range: 0 to 10 from most heavy regulation Source: World Bank (methodology adapted from Djankov et al. to lowest regulation. 2002). EU-14 (minus LU). EU values are non-weighted averages.

Table 8: Closing a business

Closing a business (January

2003) Actual time (years) Actual cost (%of estate) Goals-of-insolvency index*

USA 3 4 88

Av. EU-14 1.8 8.9 73

Best EU-14 0.4 (IE) 1 (FI, NL) 99 (FI)

Worst EU-14 4.2 (DK) 18 (AT, FR, IT) 42 (EL)

Note: * “The goals of insolvency ratio documents the success in reaching the three goals of insolvency, as stated in Hart (1999). It is calculated as the simple average of the cost of insolvency (rescaled from 0 to 100, where higher scores indicate less cost), time of insolvency (rescaled from 0 to

100, where higher scores indicate less time), the observance of absolute priority of claims, and the efficient outcome achieved. The total Goals-of Insolvency Index ranges from 0 to 100: a score 100 on the index means perfect efficiency (Finland, Norway, and Singapore have 99), a 0 means that the insolvency system does not function at all”. (Source: World Bank).

Source: World Bank (methodology adapted from Djankov et al. 2003b). EU-14 (minus LU); EU values are non-weighted averages.

The second group encompasses Austria, Denmark, interpreted by producers as “lines in the sand” and they Ireland, Luxembourg, and Sweden and shows a relative could tacitly agree to sell at the maximum price although

ease of starting a new business with few perceived they would be able to supply at a lower price. 348 As one

administrative obstacles. The third group, made out of believes that price controls have a distortionary impact The Netherlands and the United Kingdom, exhibits an on economies, the higher degree in Europe comes as bad

equal perceived ease of starting a business but more news.

perceived administrative obstacles. 345 As “average”

performers, Belgium, Germany, and Spain show Irregular payments to official (as well as the size of the indicators that are close to the EU-15 average with underground economy) are equally bad for productivity Germany slightly better and the other two doing more because they bias competition. Productive firms may be poorly. Finally, the fifth group exposes a strong unease driven out by less productive ones simply because they to start up a new company and very high administrative do not compete on the same level of the playing field. burden related to this issue. It consists of France, The Fraser indicator does not indicate a significant

Greece, Italy, and Portugal. difference between the EU average and the US level for irregular payments to officials.

5.2.3 Business-friendly environment. A business-friendly environment also encompasses a

Besides indicators on the ease to set up a new business, level playing field for competitors and sound and certain the index of regulations on business includes an rules of law. Competition policy plays an important role indicator on the time spent with bureaucracy, one on in this respect. Although the EU and US competition price control and another one on irregular payments to laws have many common features, some commentators officials. On average, and contrary to common believe, have remarked on apparent differences in the underlying the time spent with bureaucracy in the European Union philosophies, which allegedly have a significant appears to just slightly higher than in the United States. influence on the outcomes of competition cases. For Although the Fraser indicators come out to be very close example, the US authorities are said to be more for all Member States, the analysis reveals that Spanish, concerned with the efficiency effects of business Belgian, Italian, Luxemburg, and Swedish companies practices or mergers, whereas the EU's approach places are those for which senior management spend the least greater emphasis on market structures and the impact on

time dealing with government bureaucracy. competitors.

As in the case for several other indicators, the results Graph 5: Time spent with bureaucracy 2002

should be taken with caution as they are based on

346

businesses’ perception - which can vary with business

culture - but it gives an interesting indication that 8

bureaucracy might not be the key explanatory variable 7

of the gap between the United States and Europe. 347 The y

results seem to be somewhat confirmed by the World ac

6 cr

Bank indicator on enforcing contracts which looks at the re au 5

procedures necessary to recover a debt and shows a t bu as 4

similar number of procedures in Europe and the USA le

combined with a longer duration in the USA. The cost t

to 3 os

and, to some extent, the complexity of the procedure m 2 seem however lower in the United States. f ro

m 1

0- 10 Price control is usually used to protect citizens from 0

large price increases on basic product that are deemed ES BE IT LU SE DE FR IE AT GR NL DK UK PT FI EU15 necessary. However, price control can sometime by Source: Fraser Institute.

345 Graphically, groups 2 and 3 seem to be good candidates for However, recent reforms in the EU, particularly with

a regrouping under a label “medium-high” performers. regard to agreements between companies, are likely to

However, formally, the cluster analysis would favour first a

regrouping of groups 4 and 5 with a slightly lower share of reduce some of the differences. For example, it is

the explained original variance. probably true that, in the past, the European

346 In addition, time spent with bureaucracy could conceptually Commission devoted too much effort to policing be beneficial if this allows public authorities to make better relatively innocuous agreements and not enough to

decisions and actions, for example because of better detecting and breaking up hardcore cartels. A series of information. radical legislative changes enacted since 1999 are

347 To be fair, the indicators on regulation of entry and

administrative burden are purely describing the domestic situation. They do not measure the difficulties of creating a business in another EU Member State, nor to deal with an

EU foreign administration. 348 See European Commission (2001a).

Table 9: Regulation on business and sub-indicators

2002 Regulation on Price control Time spent with Irregular payments to business bureaucracy government officials

US 6.7 7.0 6.8 8.0

Av. EU-15 6.0 6.3 6.5 8.0

Best EU-15 7.5 (FI) 9.0 (FI) 7.3 (ES) 9.5 (DK)

Worst EU-15 4.9 (EL; IT) 5.0 (BE; IT) 5.3 (FI) 5.9 (EL)

Source: Fraser institute. Range: 0 to 10 from most heavy regulation to lowest regulation. EU-15 is the GDP-weighted value.

Table 10: Enforcing contracts

Enforcing contracts Number of Duration (days) Cost Procedural complexity January 2003 procedures (% GNI per capita) index

US 17 365 0.4 46

Av. EU-14 19 225 5.9 55

Best EU-14 12 (UK) 39 (NL) 0.5 (NL; UK) 36 (UK)

Worst EU-14 27 (BE) 645 (IT) 15.8 (FI) 83 (ES)

Note: The Procedural Complexity Index varies from 0 to 100, with higher values indicating more procedural complexity in enforcing a contract. This index measures substantive and procedural statutory intervention in civil cases in the courts.

Source: World Bank (methodology adapted from Djankov et al. 2003a). EU-14 (excluding LU).

Table 11: Size of government

2002 Size of Government Transfer and Government Top marginal income Government consumption subsidies enterprises and tax rate

investment US 7.4 5.5 6.7 10.0 7.5

Av. EU-15 4.6 4.1 4.2 6.8 3.5

Best EU-15 6.8 (UK) 6.2 (EL) 6.3 (UK) 10.0 (AT; BE; DK; 6.0 (UK) IE; NL; UK)

Worst EU-15 2.8 (FR) 1.0 (SE) 2.3 (DE) 4.0 (ES; FR) 0.5 (DK)

Source: Fraser institute. Range: 0 to 10 from highest to lowest involvement.

EU-15 is the GDP-weighted value

designed to enable the Commission to redirect resources that network industries in the EU are liberalised to a to the most serious problems and to introduce a more degree similar- if not superior - to that in the USA, in economics-based approach to competition policy. It particular in the energy and postal services. remains to be seen what impact these changes will have.

As far as state aid is concerned, the EU exercises a

control over the Member States that has no equivalent in 6. Policy implications

the USA. One result of this is that there is much more

transparency about aid expenditure in the EU. Another The previous section showed that the backwardness of outcome is that competition between Member States and Europe in product market reforms seems to be regions to attract investment is strictly disciplined in the concentrated in measures that promote entry and exit of EU, whereas states and local authorities are engaged in firms and in a lower degree of trade integration. The an escalating subsidy war in the USA, the overall effect European Union is opened to international competition of which is probably welfare-reducing. On the other and its network industries are liberalised to a degree that hand, national authorities in the EU seem to be much equals if not exceeds the United States. If European more willing than authorities in the USA to give companies do not perceive regulations as more timefinancial support to ailing firms or sectors, while US consuming than US companies do, their cost seems to be authorities are more likely to take a forward-looking higher. Finally, State’s involvement in the economy is approach, targeting firms with good growth prospects. higher in Europe but the consequences of this are

debatable. Obviously, a lack of flexibility in labour

Finally, corporate tax levels in the USA do not seem to markets and to some extent more regulations on credit – be lower than in the EU. One main difference between two issues not reviewed here – may also explain a

the EU and the United States may be the additional cost 349

for European companies having cross-border activities sizeable share of the US-EU gap in productivity.

of dealing with 25 different accounting and tax systems. Europe’s poor performance in promoting business Recent surveys show that companies face important dynamism may actually well explain its lower problem and compliance costs related to transfer pricing productivity as the theoretical and empirical findings issues and refund of VAT across Member States. have shown entry and exit of firms as an important if not

necessary condition for product market reforms to deliver their full effects via the three channels. The

5.2.4 State’ involvement in the economy relatively poor performance of the EU in terms of trade The last category of product market reforms concerns integration could seem at first more surprising given the those reforms aiming at reducing the State’ involvement efforts made to create an Internal Market. in the economy. All indicators, be they the government However, despite its many successes, the Internal consumption, the level of transfer and subsidies, the Market is not functioning as it should and some key level of taxes, or the size of State participations in indicators of Internal Market integration (such as growth enterprises point to a smaller government intervention in in trade amongst the euro area Member States and price the USA than in Europe (see Table 11). It is nevertheless convergence) show that progress has stalled. Similarly, difficult to univocally depict all governmental recent Internal Market scoreboards have highlighted an intervention with economic distortions. The indicators increase in the transposition deficit of Internal Market do not pick up efficiency issues, social preferences, directives and substantial delays in the transposition of differences in the organisation of the welfare state, or these directives into national legislation.

the extent to which government intervention tries to fix

market failures. In the context of the mid-term review of the Lisbon

strategy, our analysis can contribute to the choice of priorities for reforms in the area of product markets. We

5.2.5 Liberalisation of network industries conclude that reforms to ease entry and exit are

Finally, comparing the liberalisation process of network important. These should go beyond measures to reduce industries in the EU and the USA brings interesting time and cost to start up a company and should include insights. This analysis is not based on a summary reforms making Europe an attractive place to do indicator but on an analysis of the changes in the business. Similarly, making sure that the internal market regulatory framework observed in the EU and the USA. is working at full capacity should be a clear objective for Network industries make up for an important share of the Union.

the economy with around 5 per cent of total employment in both the EU and the USA. In addition, they provide

services that are economically and socially important to 349 We do not review here other explanations such as

households and business users. In Europe, large management practices, IT spending, innovation, education, productivity gains have accompanied the liberalisation etc. See Lewis (2004), Lewis et al. (2002), Dorgan and

of network industries (European Commission 2004a). Dowdy (2002) and McKinsey (2001) for some of these By comparing the degree of liberalisation of network issues. For example, there are indications that relative

industries in the EU and the USA (see annex), it appears “management scores” match relative total factor productivity levels.

References

A CS , Z.J. AND B.D. A UDRETSCH (1987), “Innovation, market structure and firm size”, Review of Economics and statistics, 69, pp. 567-575.

A GHION , P. AND P. H OWITT (1992), “A model of growth through creative destruction”, Econometrica, 60, pp. 323- 351.

A GHION , P., H ARRIS , C. AND J. V ICKERS (1997), “Competition and growth with step-by-step innovation: an example”, European Economic Review, Papers and Proceedings, pp. 771-782.

A GHION , P. AND P. H OWITT (1998), “Endogenous Growth Theory”, MIT Press, Cambridge and London.

A GHION , P., H ARRIS , C., H OWITT , P. AND J. V ICKERS (2001), “Competition, imitation and growth with step-by-step innovation”, Review of Economic Studies, Vol. 68, pp. 467-492.

A GHION , P., B LOOM , N., B LUNDELL , R., G RIFFITH R., AND P. H OWITT (2002), “Competition and innovation: an inverted U relationship”, NBER Working paper, No 9269, October 2002.

A GHION , P, B LUNDELL , R., G RIFFITH , R, H OWITT , P, AND S. P RANTL (2003A), “Firm entry and growth: theory and micro evidence”, Mimeo.

A GHION , P., B URGESS , R., R EDDING , S, AND F. BILIZOTTI (2003 B ), “The Unequal Effects of Liberalization: Theory and Evidence from India”, mimeo, http://econ.lse.ac.uk/staff/rburgess/wp/abrz031002.pdf .

A GHION , P. AND E. C OHEN (2004), “Education et croissance“, Rapport pour le Conseil d’Analyse Economique, Paris.

A HN , S. (2001), “Firm dynamics and productivity growth: a review of micro evidence from OECD countries”, OECD Economics Department Working Papers, No 297.

A HN , S. (2002), “Competition, innovation and productivity growth: a review of theory and evidence”, OECD Economics Department Working Papers , No 317.

B AILY , M.N., H ULTEN , C. AND D. C AMPBELL (1992), “Productivity dynamics in manufacturing plants”, Brookings Papers on Economic Activity: Microeconomics, pp. 187-267.

B AIN , J.S. (1951), “Relation of profit rate to industry concentration: American manufacturing 1936-1940”, Quarterly Journal of Economics, 65, pp. 293-324.

B AIN , J.S. (1956), “Barriers to new competition”, Harvard University press, Cambridge MA.

B ARNES , M., J. H ASKELL , AND M. M ALIRANTA (2001), “The sources of productivity growth: micro-level evidence for the OECD”, Paper presented at the OECD workshop on firm-level statistics, OECD, November 2001, Paris.

B ASU , S., F ERNALD , J. G., O ULTON , N., AND S. S RINIVASAN (2003), “The case of missing productivity growth: or, does information technology explain why productivity accelerated in the United States but not in the United Kingdom?”, NBER Working paper, No 10010, Cambridge, MA.

B AYOUMI , T., D. L AXTON AND P. P ESENTI (2004), “Benefits and spillovers of greater competition in Europe: a macroeconomic assessment”, European Central Bank Working Paper Series, No 341.

B ILS , M. (1987), “The cyclical behaviour of marginal cost and price”, American Economic Review, Vol. 77, pp. 838- 885.

B LANCHARD , O LIVIER AND G IAVAZZI (2001), “Macroeconomic Effects of Regulation and Deregulation in Goods and Labor Markets”, MIT Department of Economics Research Paper Series, No 01-02, January 2002.

B LUNDELL , R.W., R. G RIFFITH AND J. V AN R EENEN (1999), “Market share, market value and innovation in a panel of British manufacturing firms”, Review of Economic Studies, No 66, pp. 529-554.

B RANDT , N. (2004), “Business dynamics, regulation and performance“, OECD STI Working papers, No 2004/3.

C AVES , R.E. AND D.R. B ARTON (1990), “Efficiency in US manufacturing industries“, MIT Press, Cambridge, MA.

C AVES , R.E. ET AL . (1992), Industrial efficiency in six nations, Cambridge, MIT Press.

C LARK , S. (2002), “Corporate tax incentives for foreign direct investment”, paper presented at the OECD global forum on international investment, Shanghai, 5-6 December 2002.

C ULLEN , J.B AND R.H. G ORDON (2002), “Taxes and entrepreneurial activity: theory and evidence for the US”, NBER Working paper, No 9015.

D ASGUPTA , P.. AND J. S TIGLITZ (1980), “Industrial structure and the nature of innovative activity”, Economic Journal, 90, pp. 266-293.

D EMSETZ , H. (1974), “Two systems of belief about monopoly”, in H.J. Goldschmid, H.M. Mann and J.F. Weston (Eds), Industrial concentration: the new learning, Little, Brown, MA.

D IERX , A., I LZKOVITZ , F. AND S. K HALID (2004), “European integration and the functioning of product markets: selected issues” in European integration and the functioning of product markets, Edward Elgar (forthcoming).

D IERX , A., K. P ICHELMANN AND W. R ÖGER (2004), “Product market reforms and macroeconomic performance in the European Union”, in European integration and the functioning of product markets, Dierx, Ilzkovitz and Sekkat eds., Edward Elgar (forthcoming)

D ISNEY , R.F., J.H ASKEL , AND Y. H EDEN (2000), ”Restructuring and Productivity Growth In UK Manufacturing”, CEPR Discussion Papers No 2463.

D JANKOV , S., L A P ORTA , R., L OPEZ - DE -S ILANES , F. AND A. S HLEIFER (2002),”The Regulation of Entry”, Quarterly Journal of Economics, Vol. 117, pp. 1-37.

D JANKOV , S. L A P ORTA , R.; L OPEZ - DE -S ILANES , F. AND A. S HLEIFER (2003a), “Courts”, Quarterly Journal of Economics, May 2003.

D JANKOV , S., H ART , O., N ENOVA , T., AND A. S HLEIFER (2003 B ),”Efficiency in Bankruptcy”, Harvard University, Department of Economics, Working paper.

D ORGAN , S. AND J. D OWDY (2002),“How good management raises productivity”, The Mc Kinsey Quarterly, 2002, No 4.

E UROPEAN C OMMISSION (1996),“Economic evaluation of the Internal Market”, European Economy, reports and studies, No 4.

E UROPEAN C OMMISSION (2001a), “Price levels and price dispersion in the EU”, European Economy, Supplement A, No 7, July 2001.

E UROPEAN C OMMISSION (2001b), “Company taxation in the internal market“, COM(2001)582final i.

E UROPEAN C OMMISSION (2002a), “Corporation tax and innovation”, Innovation papers, No 19.

E UROPEAN C OMMISSION (2002b), The Internal Market – ten years without frontiers.

E UROPEAN C OMMISSION (2004a), Horizontal evaluation of the performance of network industries providing services of general economic interest: 2004 report, Commission Staff Working Paper, SEC(2004)866, June 2004.

E UROPEAN C OMMISSION (2004b), European Tax Survey, SEC(2004)1128, September 2004.

F RASER I NSTITUTE (2004), “Economic Freedom of the World“, 2004 Annual Report.

G ENTRY , W.M. AND R.G. H UBBARD (2004), “Success taxes, entrepreneurial entry, and innovation”, NBER Working paper, No 10551, June 2004.

G EROSKI , P., P. G REGG AND J. V AN R EENEN (1996), “Market imperfections and employment”, OECD Economic Studies, No 26, I, pp. 117-156.

G RIFFITH , R., S. R EDDING AND J. V AN R EENEN (2000), “Mapping the two faces of R&D: productivity growth in a panel of OECD studies”, IFS working paper, No 02/00.

G RIFFITH , R. (2001), “Product market competition, efficiency and agency costs: an empirical analysis”, Institute for Fiscal Studies, Working Paper 01/12, June.

G RIFFITH , R. AND R. H ARRISON (2004), “The link between product market reform and macro-economic performance”, Economic Paper, No 209, European Commission.

G RIFFITH , R. AND H. S IMPSON (2003), “Characteristics of Foreign-Owned Firms in British Manufacturing”, NBER Working paper, No 9573, Cambridge, MA.

G RILICHES , Z. AND H. R EGEV (1995), Journal of Econometrics, vol 65, No 1, pp.175-203.

G REEN , A. AND D.G. M AYES (1991), “Technical efficiency in manufacturing industries”, Economic Journal, 101, pp. 523-538.

H ARBERGER , A.C. (1954), “Monopoly and resource allocation”, American Economic Review, Vol. 44, No 2, pp. 77- 87.

H ART , O. (1983), “The market as an incentive mechanism”, Bell Journal of Economics, 14, pp. 336-382.

H ART , O. (1999), “Different approaches to bankruptcy”, Paper presented at the Annual World Bank conference on development economics, Paris.

I NTERNATIONAL M ONETARY F UND (2003), “Unemployment and labour market institutions: why reforms pay off”, IMF World Economic Outlook, Chapter 4.

J ACQUEMIN , A. AND A. S APIR (1991), “Competition and imports in the European market”, in L.A. Winters and A.J. Venables (Eds), European integration: trade and industry, Cambridge University Press, Cambridge.

K EUSCHNIGG , C. AND S.B. N IELSEN (2000), “Tax policy, venture capital and entrepreneurship”, CEPR Discussion Papers No 2626.

K ONINGS , J., P. V AN C AYSEELE AND F. W ARZYNSKI (2001), “The dynamics of industrial markups in two small open economies: does national competition policy matter?”, International Journal of Industrial Organization, 19, pp. 841- 859.

L EIBENSTEIN , H. (1966), “Allocative efficiency versus X-efficiency”, American Economic Review, 56, pp. 392-415.

L EWIS , W. ET AL . (2002), “What is right with the US economy”, The Mc Kinsey Quarterly, 2002, No 1.

L EWIS , W. (2004), “The power of productivity”, The Mc Kinsey Quarterly, 2004, No 2.

M ACHIN , S. AND J. V AN R EENEN (1993), “Profit Margins and the Business Cycle: Evidence from UK Manufacturing Firms”, Journal of Industrial Economics, Vol. 41, No 1, March 1993.

M C K INSEY G LOBAL I NSTITUTE (2001), US productivity growth: understanding the contribution of information technology relative to other factors, Washington DC, October 2001.

M ELITZ , M. (2003), “The impact of trade on intra-industry reallocations and aggregate industry productivity”, Harvard, Mimeo, forthcoming in Econometrica.

M EYER , M.A. AND J. V ICKERS (1997), “Performance comparisons and dynamic incentives”, Journal of Political Economy, 105(3), pp. 547-581.

N ICOLETTI , G. AND S. S CARPETTA (2003), “Regulation, productivity and growth: OECD evidence”, Economic Policy, Vol. 18, No 1, April 2003, pp. 9-72(64).

N ICKELL , S.J., S. W ADHWANI AND M. W ALL (1992), “Productivity growth in UK companies, 1975-86”, European Economic Review, 36, pp. 1055-1091.

N ICKELL , S.J. (1996), “Competition and corporate performance”, Journal of Political Economy, Vol. 104, No 4, pp.724-746.

N ICKELL , S.J., D. N ICOLITSAS AND N. D RYDEN (1997), “What makes firms perform well”, European Economic Review, 41, pp. 783-796.

O LIVEIRA M.J., S. S CARPETTA AND D. P ILAT (1996) ”Mark-up ratios in manufacturing industries - Estimates for 14 OECD countries”, OECD Economics Department Working Papers, No 162, Paris.

O RGANISATION FOR E CONOMIC C OOPERATION AND D EVELOPMENT (2001), Measuring productivity, OECD Manual, Paris.

O RGANISATION FOR E CONOMIC C OOPERATION AND D EVELOPMENT (2002), “Productivity and innovation: the impact of product and labour market policies”, OECD Economic Outlook No 71, Chapter 7.

O RGANISATION FOR E CONOMIC C OOPERATION AND D EVELOPMENT (2004a), Understanding economic growth.

O RGANISATION FOR E CONOMIC C OOPERATION AND D EVELOPMENT (2004b), OECD Economic Surveys: United States, Paris.

P ILAT , D. (1996), “Competition, Productivity and Efficiency”, OECD Economic Studies, No 27, Paris.

R AO , S.; A HMAD , A.; H ORSMAN , W. AND P. K APSTEIN -R USSELL (2001), “The importance of innovation for productivity”, International Productivity Monitor, No 2, Spring 2001.

R AO , S.; T. J IANMIN , AND W. W ANG (2002), “The importance of skills for innovation and productivity”, International Productivity Monitor, No 4, Spring 2002.

R OMER , P.M. (1990). “Endogenous technological change”, Journal of Political Economy, Vol. 98, pp. 71-102.

S AUNER -L EROY , J.-B. (2003), “The impact of the implementation of the Single Market Programme on productive efficiency and on mark-ups in the European Union manufacturing industry”, European Commission Directorate General for Economic and Financial Affairs, Economic Papers, No 192.

S CHERER , F. AND R OSS , D. (1990), Industrial Market Structure and Economic Performance, 3rd ed., Houghton Mifflin, Boston.

S HARPE , A. (1995), “Productivity concepts, trends and prospects: an overview”, Review of economic performance and social progress, Vol.2.

S CHMALENSEE , R. (1989), “Studies of structure and performance”, in R. Schmalensee and R. Willig (Eds), Handbooks of industrial organization, Vol. 2, Elsevier Science Publishers, Amsterdam, North-Holland.

S UTTON , J. (2002), “Market structure: the bounds approach”, mimeo, http://personal.lse.ac.uk/sutton/market_structure_bounds_sutton.pdf .

US B UREAU OF T RANSPORTATION S TATISTICS (2004), Commodity Flow Survey, http://www.bts.gov/ntda/cfs/prod.html .

US DEPARTMENT OF COMMERCE AND US C ENSUS B UREAU (2004), Exports from manufacturing establishments: 2001, July 2004.

US E NERGY I NFORMATION A DMINISTRATION (2004), http://www.eia.doe.org .

V ICKERS , J. (1995), “Concepts of Competition”, Oxford Economic Papers, pp. 47.

W ANG , J.-C. AND K.-H. T SAI (2003), “Productivity growth and R&D expenditure in Taiwan’s manufacturing firms”, NBER Working paper, No 9724, Cambridge, MA.

W INSTON , C. (1993), “Economic deregulation: days of reckoning for microeconomists”, Journal of Economic Literature, Vol. 31, September, pp. 1263-1289.

W ORLD B ANK (2004), Doing Business Indicators, http://rru.worldbank.org/doingbusiness/ .

ANNEX: A COMPARISON OF THE DEGREE OF LIBERALISATION OF NETWORK INDUSTRIES IN THE EU AND THE USA

The liberalisation of telecommunications industry in the US really started with the break-up of AT&T in 1984. The 1996 telecommunications act removed all barriers to competition across the various telecommunications segments and set up the Federal Communications Commission (FCC) as regulator with the powers to deregulate further if regulation is deemed unnecessary for competition and consumer protection. At the end of 2003, 75 per cent of the United States’ zip codes, covering 96 per cent of the US population, had the choice of supplier. In contrast, Europe liberalised its telecommunications industry in 1998 and, in 2000, an EU regulation ordered the unbundling of the local loop. Alternative providers are available in all old Member States, although in the great majority of EU countries there are no more than three to four large competing players for public voice telephony. In addition, the

development of local loop unbundling is still rather unbalanced across countries

st

Since July 1 2004, freedom of choice of energy supplier is available to all professional users in the European Union with all consumers to follow in 2007. Currently, seven Member States have already fully liberalised their electricity

and six have done so in gas. In the USA, 350 the 1978 Public Utility Regulatory Policies Act opened the way to

deregulation of electricity and opened wholesale trade to competition. As of February 2003, twenty-three US States

and the District of Columbia have passed legislation to open up their retail electricity market to competition.

In gas, as of January 2004, five States and the district of Columbia allow all residential costumers to choose their supplier. Eight other States have begun state-wide unbundling programmes and another eight have partial or pilot programmes.Therefore, less than half of the US states have opened up household consumption to competition in

energy sectors, even if the most populated States are usually liberalised.

In Air transport, the 1978 Airline Deregulation Act liberalised the sector in the US. The air transport sector in the European Union was liberalised in three successive stages. In 1987, a first package of measures started to relax the established rules. For example, it limited the right of governments to object to the introduction of new fares. In June 1990 a second "package" of measures allowed greater flexibility over the setting of fares and capacity-sharing, extended the right of an airline of one country to carry traffic to and from third countries en route and opened up the the right to carry traffic to and from the home state to all Community carriers. These measures, which were initially limited to passengers, were extended to freight in 1990. The third package adopted in 1992 gradually introduced

351

freedom to provide services within the European Union and led in April 1997 to the freedom to provide cabotage. Since April 1997, unconditional access to all domestic markets has been granted to all airlines in the European Union.

The 1970 Postal Reorganization Act created the United States Postal Service. The current law is unclear but in practice USPS has a monopoly on all mail that is not priority mail, expedited mail, mailgrams, international mail or parcel post. In contrast, the European Union has opened in 2003 the postal markets for mail weighing more than 100 grams or costing more than three times the price of a standard letter and all cross-border mail. Beginning in 2006, the market will be further liberalized to allow for competition for all mail weighing more than 50 grams or costing more than two and a half times the cost of a standard letter. After that, the European Parliament will initiate a review of the

feasibility of opening the entire postal market to competition by 2009.

350 US Energy Information Agency.

351 I.e. the right for an airline of one Member State to operate a route within another Member State.

• 6. 

  PROTECTING THE ENVIRONMENT AND

  ECONOMIC GROWTH: TRADE-OFF OR

GROWTH-ENHANCING STRUCTURAL

ADJUSTMENT?

Summary

While environmental sustainability is an integral part of the Lisbon strategy, protection of the environment and economic growth are often seen as competing aims. Proponents of tighter environmental regulation challenge this view. They highlight the financial benefits of increased eco-efficiency and the emergence of a European eco industry with million of jobs together with the need to improve how we protect public health and manage natural resources. European industry and business, meanwhile, often claim that tightened European environmental regulation is hampering their growth, undermining their international competitiveness, and destroying jobs, and will force them to eventually relocate their activities to emerging market economies outside the EU.

This chapter tries to shed some light on this controversy by identifying and analysing mechanisms and driving forces that could work in one direction or the other, by looking for empirical evidence for or against the above claims, and by coming up with some recommendations for better policy making.

The controversy surrounding environmental policy has, perhaps surprisingly, arisen not so much from the issue of conserving non-renewable commodities such as fossil fuels or industrial metals, but from the increasing scarcity or overuse of renewable natural resources, causing problems such as water and air pollution, or damage to global commons such as the atmosphere or the ozone layer. This apparent paradox reflects the fact that, while functioning markets exist for the non-renewable commodities, there are typically no markets for environmental commons. This has not posed a problem in the past, since there was an abundance of natural resources. However, due to rising demand linked to growing populations, industrialisation based on the burning of fossil fuels and the associated pollution, and new insights into the cause-effect relationship between pollution and public health, it has become necessary to find ways of managing these “goods” efficiently.

Normally, rising scarcity tends to move goods up a “property-rights hierarchy”, that is, free goods are first made subject to a common-property regime, and then, eventually, turned into private goods. Environmental policy aims at putting environmental resources such as land, water, air, the atmosphere and specific habitats under a commonproperty regime, with clear and enforceable rules. The tools at the environmental policymaker’s disposal are various forms of restriction on activity: access to these resources may be limited (for example, by placing limit values on emissions), or their use may be limited (by restricting the kind of activities allowed in natural habitats or drinking-water reservoirs) or made subject to specific conditions (such as paying a tax or an environmental levy or the obligation to clean or recycle them after use).

The theory of the property-rights hierarchy has been borne out in practice. Rising incomes and rising pollution have brought with them a rising demand for environmental protection (policies). Market forces themselves have led to a reduction in the pollution intensity of economic activity in Europe, both because of the dynamic growth of the ”cleaner” services sector, and because the private rates of return for local and regional pollution are closer to social rates than for global commons. However, strong policy action has nevertheless been needed to decouple economic activity and emission levels. These policies have been most successful in the context of ambient air pollution and acidification, while progress still needs to be made on cutting back greenhouse gas emissions.

There is no evidence to support the assertion that this decoupling has been achieved by exporting pollution through large scale delocalisation, as this process tends to be determined by factors other than environmental legislation. Moreover, the environmental ambitions of emerging market economies such as China are also rising, and standards seem to be converging globally, suggesting that “pollution havens” are at most a temporary phenomenon.

While demand for environmental protection is growing, it comes at a cost. The costs and benefits of taking action or not must therefore be estimated when environmental legislation is being drafted. However, it is rare for the costs and benefits – particularly the benefits – that actually materialise to be assessed after the policy has been implemented. Where they are, it appears that costs tend to be overestimated, possibly owing to both asymmetric information and a tendency to underestimate innovation and progress in abatement technologies. That said, spending on environmental protection – estimated by Eurostat at about 1.5 per cent of GDP in the late 1990s – does divert the resources of regulated industries from their core business. Typically, it makes their production more capital intensive and more expensive, with a negative knock-on effect on the productivity of other production factors, and on demand. If competitors do not have to comply with similar policy constraints, this spending also worsens the (international) competitiveness of the industries affected.

On the plus side, gradual but credible long-term tightening of environmental standards and ambitions helps to establish new markets for environmental technologies - both abatement and clean technologies. It is estimated that spending on environmental protection accounts for 2 million jobs in the EU15, or about 1.2 per cent of total employment.

In addition, environmental policies cause an adjustment of economic structures, mainly by changing the propertyrights regimes for natural resources. The price (in the widest sense of the word) of using environmental resources and of exposing the public to health risks should thus be brought closer in line with the social cost, with the consequence that pollution and risks to public health should decline, and GDP become less pollution intensive. Polluting industries will thus be held in check while cleaner industries will be boosted, and the net effects on welfare – though not necessarily on economic activity as measured in national accounts statistics – should be largely positive.

However, this adjustment comes at the price of friction between regulated industries, their suppliers and their customers, which could offset potential welfare gains. A cost-effective environmental policy should aim to minimise the costs incurred in achieving an environmental objective by taking into account this kind of friction, the dynamic character of adjustment needs, and the huge uncertainties surrounding cost and benefit estimates in the absence of well-functioning markets. In this way it could contribute to significantly relaxing the potential trade-off between environmental protection and economic growth, and support welfare-enhancing structural adjustment.

TABLE OF CONTENTS

• 1. 

  I NTRODUCTION .............................................................................................................................................226

• 2. 

  (W HY ) DO WE NEED ENVIRONMENTAL POLICY ? ...........................................................................................227

2.1 Renewable and non-renewable resources ............................................................................................................ 227

2.2 Inappropriately defined property rights ............................................................................................................... 229

• 3. 

  H OW POLLUTION AND ENVIRONMENTAL POLICY AFFECT THE ECONOMY .....................................................230

3.1 The mechanisms .................................................................................................................................................. 230

3.2 The valuation problem......................................................................................................................................... 231

• 4. 

  G ROWTH AND THE ENVIRONMENT – THE K UZNETS CURVE ...........................................................................232

4.1 Some evidence..................................................................................................................................................... 233

4.2 The driving forces behind decoupling ................................................................................................................. 234

• 5. 

  E FFECTS OF ENVIRONMENTAL POLICY ON E UROPEAN BUSINESS ..................................................................236

5.1 Costs of environmental policies........................................................................................................................... 237

5.2 Benefits of environmental policy to business ...................................................................................................... 241

5.3 Overall impact ..................................................................................................................................................... 242

• 6. 

  I MPLICATIONS FOR REGULATION – FINDING THE RIGHT BALANCE ................................................................244

• 7. 

  C ONCLUSIONS ..............................................................................................................................................246

R EFERENCES ...........................................................................................................................................................248

PROTECTING THE ENVIRONMENT AND

ECONOMIC GROWTH: TRADE-OFF OR

GROWTH-ENHANCING STRUCTURAL

ADJUSTMENT?

• 1. 

  Introduction damage may hamper economic activity. This lack of

data, often due to the absence of market transactions in

There is probably a fairly broad consensus that, in the these fields, is a severe barrier to integrating long-term, high material living standards and high levels environmental and economic policies. In particular, the of environmental quality and public health are mutually absence of figures on the effect of environmental consistent, if not interdependent goals. However, at least damage on economic activity makes it difficult, if not in the short- to medium-term, environmental policy and impossible, to identify the scope for “win-win” economic growth are often portrayed as being in conflict measures. with one another. That is, an increase in economic The structure of the chapter is as follows. As a activity is seen as being inevitably bad for the preliminary to the main theme of the chapter, the environment, while environmental policy is regarded as question of why – or whether – we need environmental imposing a drag on growth. policy is discussed. From this basis, the scope for both This chapter sets out to examine the validity of this synergies and trade-offs between environmental quality perception: is it true that environmental quality and and economic growth is considered. The next part of the economic growth are competing goals, or can chapter in a sense reverses the direction of causality by environmental policy lead to more efficient use of scarce looking at the relationship between economic activity resources, so fostering growth-enhancing structural and changes in pollution, drawing on the “environmental adjustment of the economy? The focus of the following Kuznets curve” literature. The final part of the chapter, pages is not whether environmental policy is successful in line with the overall theme of this year’s review, in delivering its objectives of improvements in the looks at the possible contribution of environmental environment and public health – in a sense, these are policy to improving the short- and medium-term taken as given – but on the rather narrower issue of the framework conditions for growth. It examines how costs and benefits to the economy of environmental environmental policy causes costs and benefits for policies. business, and suggests how policy should be designed to

minimise the former and maximise the latter, without

The chapter draws on theory and empirical evidence, compromising the environmental objectives of the where the latter is available. However, one of the policy.

conclusions is that there is an acute lack of data, both on the impacts of environmental policy on economic growth, and on the degree to which environmental

 Box 1: “Growth” and “welfare”

Throughout this chapter the terms “growth” or “economic growth” are used in the sense of “changes in real Gross Domestic Product (GDP)”. Although standard economic theory deals more with “welfare”, and changes in real GDP do not necessarily correlate perfectly with changes in national well-being, or welfare, a focus on the narrower concept of economic growth has been taken for two reasons.

A first, pragmatic, reason is that no comprehensive measure of welfare exists. Attempts to measure and compare the relative contributions of environmental quality and production of marketed goods and services quickly run into problems of “incommensurability”. That is, different units are used to measure changes in environmental quality and changes in market output of goods and services. The fundamental underlying difficulty is that aggregates such as GDP derived from the national accounts are mainly based on transactions that take place in the market. The perceived need for an alternative measure, such as a “green” GDP, arises precisely because markets for environmental resources do not generally exist.

Although considerable work has been undertaken to link uses of environmental resources with national accounts (see, for example, Schoer et al. (2001) or Eurostat (2001a), this does not yield a single, integrated measure of “welfare”. Indeed, as noted in the joint UN/EC/IMF/OECD/WB manual of integrated environmental and economic accounting, these integrated approaches are themselves open to criticism on the grounds that they fail to take adequate account of other dimensions of welfare, in particular its social dimension.

A second reason for using the conventional, albeit flawed, concept of growth in GDP is that trying to replace it with an overall measure of welfare would have fudged the issues the chapter tries to address. The aim here is not to assess whether environmental policy contributes to overall welfare – it is taken for granted that this is so – but whether and to what extent the pursuit of enhancements in environmental quality have been bought at the expense of improvements in GDP. This is a crucial question, given the Lisbon strategy’s aims of seeking simultaneous improvements in economic, environmental and social well-being.

• 2. 

  (Why) do we need environmental policy? Brundtland et al. in Our common future (1987), while not making dramatic predictions of this sort, highlighted

the implications for world energy consumption of the

Views about the interaction between environmental combination of a rising world population with the need policy and economic growth frequently fall into two to achieve much higher living standards of the camps. populations of poorer countries.

2.1 Renewable and non-renewable resources The recent rise in oil prices has revived fears of looming shortages, 352 even if it is generally accepted that part of

On one side, there are those who point to the finite the price rise reflected a perceived increase in the risk of nature of many of the earth’s natural resources on which supply disruptions due to heightened political tension in much economic activity depends, the seemingly the Middle East and other parts of the world. A period of inexorable rise in human consumption of those sustained, rapid commodity price increases would tend resources, and consequent inevitable shortages. Everto strengthen the arguments of those who argue that our increasing rates of exploitation of natural resources societies are developing along fundamentally could lead to the depletion of non-renewable resources unsustainable paths. such as oil or industrial metals, to high levels of

biodiversity loss and a subsequent reduction in the Others take a more optimistic view. While quality of life, as this also depends on the natural acknowledging that natural resources such as fossil fuels environment and species diversity (Balmford et al. and minerals are indeed finite, they foresee considerable (2002)). The unsustainable “footprint” of economic potential for society to adapt to possible future shortages activity would first lead to sharply rising input prices, through innovation and technical progress. This view and ultimately to the depletion of crucial inputs, pushing rests in part on historic evidence of huge improvements substitution costs to unaffordably high levels. This could in the efficiency of resource use: for example, the have significant impacts on growth, both in developed efficiency by which the energy in coal is converted to but even more so in developing countries. Even wars for steam has increased over time by a factor of 25.

353

access to limited resources (water, oil?) could be expected.

This type of “doomsday” standpoint achieved particular prominence with the publication by the Club of Rome of

The Limits to Growth (Meadows et al. 1972). They predicted that if the then current trends in population, industrialisation, pollution, food production and resource depletion were to continue unchanged, then within the

following one hundred years, “the most probable result 352 will be a rather sudden and uncontrollable decline in See for example “The end of cheap oil”, National

both population and industrial capacity.” Geographic magazine, June 2004. 353

  Shell International (2001).

Graph 1: Commodity prices, 1980-2004 Graph 2: Long-term trends in CO 2 - concentrations and global temperatures

140 Commodity Non-Fuel Price Index, 1980 = 100 40 800

120 Crude Oil (petroleum), Price index, 1980 = 100 CO 2 (rhs)

30

600 v)

100

at iv e °) C Current (p pm

20

80 levels

n

re r

el

nt (

in

400 ra

tio

ra tu nt ce

60 pe re

se 10

T em to

 p

2 C

on

40 200 0 C

O

20 T° (lhs)

-10 0

0 -400 -300 -200 -100 0

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 Thousands years from now

Source: IMF. Source: IPCC (2001), Jouzel et al. (1987, 1993, 1996)

From this perspective, increases in the prices of what we To date, the targets of environmental policy-makers, today regard as essential raw materials will act as a perhaps to the surprise of some, tend to support the stimulus to resource-saving innovation. Moreover, seen optimists: preserving non-renewable resources has not

in a longer-term perspective, as in Graph 1, the case can been the main driver of environmental policy. 354 In fact,

be made that recent rises in commodity prices have done contrary to what one might expect, the most pressing little more than return them to their levels of quarter of a environmental issues are human health and century ago: neither the level of prices, the scale nor the environmental problems caused by overuse (in terms of speed of the recent increases look particularly overstretching the carrying and recovery capacity) of exceptional. renewable resources: air and water pollution, climate

change and biodiversity loss (see Box 2 “The priorities Optimists also assume that what holds for commodities of environmental policy”). As argued below, this might also hold for other kinds of environmental apparent paradox of relative shortage in renewable pressures. Lomborg (2001) may be regarded as a recent resources and relative abundance of non-renewable example of this outlook, according to which far from resources can be explained in terms of the presence or leading inevitably to environmental (and ultimately, absence of enforceable property rights. economic) disaster, economic growth has generally been

associated with declining, not increasing levels of The problem of climate change is a particularly forceful

environmental damage. example of the contrast between relative abundance of non-renewable resources and relative shortage of

These optimistic interpretations tend to leave renewable ones. unanswered the question of the extent to which current

prices reflect both the needs of the present and those of According to the Intergovernmental Panel on Climate future generations. They also overlook the question of Change, increased atmospheric concentrations of those resources for which no markets exist. Dasgupta greenhouse gases – mainly due to emissions of carbon and Heal (1979) show that, in general, markets will dioxide from the burning of fossil fuels – are likely to be allocate non-renewable resources efficiently over time warming the earth’s atmosphere, thus affecting the only under quite restrictive conditions. climate.

355 The likely impacts include more extreme

weather conditions, with an increased risk of heat waves, droughts and floods and their associated damages. In the longer term, global warming could

354 As an example, the European Commission is currently

developing a “thematic strategy” on natural resources which is expected to focus on the environmental impacts of using non-renewable resources like metals, minerals rather than on their possible scarcity. See European Commission (2003b).

355 For this paragraph, see IPCC(2001a, b, c).

Box 2: The priorities of environmental policy

Traditionally, environmental policies have dealt with three core issues: (i) threats to public and occupational health, where the environment (mainly water and air) is the medium transporting the cause of disease or health risks, so that tighter air quality or water quality standards could help to significantly reduce these risks, (ii) biodiversity issues, such as natural equilibria, food chains, or existence values of rare species or the gene pool, especially in largely unexplored biotopes, such as deep seas or tropical rain forests, and (iii) the overuse of natural resources, such as commodities, fish stocks, global commons (tropical rain forests or the atmosphere and the ozone layer).

Historically, the first of these – concern over the public health impacts of pollution – has been the main driver of environmental policy. The existence of a relationship between polluted air and water and adverse health impacts has been recognised for a long time,* even if the precise nature of the cause-effect links and their scale remains uncertain in some respects. First policy reactions (at local level) to this insight typically took the form of action to establish waste-water collection systems and protect drinkingwater reservoirs; later, policies to improve air quality and reduce exposure to potentially harmful substances complemented efforts to protect citizens and workers against the negative fall-out from human activities.

A more modern, but still long-standing additional rationale for environmental policy, such as the policy combating acid rain, has been to reduce the impact of pollution – particularly air pollution – on buildings and crops. More recently, as the scale and scope of human activity has continued to expand, issues relating to preserving the global commons – climate change, the ozone layer, biodiversity, for example – have become prominent. Here too, part of the rationale for policy action is motivated by fears of the negative feedback from human activity to public health and economic activity: a significant acceleration in the rate of climate change could have adverse impacts on human health by expanding the range of infectious diseases such as malaria, for example. However, most concerns with respect to climate change are related to its potentially dramatic effects on economic activities.

Notwithstanding the broadening of the range of issues tackled by environmental policy, protecting human health remains a key factor, not least because improvements in knowledge highlight previously unknown sources of harm. For example, most of the outstanding health problems due to air pollution are now believed to be caused by very fine particulate matter, emissions of which are not directly regulated at EU level.

• * 

  Lomborg (2001) reports that a first attempt to ban coal burning in the United Kingdom was made in the 14 th century!

cause – besides a general rise in sea levels - severe biodiversity, so individual decisions on land use, for shocks such as shutting down or substantially weakening example, are unlikely to take account of the wider social the Gulf Stream. This would give much of Europe a less and economic benefits that may flow from a higher level temperate climate, with significant impacts for economic of species diversity. The lack of markets for activity. Yet cumulative emissions of carbon dioxide environmental resources thus gives rise to a difference

th

from the middle of the 19 century to date – that are between the private benefits of their use and the benefits already judged to be causing climate change – result to society at large. Action to reduce these gaps, or from the burning of no more than 6 per cent of the “externalities”, between private and societal benefits, (so world’s estimated total fossil fuel resources. Thus, the called because the effects of individual action on the problem for environmental policy is not that we are wider society are not “internalised” in prices) will running out of a non-renewable resource – fossil fuel – therefore potentially be beneficial for the overall wellbut that we are overstretching the capacity of the earth’s being of society.

atmosphere, a renewable resource. As long as environmental resources were abundantly

2.2 Inappropriately defined property rights available, the lack of enforceable property rights was not really an issue and could be largely neglected. However,

The superiority of a market economy over other forms rising demand for natural resources due to growing of economic organisation – in terms of the ability to populations, industrialisation based on the burning of deliver high and rising levels of material comfort to fossil fuels and the associated pollution, new insights people – is based in part on well-defined, enforceable into cause-effect relationships (that is, the link between and tradable property rights, which in turn requires the pollution and public health threats), better knowledge existence of effective public institutions. Enforceable about how ecosystems function, their potential fragility property rights enable owners of resources to use them and the services they provide and increasing awareness to produce goods and services for sale to willing buyers; of the limits of current knowledge have led to the need when property rights are tradable, they may be bought to change how these “goods” should be managed.

and sold for the benefit of both buyer and seller. Normally, rising scarcity tends to move goods up a

However, there are frequently no property rights for “property-rights hierarchy”. That is, free goods are first environmental resources such as air and water. When turned into goods falling under a common-property this is so, they can be used for free and in unlimited regime, before they eventually turn into private goods. quantities as a dump for waste by-products of human However, for this to happen property rights must first be and economic activity. Similarly, there are typically no defined and assigned, and then they must become property rights – and hence no markets – for maintaining enforceable, normally with the help of both the

institutional and legal framework, and technical value added enters national accounts statistics. Indeed, exclusion mechanisms. services such as waste-water management or municipal

In the case of some natural resources the problem may waste collection and treatment have turned into be that while property rights exist, they are not important service providers with an annual turnover (in adequately defined and/or enforceable or enforced, 1999) of € 48 billion each.

356

leading to an overuse of these resources. The Environmental policy usually aims to prevent, reduce or management of fish populations in (inter)national waters at least manage better such waste streams. Pollution may serve as an example of this. For example, the damages the natural environment, but may also affect decline in fish stocks in European waters is not because the amount and quality of the inputs available to be used it was not possible to establish ownership of fish but for production. Indeed, as already observed, one of the because the fishery quotas Member States agreed upon main drivers of environmental policy is the effect of have often been too high (if measured against scientific pollution on human health. This gives rise to economic advice) to avoid overexploitation of these resources, and costs in the form of higher health care spending and they are often monitored in an insufficient way. There is reduced labour supply. Pollution also affects natural a striking contrast between the threat to the continued resources such as soil and water, reducing their existence of some types of fish – in principle a productivity, and requiring significant resources to be renewable resource – and the continuing availability of spent on their remediation. non-renewable resources such as precious metals, for

which exclusive property rights have been established. However, reducing the emissions that cause pollution and environmental damage may imply diverting

The same contrast between the relative abundance of resources from production of goods and services fossil fuels and the relative scarcity of the atmosphere demanded by market actors (such as power steering or has already been highlighted in the context of climate air-conditioning in cars) to pollution abatement activities change. The link to the presence or absence of well(such as catalytic converters), that is, the production of

defined property rights should be immediately apparent. goods and services imposed on market actors. 357 If this is

Indeed, it is noteworthy that the first significant global the case, there may be a trade-off between providing attempt to address climate change – the Kyoto Protocol goods and services to clean up the environment and – limited developed countries’ access to the global producing economic goods and services requested by commons of the atmosphere by placing a cap on their pure market considerations. greenhouse gas emissions. It is equally noteworthy that

subsequent problems in implementing the Protocol (in Any given policy proposal is likely to give rise to both particular, the withdrawal of the United States) are these effects. That is, cutting back on emissions is likely related to both dissatisfaction with the size of the limits to require that resources are allocated to abatement, on emissions (that is, the volume of property rights thereby reducing the level of the primary economic allocated), the fact that access to the atmosphere remains output of the regulated sector, while the improvement in unrestricted for some large emitters (so that for these environmental quality that results from lower emission emitters, the atmosphere remains a global commons) levels may enhance the availability and productivity of and the absence of mechanisms to enforce the resource inputs. The issue then is which of these effects

agreement. is the larger, that is, whether the fall in output in the regulated sector (and in up- and downstream industries)

due to reducing emissions is offset by the rise in output

• 3. 

  How pollution and environmental policy in pollution abatement industries and in the rest of the economy due to lower levels of pollution.

affect the economy These competing effects on output of reducing

emissions and reducing pollution levels help to explain

3.1 The mechanisms some of the controversy about the impact of

The output of any economy depends on both the environmental policy on economic output. If those who quantity of inputs it uses and the efficiency with which have to incur the cost of reducing emissions are not the these inputs are used: typically, the greater the quantity same as those who benefit from lower levels of pollution of inputs and the more efficient the use of these inputs, (as will very often be the case), then it will not be

the greater the amount of output.

Most forms of production also generate pollution. That 356 Eurostat (2001).

is, on top of the primary output produced for the market, 357 This is not the only shift. To assess the economy-wide

they also produce waste, a public bad, in the form of air impact it is necessary to take account of the substitution

or water pollution, or other forms of liquid or solid and income effects triggered by a given policy measure. waste, which are typically released into the environment Increased energy taxation for example will induce

(air, water, soil), unless waste-management systems companies to substitute other factors of production for

have been put in place. In the latter case, these systems energy, and less energy-intensive products will constitute a

themselves contribute to economic activity, and their larger share of final goods. This will entail transfers of income within the economy beyond those set out here.

surprising if the two groups have differing views about particularly suitable in cases where people are the desirability of action to reduce emissions. Moreover, unaware of a certain dose-response relationship and as in all likelihood the members of the group of those would therefore not have well established potentially negatively affected by tightened regulation preferences. will individually lose much more than the individuals of

society at large, they will articulate their opposition − “Avoidance costs”: This frequently used technique

much more loudly and visibly than the individuals who takes the costs of measures to reduce externalities as benefit. an approximation of their benefits. The main

The time dimension may also be relevant: the sequence advantage of this approach is that avoidance costs of events that results from implementing an are comparatively easy to establish, as the costs of environmental policy measure is that first emissions are end-of-pipe technologies (like catalytic converters) reduced, so output falls, and then the positive effects of or other defensive expenditure (such as double reduced pollution levels materialise, so output rises. In glazing for sound-proofing) are usually well known. other words, benefits occur later than costs. So if The main disadvantage is the risk of circular different interest groups have differing views (explicit or reasoning when one would like to establish policy implicit) about the appropriate discount rate, this may be priorities in the first place.

enough to lead them to opposing conclusions about

whether the measure is good or bad for the economy. − “Hedonic pricing”: This method tries to estimate

Differences in the timing of costs and benefits are how the prices of otherwise similar goods are especially relevant in dealing with problems such as air affected by differences in their environmental pollution and climate change. The benefits of action characteristics. For example, differences in the taken now in these areas may only be felt many years or prices of houses in quiet and noisy streets may be even decades in the future. used to place a value on measures to reduce noise

pollution. This method can only be used to value

Further scope for debate comes from our imperfect impacts of which people are aware.

understanding of both the exact nature of the “doseresponse”

 function, that is, the relationship between − “Contingent valuation”/“stated preferences”: emissions, levels of pollution and adverse environmental Individuals are questioned about how much they and health impacts. Although it may be possible to feel their well-being is affected by a particular forecast the costs of action to reduce emissions environmental issue. The approach may be based on reasonably precisely, there may be considerable “willingness to pay”, that is, determining how much uncertainty about the scale of the benefits. This opens people would pay to avoid or reduce a particular another avenue for disagreements about the net impact externality, or on “willingness to accept”, that is, of environmental policy on the economy. the amount of compensation people would require

in return for a deterioration in the environment.

3.2 The valuation problem Which of the concepts is more appropriate is likely

As well as these issues of the distribution of costs and to depend on the (explicit or implicit) allocation of benefits between different economic agents, the timing property rights.

of these costs and benefits and their extent, a further

major source of potential uncertainty and disagreement The contingent valuation/stated preference approach arises precisely because of the lack of markets for many tends to be more costly than the others because it of the benefits of environmental policy, such as requires information from individuals, obtained through increased life expectancy, improved health in general, or interviews or questionnaires. Offsetting this maintaining biodiversity. A number of techniques have disadvantage, it gives more complete estimates of the been devised to value these benefits, to provide input to impact of environmental damage on well-being, because policy making: it is able to capture “quality of life” aspects that some of

the other methods do not. For this reason, it is often

• - 

  “Damage function/ dose-response”: Based on regarded as the preferred, or “first best” way to value scientific knowledge, a relationship is established environmental externalities for which there are no between the observed environmental pressure (for markets. 358 However, this approach has to be carefully example, particulate emissions or noise) and the applied, as answers to questionnaires may differ observed impact (for example, increased morbidity significantly from actual behaviour once it comes to or mortality). It is only with respect to the latter that implementing a willingness to pay or to accept.

  a monetary valuation is attempted. However, the

monetary valuation is limited to the costs that are Placing a value on human health or species diversity visible in the market (hospital costs, labour may be considered by some to be morally offensive, but productivity, and so on.). In practice, a damage is necessary if the costs and benefits of implementing or function approach can therefore often be expected not implementing a particular policy action are to be

to underestimate the welfare costs of a given

externality. On the other hand, it might be 358 See European Commission (1995).

analysed in a rational way. Given an estimate of the production losses due to “restricted activity days” could expected costs of a measure, a decision to proceed or not add about € 1 billion to their estimate. The willingnessto proceed with it places an implicit floor or ceiling to-pay approach gives much higher values, with total air respectively on the value attached to its benefits. The pollution-related costs in the three countries estimated at techniques outlined above for making this implicit € 50 billion, equivalent to the order of 3 per cent of GDP valuation explicit do not aim to exercise an ethical

judgement, but rather to facilitate rational policy debate. A recent report on the costs and benefits of Natura 2000 sites in Scotland throws particular light on how different

Examples methods of valuing environmental assets can yield

Pretty et al. (2000) undertook an assessment of total completely opposing cost-benefit ratios. 359 Designating

external environmental and health costs of agriculture in an area as a Natura 2000 site implies costs such as the the United Kingdom. Their approach was close to the costs of managing and maintaining the site and “damage costs” method. Wherever possible, they valued opportunity costs in terms of restrictions on the externalities based on the financial costs they imposed, economic activities that may be undertaken on the thereby aiming to overcome uncertainties in valuing protected area. Benefits from classification as a Natura non-marketed goods and services such as landscape or 2000 site include direct use values – essentially related biodiversity. This approach yielded an estimate of total to tourism – and non-use values, reflecting individual annual external costs of UK agriculture of £ 2.3 billion willingness to pay for the continued existence of natural in 1996, equivalent to 89 per cent of average net farm resources.

income for the 1990s. When both use and non-use benefits were taken into

Pretty et al. claim these estimates are likely to be consideration, the report estimated that the ratio of conservative. For example, agriculture’s negative impact benefits to costs of designating areas as Natura 2000 on biodiversity is estimated based on the cost of plans to sites in Scotland was about 7 to 1, so that the policy return species and habitats to acceptable levels for represents good value from the perspective of society at society (after taking account of impacts of other sectors large. However, almost all of the benefits relate to nonon biodiversity), but this does not adequately include use values, so that from the narrower perspective of the non-use values of biodiversity; external costs due to impact on economic activity, the policy has negative chronic health effects of pesticide use are excluded due impacts. If these non-use values are excluded, the ratio to uncertainty in current scientific knowledge. On the of benefits to costs is considerably less than 1. other hand, their estimates do not take account of

possible positive externalities of agriculture, such as In circumstances such as these, the higher the value a

landscape and amenity values or carbon sequestration. society attaches to intangible or non-traded benefits, the more willing it will be to trade economic growth for

An example of the costs to the economy of air pollution environmental quality. As individuals and groups in is given by Sommer et al. (1999), who report the results society will have different views about the importance of an assessment of the health and related economic of issues such as nature conservation, whether because impacts of air pollution in Austria, France and of incomplete information or because they are Switzerland. They find that some 40,000 deaths per differently affected (that is, potential losers or winners), year, or 6 per cent of all deaths in these countries, are or for other reasons, this offers another reason for attributable to air pollution. In addition, air pollution disagreements about the right level of ambition of caused large numbers of additional cases of chronic environmental policy. Differences of opinion about the bronchitis and asthma attacks, giving rise to over 28 desirability of environmental policy may arise as much million “restricted activity days” per year among the from differences in value systems as from disagreements adult population (aged 20+) in the three countries. Road about its physical effects. traffic was identified as the major source of air pollution causing these impacts.

The authors tried to give an economic value to these 4. Growth and the environment – the

impacts in two ways, by estimating the value of the lost Kuznets curve

production or income due to premature death or ill health, and by estimating “willingness-to-pay” to reduce It is a widely observed phenomenon that as economies the risk of death or illness due to air pollution. As grow over time, emissions of many pollutants first grow, already noted, the latter is generally considered to be the and then decline. This stylised fact is illustrated in appropriate way to measure the cost to society of death Graph 3. First to be addressed are local pollution and illness, because in addition to the cost of lost problems, such as lack of access to safe drinking water. production or income, this method includes intangible Next to be tackled as incomes rise are regional factors such as pain and suffering. problems, such as pollution due to sulphur dioxide (acid

The first approach gave an estimate of € 6.5 billion (in rain, for example). The last to be dealt with

1996 prices). This excludes the cost of “restricted

activity days” because of a lack of precision in how this 359 See Jacobs (2004). Natura 2000 is a European Union-wide impact was defined. The authors indicate that including network of nature conservation sites.

(successfully?) are global pollutants, of which international dimension of environmental pollution to greenhouse gases are a notable example. the general public for the first time.

Questions to be answered in this context are, (i) how far As regards regional and global pollution, Graph 4 allows these stylised trends are matched by empirical evidence the broad validity of this sequence to be assessed for the (for the EU), (ii) what drives this differentiated EU15, for sulphur oxides, nitrogen oxides and carbon decoupling of economic growth and pollution, and (iii) dioxide emissions from energy. The graph shows three whether a price has been paid for this decoupling in the distinct patterns: sulphur oxide emissions have fallen form of foregone economic growth and delocalisation of throughout the period, so that they are now less than industries? This chapter and the next try to at least one-fifth of their levels in the early 1980s; emissions of partially answer these questions. nitrogen oxides did not start to fall until around 1990,

since when they too have shown a steady decline; finally

Graph 3: Stylised relationship between economic growth emissions of carbon dioxide from energy use, a typical and different types of pollution global pollutant, show no sign as yet of turning down.

The graph lend support to the hypothesis that the priority

Local pollutant Global pollutant attached to tackling different types of pollution changes

as income rises. They show a clear absolute decoupling

Regional of local and regional levels of pollution from GDP

n pollutant levels. However, decoupling for the global pollutant,

tio

llu carbon dioxide, has so far occurred only in relative

f po

l o terms, that is, absolute emissions are not falling

ve dramatically as for the other pollutants, but have

Le remained rather stable over the last two decades.

In the early 1980s, emissions of sulphur oxides came predominantly from stationary sources, such as fossil fuel power plants, and were a significant source of local Income pollution. Pollutants whose causes and effects are mainly local may be tackled first as almost all the

Source: based on World Bank (1992). benefits of action accrue to the members of local

communities, and as the latter are able to agree

4.1 Some evidence 360 appropriate solutions among themselves than more heterogeneous bigger communities. Compared to

Typical local pollutants are water pollution, solid waste sulphur oxides, a greater share of nitrogen oxide streams and local air pollution due to the dirty burning emissions come from transport. Pollutants which are of fossil fuels. While waste water and solid waste emitted from a larger number of sources, and whose streams have not really declined over time, their effects are widely spread, require national action: this management has significantly improved over the past requires mobilising and co-ordinating greater amounts century, and nowadays private households or enterprises of administrative resources, and takes longer to organise. not connected to solid waste and waste-water collection Finally, carbon dioxide is the major greenhouse gas and treatment networks are the exception and no longer contributing to human-induced climate change. Such the rule in the EU. Indeed, initially, such waste was only pollutants with global effects cannot be effectively collected and then disposed of in rivers. Later it was tackled in the absence of global co-operation, so their treated before being released into rivers. volume may continue to rise with rising income,

Local air quality has also improved dramatically over possibly until long after trends in local and national the last seven decades, both as a result of less dirty pollutants have turned downwards.

361

burning of fossil fuels and tendencies to export pollution outside the local jurisdictions where it is generated: wherever it was possible (at low costs) – as in the case of large combustion plants by fitting them with higher smokestacks – local air pollution was “exported”, turning it into regional or even trans-boundary pollution.

However, the price of a policy aiming at a “blue sky” over the regions with large heavy industry in western

Europe was environmental damage such as acid rain and

“dead lakes” in Scandinavia, highlighting the

360

  For a more complete discussion of trends in pollution in

the European Community, see the EU E CONOMY R EVIEW 361 See also World Bank (1992) and European Commission

2000, Chapter 4. (1994).

Graph 4: Trends in emissions of various pollutants, EU15, The role of factor endowments is, for example,

index 1990=100 highlighted by Copeland and Taylor (2004). According

to them, if output is made up of a “dirty” good X

180 (industry) and a “clean” good Y (services), it is a simple

matter to decompose the level of emissions of any

160 Sulphur oxides pollutant z in the form of an identity:

140

z = Q

120 * S * e, Carbon dioxide where Q is the level of output, S is the share of the

 1 00 100

 = “dirty” good X in total output, and e is the level of

89 80 emissions produced by one unit of X. Changes in the

19

60 Nitrogen level of pollution are then determined by changes in

oxides output, the share of the “dirty” good in output, and the

40 emissions intensity of the “dirty” good.

20

It is immediately obvious from this identity for z that a

0 “neutral” increase in output, leaving S and e unchanged, 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 will lead to a rise in pollution, and equally, that a fall in

Source: Commission services. z that leaves S and e unchanged must lead to a fall in output. Less obviously, Copeland and Taylor show that,

if growth occurs due to an increase in the supply of the

4.2 The driving forces behind decoupling factor used intensively in the production of the “clean” good, pollution levels will fall. This is a consequence of

The bell-shaped relationship between growth and the Rybczynski effect in a two-good model, whereby an pollution has been called the “environmental Kuznets increase in the supply of one factor leads to a rise in the curve”, following Kuznets’s observation (1955) that output of the good that is produced using that factor rising per-capita incomes were associated with an initial intensively, and an absolute fall in the output of the increase in inequality and a subsequent decline. The other good.

cause of this relationship between growth and the

environment is a crucial issue: what is the “transmission However, the assertion by Copeland and Taylor that the mechanism” from higher levels of output to lower levels Rybczynski effect shows that “a strong policy response

of pollution? to income gains is not necessary for pollution to fall with growth” is surely of little relevance in the real

It may be helpful to distinguish between market driven world. Altering the model slightly, so that output is and policy driven mechanisms when trying to explain made up of a “high pollution” good X and a “low the driving forces behind the relative and absolute pollution” good Y, is enough to make the impact of decoupling of pollution trends from economic growth. higher output of Y on pollution indeterminate. The first might shed some light on why there has Moreover, observed growth patterns both in the EU as occurred a relative decoupling of economic activity and well as in other industrialised and developing countries pollution, while the second might be necessary to do not generally support the contention that higher explain the evidence of absolute decoupling and reduced output in one sector is accompanied by absolute falls in environmental pressure from certain pollutants. output in others. An increasing share of “clean” services

relative to “dirty” industry in an economy in which both Market mechanisms

Market-driven changes in economic structures, sectors are growing will produce a fall in pollution per

including the pollution intensity of an economy, are unit of output, not necessarily an absolute decline in pollution levels.

determined by factor endowments, relative prices,

competition and innovation, rates of return, market In varying this theme and focussing on labour supply saturation, and so on. and relative prices, the same mechanism would have

The change in the relative importance of the three worked when, as a result of the emerging “clean” service sector with its “clean” jobs, demand for jobs in

sectors agriculture, industry and services over the last

centuries – the change itself driven by changing factor the service sector would increase, while demand for jobs endowments, technological progress, market saturation in the “dirty” industry would decline. Then industry

and changing needs of the population – is definitely the would – unless it replaced labour by capital - either have to pay a supplement or to invest in abatement

most important force behind the changing pollution technologies so as to make jobs “cleaner” and less

intensity of economies: with the emergence and rapid

growth of dirty heavy industries and industrialisation the dangerous. In both cases production costs in industry

pollution intensity typically sky-rockets, the emergence would rise relative to costs in the service sector, and its share in GDP would decline, leading to a fall in the

and rapid growth of the cleaner service sector then pollution intensity of the economy. A similar reverses this trend. mechanism would be triggered if labour demand shifted

due to new insights in dose-response functions so that workers became more aware of the risks in dirty and If correct, this explanation for the environmental dangerous industries. Kuznets curve implies a clear trade-off between growth

An alternative mechanism through which growth may and the environment, certainly in the short-term as the lead to lower emissions – without policy intervention – economy adjusts to the effects of the regulation. In the is if there are increasing returns to scale in pollution longer term, since pollution generally tends to be abatement. Andreoni and Levinson (1998) develop a associated with more capital-intensive industries, the model in which the relationship between pollution and implication could be a shift towards less capitaloutput is monotonically increasing, U-shaped, or bellintensive activities with adverse consequences for labour shaped, depending on whether abatement shows productivity. In addition, this would imply that the constant, declining, or increasing returns to scale, environmental Kuznets curve will not persist into the

respectively. long-term: as poorer countries get richer, they too will impose tighter environmental regulation, so that at some

In their model abatement is undertaken by individuals stage, outsourcing of pollution cannot continue. because pollution lowers their utility. In consequence, as

of a certain point in income and pollution, the rate of In order to check the appropriateness of this explanation return on increasing traditional output combined with an evidence must be found for both the existence of increase in pollution turns negative and makes pollution significant pollution havens and the importance of abatement rewarding. However, in such a scenario international differentials in environmental standards for individual abatement efforts take no account of location decisions of large scale investors. externalities (except to the extent that individual utility Unsurprisingly, the mechanisms described here have is enhanced by concern for the welfare of others). been quite extensively examined. The typical approach Consequently, even if there are increasing returns to is to examine the relationship between trade and scale in pollution abatement, the break-even point for investment flows and differences in environmental pollution abatement would remain higher than that regulation. In one of the most widely cited references, which would have resulted had total social costs and Jaffe et al. (1995) concluded that there was little benefits been taken on board, implying that pollution evidence to support the argument that increasing

will remain at a socially-inefficient level. environmental regulation had led to significant changes in US net exports, or to relocation of US manufacturing.

Environmental policies They also found no evidence that environmental

A plausible explanation for the relationship shown by regulation stimulated innovation and international the Kuznets curve is that at low levels of income, competitiveness. Similarly, Leonard (1988) found that increased consumption of material goods is valued more lax environmental standards had not been successful in than environmental quality, so that the utility gain from attracting foreign direct investment. consumption is greater than the loss of utility due to a

deteriorating environment; as consumption levels rise, Copeland and Taylor (2004) offer a less sanguine view. further increments produce ever smaller gains in utility, They argue that the earlier studies on which Jaffe et al. so there is a willingness to trade off a slower increase in based their conclusions, failed to take adequate account material consumption against welfare-enhancing of other differences – notably, factor endowments – improvements in environmental quality. Because of the between countries that influence trade flows (although presence of externalities, uncoordinated action by these differences were mentioned as possible individuals will have at best limited effect, so this explanations for the absence of a measured effect of willingness can only be fully realised by policy environmental policies). They quote more recent work

intervention. that explicitly accounts for these factors, showing that tighter environmental policy does have a negative

One possible policy-driven cause of the environmental influence on the production of polluting goods, but, in Kuznets curve is a “pollution haven” effect or “race to line with the earlier work, confirms that these other the bottom”, that is, a relocation of dirty industries to factors remain the main determinants of trade and third countries in response to tightened environmental investment flows. In short, according to them, there is a policies. The reasoning behind this is that as incomes pollution haven effect, but it is too small to explain the rise, demand for a cleaner environment increases, but so existence of the environmental Kuznets curve. does demand for goods and services that give rise to

pollution: wealthier people want more spacious and The implications of these results would be that if better heated houses, more energy-consuming domestic developing countries “catch up” with developed appliances, bigger and more powerful cars, and so on. A countries, so that differences in factor endowments possible explanation for the simultaneous increase in narrow, the influence of differences in environmental incomes and environmental quality is then that the policies on trade and investment will become more demand for a cleaner environment is met by regulation. important.

This raises the costs of polluting firms, who relocate Offsetting this, as developing countries catch up with abroad to remain competitive (this line of argument is developed countries, differences in environmental the environmental equivalent of “social dumping”). regulation may narrow as well, so that differences may

only be temporary and more a result of delayed industrialisation than the result of an active active environmental policies. Markets themselves will “environmental dumping” policy. Indeed, a recent study only remedy parts of environmental pressures, in line undertaken for the European Commission comparing with private instead of social rates of return. However, EU air pollution policies and legislation with other the more environmental policies succeed in internalising countries such as the USA, Japan, but also China, show environmental externalities in investment decisions, the

362

converging air quality limits over time. more private and social rates of return will converge.

Accentuating this more optimistic outlook, Dasgupta et al. (2002) argue that developed country firms operating

in developing countries typically do so to higher 5. Effects of environmental policy on

environmental standards than domestic firms, because European business

they might simply export their cleaner technology and production methods to these countries to benefit from This section discusses the mechanisms through which economies of scale and scope, because of pressure from environmental policy gives rise to costs and benefits for

363

activists in their home markets, or because they might businesses in Europe and gives some indication of their anticipate tighter environmental legislation in these order of magnitude, where this is possible. The section countries. Moreover, this cleaner technology might also focuses on effects showing up in economic statistics be more efficient. This serves to highlight the role of such as national accounts, and neglects the broader innovation in easing any trade-off between growth and welfare effects mentioned above. the environment. In addition, it provides a channel

through which globalisation and trade liberalisation, by A widespread starting point in environmental policy is making advanced technologies more accessible, may the “polluter pays principle”, implying that those who facilitate less polluting economic growth, and so ease wish to use the environment as a dump for their

any trade-off between growth and the environment. pollution need to buy the “right” to do so. However, Coase (1960) showed that, as long as the numbers of

With respect to the EU, Scherp and Suardi (1997) find polluters and victims of pollution are both small, so that no evidence for a significant export of pollution there are no transaction costs involved in trading triggered by a relocation of polluting European property rights, from the perspective of economic industries to developing or other third countries. When efficiency it makes no difference whether property rights ranking individual industries according to the pollution in the environment are assigned to polluters or victims. content of their production processes and analysing their If polluters receive the rights to pollute, they will be trade performance they find no evidence that the willing to sell part of these rights to victims and reduce international specialisation of EU industries has shifted their output (and pollution) if they receive a price away from relatively pollution-intensive goods towards reflecting the value to them of this foregone output; if cleaner ones. Moreover, developments in overall trade victims receive the rights to a clean environment, with less-developed and developing countries have been polluters will be willing to buy part of these rights at a found to be rather similar to those in trade with price that reflects the value to them of the resulting developed countries. They explicitly emphasise the large increase in output. While the outcome in each case will and increasing net exports of the EU’s chemical industry be the same from the point of view of economic – one of the sectors with highest pollution abatement efficiency and the environment, the issue of who costs – as a representative example in this context. On receives the rights clearly has significant issues for the other hand, trade with seven newly industrialised income distribution. economies in East Asia – which has also been increasing

in both value and as a share of total extra-EU trade – has In practice, environmental pollution only rarely respects been increasingly characterized by EU imports of the “small numbers” conditions necessary for the “Coase mainly clean manufactured goods, while pollutiontheorem” to offer a complete solution to environmental intensive products have had more weight in EU exports problems, so that other forms of policy intervention are

to that region. necessary. Despite the evidence that absence of (tradable) property rights and the consequent lack of

All in all, the existence of the environmental Kuznets markets for environmental goods and services is at the curve is not evidence that growth does not harm the root of environmental problems, policy-makers have environment: decomposing the level of pollution into generally been reluctant to apply what to economists components due to the scale of output, its composition, appears to be the obvious remedy, that is, to create and and production techniques shows that, other things assign enforceable, tradable property rights, and use equal, an increase in output will lead to higher levels of market forces to address the issues. This may be because pollution. Ultimately, absolute decoupling of economic of a perception that market forces are to blame for growth and environmental pressure seems to require environmental degradation, and that therefore the

appropriate response is to restrict their functioning in

some way.

362

  See Watkiss et al. (2004). Indeed, whether or not environmental policy makes use

363

  Legrain (2003) makes a similar point in relation to of markets to achieve its aims, the main instruments in

employment conditions.

the environmental policymaker’s tool box are various – by changing the availability and price of inputs, such forms of restriction on activity in the form of constraints as the non-availability of certain dangerous on the exercise of previously unrestricted (implicit) substances or higher energy prices; property rights. That is, resources – land, water, air, the

climate, specific habitats . . . – are put under a different – by placing restrictions and additional burdens on the regime which limits access to them (for example, limit production process, such as limit values for values for emissions), limits their use (such as the kind emissions or risk-management provisions to reduce of activities which are allowed in natural habitats or occupational health risks;

drinking-water reservoirs) or makes it subject to specific – by affecting the availability, performance and price conditions (such as paying a tax or an environmental of outputs, such as fuel efficiency of cars, design levy or the obligation to clean or recycle them after use). features to facilitate better waste management, or

These restrictions may be introduced through regulation banning or taxing certain products that could be that prescribes certain categories of production harmful for the environment or human health.

technique (“best available technology”), or proscribes The first two channels mainly burden European some types of output (genetically modified organisms). producers, negatively affecting their cost Environmental regulation may also take the form of competitiveness on European and on third-country taxation to discourage some activities (example: taxes markets if non-EU producers do not face similar on landfilling in some Member States) or subsidies to constraints. The third channel imposes the same encourage others (example: subsidies for renewable obligations on European and non-European producers energies). Negotiated agreements (also called “voluntary on European markets. However, it might affect agreements”) with industry have also been used to try to competitiveness of European producers on third-country tackle environmental problems, though concerns remain markets.

about their real impact. 364 Finally, “cap-and-trade”

schemes seem to be becoming more attractive to policy

makers. The European Community has recently Static estimates of resource costs launched a large scale “cap-and-trade” scheme to help it Recent years have seen the adoption of a considerable to meet its obligation under the Kyoto Protocol to reduce volume of environmental legislation. Table 1 shows the greenhouse gas emissions. Each of these instruments European Commission’s ex ante estimates of the annual will give rise to various types of costs and benefits for costs of complying with some of the more important

different industrial sectors. elements of this legislation, taking account of significant amendments adopted by the Council and European

Creating and assigning explicit property rights and Parliament. 365

introducing new regimes to manage environmental

resources should make polluting products and/or Although these estimates have been compiled at production processes more expensive. Alternatively, different times and for different compliance periods, so environmental policy prescribes cleaner products and/or that they cannot be added together to give a figure for production which come at a higher price. Depending on cumulative compliance costs, they nonetheless suggest the market structure (competition, price elasticity of that these policies will represent a non-trivial cost to the demand) this makes regulated products more expensive targeted sectors. At the level of the whole economy, the for end-users and/or production less profitable. Both direct costs of the legislation identified above would be result in demand and production shifting towards less of the order of 0.2 per cent of GDP. Experience and the polluting products and production processes. This is an work of Morgenstern et al. (1998) discussed below give

accepted purpose of environmental policy. some grounds to expect that the actual costs may turn out to be smaller than this. In particular there may be

The costs of environmental policies ultimately fall on room for economies of scope in reducing different types consumers, who face higher prices. However, consumers of air pollutants, and in reducing air pollutants and also benefit from environmental policies in the form of limiting emissions of greenhouse gases. improvements to their health or improved amenity.

Within the business sector, costs thus fall on those using Eurostat (2001) estimated “end-of-pipe” investment by production methods that generate greater amounts of industry (excluding spending by firms specialised in pollution, or who produce products the use of which providing environmental services) at about € 7.2 billion generates pollution. Benefits accrue to businesses that in 1998. A study by Ecotec (2002) for the European produce pollution abatement equipment, or goods whose Commission found that in 1999 operating expenditure

use generates little or less pollution. relating to air pollution control amounted to some € 7.4 billion and accounted for 30,000 jobs. It seems

5.1 Costs of environmental policies

Environmental policies create costs for industry through

three channels: 365

  Other examples of cost and benefit estimations of EU environmental legislation can be found in the EU

364

  See OECD (2003). E CONOMY R EVIEW 2000, Chapter 4.

countries that have set comparatively demanding standards”. All in all, spending on environmental

Table 1: Estimated ex ante annual costs of various protection appears to be at roughly the same level in the categories of European environmental legislation EU, the USA and Japan.

(EU 15) According to the Eurostat data, about one-third of

Category Estimated cost (billion €) environmental protection expenditure by industry in the

late 1990s was for investment. Most investment Air quality & acidification 8.9-15.3 spending by industry was in “end of pipe” equipment

Climate change (fuel quality rather than on integrated, process-oriented investments. + sectors covered by 3.2 However, Eurostat points out that the latter type of

emissions trading) investment spending – that is, investment that integrates

Waste 1.4-1.9 pollution prevention in the production process rather than reducing or cleaning emissions after they have been

Product safety 0.3-0.5 generated – may be underestimated, as it is not always

Environmental liability 0.9-2.3 possible to distinguish the “environmental” component of such investments. The crucial difference from an

Source: The data are derived from the Explanatory Memoranda economic perspective between “end-of-pipe” and

and Impact Assessments accompanying the proposals, taking integrated, process-oriented investments is that the account where possible of significant differences between the former are unlikely to lead to efficiency or productivity

Commission’s proposals and the legislation actually adopted by the

European Council and Parliament. The proposals/directives gains, as they are an “add-on” to the firm’s production

included under each category are: for Air quality and acidification, process. Integrated investments, in contrast, are likely to

1 st , 2 nd , 3 rd “daughter directives” setting limit values for various imply a shift to cleaner, more energy-efficient pollutants, and proposals on large combustion plants, national technology, enabling the firm to offset at least part of the

emission ceilings, and volatile organic compounds; for Climate

change, the emissions trading directive including Kyoto project cost of complying with environmental regulation.

mechanisms, and low sulphur fuels to enable CO 2 reductions from

transport; for Waste, proposals relating to waste electrical and In total, investments in environmental protection

electronic equipment, end of life vehicles, and packaging; for represented about 4 per cent of industry gross fixed

Product Safety, proposal to restrict hazardous substances, and capital formation on average, with the share rising to

REACH (the cost of REACH has been spread over the 11 years 20 per cent or more in some branches and countries (see

over which testing is expected to take place to derive the figure in

the table). Table 2). The large number of “outliers” in the data limit their analytical value, as does the lack of time series. It

is not possible to judge whether the variability within

reasonable to assume that these figures indicate that and between sectors and countries is representative of resources were being diverted within companies from the impact of environmental policy on investment producing marketable goods and services to reducing spending in different industrial branches and countries. pollution. However, it seems unlikely that any difference in policy

could explain the range observed for refineries in

Data on pollution abatement and control expenditure different countries, for example. A more likely (PAC) are collected jointly by Eurostat and the OECD. explanation is that the differences are due to differences These data need to be interpreted with caution, but in the timing of investments.

nevertheless give an indication of the scale of the direct economic impact of environmental policy. Eurostat estimates that total environmental protection expenditure in EU15 in 1998 was about € 120 billion, or about

1.5 per cent of GDP. Of this, some 28 per cent, or about

€ 32 billion was funded directly by industry. OECD

(2004) suggests that environmental protection costs are

“likely to be equal to around 2 per cent of GDP in

Table 2: Environmental protection investments as a share of gross fixed capital formation in different

branches of industry, various countries and years

BE NL AT PT FI UK 1996 1997 1998 1997 1998 1997

Mining & quarrying 1.0 3.7 7.6 2.8 4.9 0.3

Food, beverages 3.3 3.0 6.0 1.7 4.1 3.4

Textiles, leather 1.8 2.8 3.3 0.9 2.2 0.9 Wood, wood

products 1.6 3.3 19.2 2.7 1.9 34.0

Pulp, paper, printing 1.9 2.1 6.5 6.6 7.1 5.8

Refineries 15.1 22.1 0.0 18.9 12.5 2.8

Chemicals, rubber 3.9 8.6 5.8 1.5 2.6 9.6 Non-metallic

mineral 7.0 5.1 6.2 3.1 1.2 29.7

Other

manufacturing 5.0 2.7 4.2 1.1 3.0 2.4

Electricity, gas and

water 2.6 1.0 0.7 5.0 0.6 1.0

All Industry 3.9 5.8 4.7 4.2 3.7 3.5

Source: Commission services.

None of these figures distinguish spending induced by available to firms. Compared with the previous situation, regulation from other environmental spending. generating pollution now has a cost. In trying to reduce However, it may be reasonable to assume that the this cost firms may find ways of reducing the level of overwhelming majority of spending by industry is inputs, using them more efficiently, or using cleaner related to the need to comply with environmental laws. inputs, all of which offer the potential for cost savings.

Unfortunately, comprehensive data on the interaction Accordingly, once one departs from a static, full between the environment, environmental policy and information competitive equilibrium, the notion that economic performance are not available. Although innovation can cut the costs of regulation hardly European environmental policies are usually supported represents a significant departure from conventional by ex-ante assessments of costs and benefits, ex-post economic analysis. It is in this context that Porter and policy evaluations are generally notable for their van der Linde criticise existing regulation for failing to absence. stimulate innovation, and develop a set of

Dynamic effects recommendations to ensure that future environmental In firms that are operating efficiently before the policy regulation is designed to give firms as much scope as takes effect – that is, producing as much output as possible to innovate as a way of cutting compliance possible from the inputs they use – the immediate effects costs.

of the spending it induces will thus be a decline in If environmental – or other – regulation succeeds in productivity as resources are moved from producing highlighting inefficiencies in the firm’s production output towards pollution abatement and control. process, it may yield benefits, even in the regulated However, the assumption that firms are operating firms. The scope for this depends in part on the details efficiently and that (environmental) regulation must of the particular regulation. For example, firms will face therefore inevitably lead to productivity declines is a additional costs if the measure requires process-oriented very strong one. In practice, a variety of “principalinvestment that makes existing equipment obsolete agent” problems (see Box 3) may mean that firms do not before the end of its useful life because it cannot be always operate at maximum efficiency. adapted to the needs of the new policy measure. To

Moreover, even when firms are trying to maximise calculate the costs of the policy in such cases one ideally profits, the notion of bounded rationality offers a needs to distinguish the gross costs of this new mechanism through which regulation can spur costinvestment from its net costs, that is, the value of the reducing innovation. Simon (1957) argues that firms prematurely depreciated equipment and the costs of the have to make their decisions based on incomplete parts of the new equipment that serve no other but the information, or on imperfect understanding of the new environmental purpose.

information available to them. In this framework, the effect of regulation is to change the information

Box 3: Environmental regulation and innovation: The Porter hypothesis

In a short article in Scientific American, Porter (1991) challenged the “conventional wisdom” of an inevitable trade-off between growth and the environment, arguing that “the conflict between environmental protection and economic competitiveness is a false dichotomy.” Subsequent articles with van der Linde (Porter and van der Linde, 1995, 1995a) developed this “Porter hypothesis”, as it has come to be known, and generated considerable interest and controversy.

Porter and van der Linde’s basic thesis is that regulation can stimulate innovation that reduces the costs of complying with it: “properly designed environmental standards can trigger innovation that may partially or more than fully offset the costs of complying with them”. It is not immediately obvious why this somewhat innocuous claim should have generated so much attention, particularly as it is widely recognised in the field of industrial economics that there are a number of reasons why firms will not always maximise profits in practice.

Possible explanations include “satisficing” (Simon, 1979), and “X-inefficiency” (Leibenstein, 1966), which may be regarded as particular examples of a wider class of “principal-agent” problems.1 Because owners of firms find it difficult to fully control the activities of their managers, as long as firms are earning an acceptable rate of profit for their owners, managers may be free to pursue other goals than maximising profits: “satisficing” on the part of owners may give rise to “X-inefficient” behaviour on the part of managers. An environmental regulation that raises the cost of pollution creates a new set of conditions. In trying to reduce compliance costs and restore profits to a “satisfactory” level, it is possible that firms may discover other potential savings. However, that this cost saving actually materialises cannot, of course, be taken for granted.

1 See also Leibenstein (1978) and Stigler (1976).

The direct resource costs of complying with amount of compliance expenditures reported by firms: environmental policy measures (as with all forms of for every dollar of reported environmental expenditure, regulation) will in all likelihood give rise to secondary overall production costs rose by 82 cents. In other effects by affecting productivity, profitability, prices, words, the economic costs of environmental regulation demand dynamics, innovation and investment decisions are less than the direct costs. The authors hypothesise of the affected businesses. As an example, fitting flue that this is because of complementarities between the gas desulphurisation units to clean the emissions of production of goods and services and pollution control: power plants can reduce the efficiency of the plant, “the costs of jointly producing conventional output and a increasing the amount of fuel input needed to generate a cleaner environment may be lower than if each were

given amount of power output. 366 produced separately”. For example, it may be cheaper to

The secondary effects also depend to a large extent on reduce air pollution by replacing a coal-fired generation how the affected businesses finance their compliance plant with a more energy-efficient gas-fired plant, rather costs (additional borrowing on capital markets, price than keeping the coal-fired station in operation and increases, cuts in dividends, cost savings by cutting fitting pollution control equipment to “scrub” the R&D spending, etc.), and market structures (price emissions after they have been produced.

elasticity of demand, international competition, etc.). If correct, this interpretation reinforces the arguments in

The relationship between direct and indirect costs is not favour of regulation that encourages integrated at all straightforward. For example, if the firm redirects approaches to pollution abatement, rather than “end-ofits research budget towards innovations that could lower pipe” solutions. It may also be that there are “economies the long-term cost of complying with an environmental of scope” in pollution abatement. That is, reducing one regulation, this may simultaneously reduce direct costs pollutant may also contribute to reducing others. This (compliance costs are lower) while increasing or seems particularly likely to be the case for actions to decreasing indirect costs (the environmentally-induced reduce the wide range of atmospheric pollution innovations may generate smaller or bigger profits for associated with burning fossil fuels.

the firm than the innovations that might have been made Morgenstern et al’s results differ from some earlier if R&D spending had not been refocused). Similarly, in research that showed indirect effects considerably higher the extreme case in which a firm were to close as a than the direct compliance costs. The authors argue that direct consequence of environmental regulation (though these earlier results failed to take adequate account of evidence that this has taken place is non-existent), differences between plants in terms of how they are recorded direct costs would be zero, but indirect costs affected by regulation and able to react to it, and assume could be substantial. that factor inputs are fixed. Indeed, taking an alternative

Ex-post estimates modelling approach that ignores these differences, they In an analysis of US data Morgenstern et al. (1998) get results that are broadly consistent with the other found that production costs actually rose by less than the studies.

Haq et al. (2001) highlight the role of unanticipated

366 innovation in reducing the expected costs of a number of

  See Stockholm Environment Institute (1999).

environmental regulations, based on a study from the underestimated the quantity of emission reductions. In Stockholm Environment Institute (SEI) (1999). The other words, market-based approaches produced greater phasing out of ozone depleting substances (largely environmental benefits at lower cost. However, chlorofluorocarbons – CFCs) under the Montreal Harrington et al. do not report any examples of Protocol was forecast (mainly by industry) to lead to regulation giving rise to negative costs to the regulated large-scale redundancy of existing equipment and a firms, as the Porter hypothesis might imply. corresponding need for high levels of investment in

replacement capital. At its peak, the market for CFCs A recent OECD review noted that the failure of was worth over $2 billion, and it was expected that the countries to systematically analyse costs and benefits main replacement substances might be up to 10 times made it difficult to assess the overall welfare more costly. In the event, costs of the phase out have implications of environmental policy measures. been much less than anticipated, both because the direct However, based on the evidence from OECD member replacements have been cheaper to produce than countries, it appeared that air pollution policies expected, and because of innovation that reduced the delivered benefits significantly greater than the marginal

need for their use. abatement costs, whereas there were doubts as to whether current programmes for greenhouse gas

In the case of the European Auto-Oil programme to emissions, waste management and water pollution had reduce emissions from road transport, in the mid-1990s “delivered benefits at the margin that are commensurate

it was estimated that meeting the Euro IV standard for with costs”. 368

cars would require advanced catalyst technology costing

at least € 100-175 per car; this estimate was itself lower Overall, these results suggest that the trade-off between than earlier figures. More recent estimates suggest that environmental policy and economic growth may not be fine-tuning existing technology can meet the standard particularly severe. However, they do not provide

for at most half this cost. grounds to argue that there is no trade-off: the seeming absence of any substantial impact of environmental

A recent review of EU air pollution policies carried out policies on economic growth to date does not mean that for the European Commission (DG Enterprise) one can ignore its potential effects. concluded that there was very limited evidence for there

being significant competitiveness effects due to 5.2 Benefits of environmental policy to business European air pollution legislation. 367 The main reasons

they give for the lack of impact are: Much of the money spent on environmental protection by sectors that have to comply with environmental

– broad similarity in the stringency of environmental regulations is paid to firms providing environmental legislation across major industrialised economies; goods and services, who thus benefit from

– technological progress offsetting cost increases due environmental policy. These firms might be part of the

to environmental legislation; regulated sector, such as the providers of catalytic converters for passenger cars, or they might belong to

– the relative lack of importance of environmental other sectors, such as the providers of scrubbers for legislation relative to other factors influencing large combustion plants. According to Eurostat (2002), location decisions, such as cost of labour, access to about 40 per cent of current spending on environmental inputs and markets, and overall economic and protection by industry goes to purchase environmental political stability services from other organisations, whether public or

Porter and van der Linde (1995, 1995a) give evidence private: this is particularly the case for waste and from a number of case studies showing how innovative wastewater treatment.

responses to environmental constraints saved firms This implies that most current spending on money. In a slightly different vein, Harrington et al. environmental protection takes place “in house”, that is, (1999) compared ex ante and ex post estimates of the in the firms that are subject to environmental regulation. cost of a sample of 25 environmental regulations in the As discussed above, this spending diverts resources United States, and found some tendency for actual from the main activities of these firms and reduces their compliance costs to be lower than forecast costs. The output. However, this money does not go up in smoke, reasons the authors identified for this tendency toward as it were, but is instead spent in a different way than overestimating costs included changes in the regulation previously. The effect of the policy is to oblige firms to after the ex ante analysis had been undertaken, using transfer resources from one type of activity – production maximum cost estimates, over-estimating the amount of of marketed goods and services – to another – pollution emissions reduction, and, “in numerous instances”, abatement. Taking account of this “in house” spending, unanticipated technical innovation. Ecotec (2002) found that spending on environmental

Moreover, all the regulations based on economic protection accounted for 2 million jobs in EU-15.

incentives either overestimated the cost or

367

  See Watkiss et al. (2004). 368 See OECD (2004).

As environmental policy directly or indirectly raises the owners’ and/or managers’ attention to various types of price of polluting, firms who use less polluting resources inefficiencies in the way firms operated before the or produce less polluting products benefit as demand measure took effect. Better resource use could be shifts towards their output. Benefits also accrue to firms triggered if the need to reduce pollution focuses who use previously polluted resources as inputs for their company attention on using its inputs more efficiently. production: reducing water pollution benefits activities This could induce positive effects on innovation, as that require clean water. Just as environmental well-designed regulatory instruments generally enable regulation may reduce the productivity of firms in the companies to seek innovative solutions that otherwise regulated sectors, it may increase the productivity of would remain unexplored.

firms elsewhere in the economy. At an aggregate level, the output of European

Moreover, entire industries, such as the manufacture of manufacturing industry increased by 29 per cent from wind turbines or photovoltaic cells for solar energy, 1985 to 1999, while energy consumption was have in large part been created by environmental unchanged. This improvement took place at a time of policies: Ecotec estimated that spending on renewable falling real energy prices. Several factors explain this energy plant was roughly € 5 billion in 1999 in EU-15. improved performance. Structural change in They also found that the EU-15 had a trade surplus in manufacturing industry has probably been towards less environmental goods and services of a similar order of energy-intensive activities, while there have also been magnitude. This is consistent with one interpretation of improvements in the energy efficiency of particular the Porter hypothesis, that regulation can generate manufacturing processes. Some of this change would international competitive advantage by giving firms and have occurred anyway, but part of it is likely to be due the economy a “first mover” advantage, notably in to the impact of regulation, including higher energy

environmental technology. taxes that partly offset falls in energy costs. 369

However, as Porter and van der Linde point out, Over the longer term, the positive impacts on human environmental regulation will not necessarily give rise to health – often the main driver for environmental policy – a first mover advantage. Whether for regulated firms, or should have wider economic benefits, both in the form for firms supplying environmental technologies, an early of reduced health spending, and also by contributing to a mover advantage only arises if “national environmental workforce that is more productive (because healthier) standards anticipate and are consistent with international and larger (and therefore cheaper). In a study focussing trends in environmental protection, rather than break exclusively on this issue, Holland et al. (1999) estimated with them.” In other words, taking the lead in deploying that in the case of EU policies to limit air pollution, the renewable energies will not yield international benefits of improved worker health, in terms of reduced competitive advantage if other countries do not follow levels of absence from work, would be of the order of suit. In this respect, it is noteworthy that Ecotec found 10 per cent of abatement costs over the period 1996-

that the EU-15 had a deficit of € 0.2 billion in trade in 2010. 370

photovoltaic products: the economic rationale for

promoting solar energy in northern Europe is not 5.3 Overall impact

immediately apparent. Econometric studies using a production function

Even when other countries do adopt similar regulation, framework (see Box 4) generally find significant the regulated sector will not necessarily be better off (though not always very large) negative impacts of than it was before being regulated. The “first mover” regulation, mainly on the productivity of the regulated advantage enables the sector to comply at lower cost industry. 371 It must be kept in mind that production than its competitors in other countries. But if the theory focuses on the microeconomic effects, taking into (partial) pass through of compliance costs leads to lower account the optimal behaviour of individual firms. It overall demand for the sector’s output, the result of the does not capture possible externalities, offsetting first mover advantage may be that firms secure a larger dynamic effects through technological innovations, or share of a smaller market, so that the net impact on more general welfare effects. For example, increased output and profits is ambiguous: “first mover” advantage environmental quality could increase the health of does not necessarily imply faster growth than would workers which increases the efficiency of labour. 372 have occurred in the absence of regulation. In the Another offsetting effect not directly modelled in this particular case of renewable energies, the industry’s framework is a possible link between the levels of development has come at the cost of higher prices for abatement costs on the rate of innovation. Some recent electricity than would otherwise have been the case and, papers dealing with the direction of technological

presumably, reduced demand for investment in conventional electricity generating technologies.

369 See European Commission (2002). Benefits to business may also result if environmental 370

regulation induces changes in firm behaviour, The study did not attempt to estimate the wider health and environmental benefits of the policies.

particularly in the longer term. As already described, 371

this depends on the ability of the regulation to draw See, for example, Gray and Shadbegian (2002). 372

  See for example Bloom et al. (2001).

change suggest that it is optimal for firms to concentrate EPA concludes that it is extremely unlikely that these innovative activities on economising on those factors uncertainties could overturn the favourable benefit-cost

whose relative price rises more strongly. 373 This ratio.

argument is used to explain why technical progress in

industrial economies tends to be labour saving and not The EPA used a macroeconomic model to estimate the capital saving. In consequence, if economic agents overall impact of the Clean Air Act on economic expect prices for environmental resources to rise more activity. They found that it had reduced the rate of than prices of other factors, innovative activities would growth of GNP by 0.05 percent on average from 1973 to

be channelled towards economising on this factor. 1990, so that by 1990 GNP was approximately 1 per cent – $ 55 billion – lower than it would have been in

While it is possible that environmental policy acts as a the absence of the policy. This was due to slower rates drag on growth in the regulated industries, it is also of capital accumulation and productivity growth. It possible that – as outlined in Section 3 – the effect is to should be noted, however, that the model was unable to accelerate growth by improving the supply of inputs. If capture feedback effects of improved health in terms of the health effects of pollution are adversely affecting reduced medical expenditure and improved worker labour supply, or the quality of natural resource inputs is productivity. Over the entire period considered, being damaged, environmental policy that successfully aggregate macroeconomic costs were estimated at $ 1 tackles these problems will be beneficial for economic trillion (in discounted 1990 dollars), that is, activity. Some recent papers show that the positive approximately twice the direct compliance costs, and welfare effects of improved health conditions can be less than 5 percent of the estimated welfare benefits.

374

large.

Evidence on crowding out of dirty industries to pollution havens in third countries seems to be very shaky and not convincing at all. This might not come as a surprise given that other factors normally drive decisions of investment locations, and given the convergence of environmental standards around the world, including developing countries.

The data and case studies above give some indication of the scale and nature of the impacts of environmental policy on economic activity, but do not allow any clear picture to be formed of its overall economic effects. So far, no comprehensive attempt appears to have been made to measure ex post the economic impacts of environmental policy in Europe. However, the United

States Environmental Protection Agency has tried to

375

estimate the costs and benefits of the Clean Air Act.

The results of this exercise, although the details are clearly valid only for the USA, may nevertheless give some broad indication of the likely order of magnitude of impacts of European policies, as air quality standards

in European and US legislation are broadly similar. 376

Overall, the EPA found that the benefits of the Clean Air

Act were substantially greater than the costs, mainly due to increased life expectancy. Over the 1970-1990 period, the central estimate of benefits was $22 trillion in 1990

US dollars, while direct compliance costs over the same period were $0.5 trillion. By far the largest component of the benefits – close to 90 per cent – was due to increased life expectancy because of reduced exposure to particulate matter and lead. Although there are considerable uncertainties about these figures – and the estimate of costs does not include indirect costs – the

373

  See, for example, D. Acemoglu (2003).

374

  See, for example, Nordhaus (2002).

375

  US EPA (1997).

376

  See Watkiss et al. (2004).

Box 4: The treatment of environmental resources and policies in neoclassical production functions

A standard tool for macroeconomic analysis is the neoclassical production function which relates total output (Y) of a certain industry to a comprehensive list of inputs. At an industry level one can distinguish between labour (L), capital (K), energy (E), raw materials (R) and intermediate inputs (M, goods and services supplied by other sectors (both domestic and foreign)) as factors of production. Obviously, the level of disaggregation of these individual input categories depends on data availability. For example energy could be further disaggregated into different types of energy. However, environmental resources enter such production functions only in so far as they are raw materials or energy inputs. In its most general form a production function can be written as follows

Y = F(L, K, E, R, M) TFP (i)

For empirical analysis specific functional forms must be chosen. For simplicity we assume a Cobb Douglas specification

Y = L α K β E γ R η ν M TFP (ii)

where, α, β, γ, η, ν represent a kind of marginal productivities or, more correct, the output elasticities of the respective factors of production (Labour L, Capital K, Energy E, natural resources R and intermediate inputs M) and TFP or total factor productivity summarises the level of efficiency of production. TFP can itself be a function of various underlying factors such as the human capital endowment, the level of knowledge generated by national innovation systems (universities, research labs) or diffusion of knowledge. It can also be influenced by institutional factors*.

Environmental regulation can affect TFP in the standard production function framework both when it materialises as an increase in the price for a specific input, such as energy and when it materialises as a regulation requiring end-of-pipe technologies:

In case of energy tax the relative price of energy with respect to output increases. Assuming that the firm behaves optimally, the demand for energy is given by a first order condition of a cost minimisation problem from which a new demand function for energy can be derived. Substituting the optimality condition into the production function establishes a direct link between Y and the tax rate on energy. In the Cobb Douglas case the output loss in the regulated industry of an increase in the price of energy is proportional to the output elasticity of energy.** In general this is an underestimate of the total output effect of the energy tax since an increase in the price of energy and the subsequent reduction of its use is predicted to be associated with a fall in the marginal product of all other factors of production by standard production theory. The degree in which the use of other factors is reduced depends on the degree of factor price rigidity of the other factors.

Similarly, the need for investing in additional end-of pipe technologies imposed upon sectors by tightened environmental regulation would show up in the production function as the need to increase the amount of intermediate inputs M, without being able to correspondingly increase the output Y, so that the output elasticity ν of this input declines. Eventually, the degree of output decline in the regulated sector will then depend on the price elasticity of demand for this output. The output increase in the sector producing the abatement technology is given by the increase in M.

• * 

  See, for example, EU E CONOMY R EVIEW 2003, Chapter 2 for an empirical analysis of TFP at the aggregate level.

** In general the output loss also depends on the elasticity of substitution between individual factors. This is hidden in the Cobb Douglas specification because the elasticity of substitution is one in this case.

• 6. 

  Implications for regulation – finding the stalemate between regulators and firms that, in their

right balance view, unnecessarily exacerbates the trade-off between the environment and growth. 377 They urge regulators to

This section aims to identify the conditions under which design regulations in ways that stimulate innovation, and environmental regulation can relax the potential tradecall on companies to discard their adversarial mind-set. off with economic growth for the regulated sectors, and In so doing, they highlight the importance of good contribute to growth-enhancing structural adjustment. policy design in reducing trade-offs.

The key to achieving such a result lies in minimizing the A recent Commission staff working paper set out a impact of regulation on costs for the regulated sector number of useful guidelines for designing environmental (without compromising on environmental and public policy so as to minimise any unavoidable trade-offs

health objectives), and in stimulating innovation and between environmental and economic policy goals: 378

adjusting price signals to new demand-supply trends instead of working as a drag on economic growth.

That this is possible is shown by Porter and van der

Linde (1995a, 1995b). They do not aim to show that 377 Schmalensee (1993) makes a similar point: “[Porter’s

there is no trade-off between environmental protection message to the business community] appears to be that the

and economic growth. Rather, by showing that social and political demand for environmental protection is environmental regulation can be designed to allow firms unlikely to diminish and that “Just say no!” is unlikely to to comply in innovative ways that enables them to be the profit-maximising response strategy...” generate a competitive advantage, they seek to end the 378 See European Commission (2004).

– Market-based proportionality: Policies should – Regulation should be as simple as possible, but no intervene as little as possible in the functioning of simpler: Companies should be clear about what they market mechanisms. Instead, they should try to have to do to comply with legislation. Unnecessarily exploit as much as possible the driving forces complicated reporting and regulatory oversight embedded in market transactions by giving actors should be avoided. However, the simplicity of incentives to achieve the environmental objectives at regulation must not negatively affect either its lower cost and by better synchronising investment proportionality or its (cost-)effectiveness.

requirements of regulation with company investment

plans. – A stable policy framework: Policies should try to avoid sudden surprise movements that make large

– Include a “safety margin”: Although from a parts of the existing capital stock prematurely theoretical perspective a policy is optimal when obsolete and overstretch the adjustment capacities of marginal benefits equal marginal costs (that is, the targeted industries. Instead, environmental standards cost of achieving additional reductions in pollution should be implemented gradually but credibly. This would be greater than the benefit of those implies that regulation should aim to enable industry reductions), uncertainty about the precise level of to incorporate environmental policy requirements benefits and regulatory prudence point to a need to into its investment decisions. The immediate losers – include a “safety margin” in the level of ambition of owners of capital and labour in the regulated sectors the policy. A serious sensitivity analysis in the – should be given adequate time to adjust.

context of an ex-ante impact assessment should give

some guidance in this respect. This might be These principles point to a clear preference for marketregarded as the economist’s equivalent of the based regulatory approaches that set the standard that

environmentalist’s “precautionary principle”. firms have to reach, but leave it up to firms as to how they reach it. This is in contrast to more widespread

There are two important qualifications to this cautious regulatory approaches that prescribe what firms have to approach. The first is the possibility that prior estimates do to comply. More recent Community environmental of costs may be higher than actual compliance costs, as legislation (such as in the context of the European suggested in the review by Harrington et al. already acidification strategy or the national emissions ceilings discussed. This may be because up-front regulatory cost directive) often tries to take account of economic estimates depend to a large extent on information from constraints such as investment cycles, abatement those who are targeted by the regulation, who have an technologies available, and so on. However, a obvious incentive to overestimate its costs. significant part of environmental protection spending

A second issue relates to the potential for regulation to continues to be on “end-of-pipe” investments.

stimulate cost-saving innovation. Porter and van der On the one hand, this may suggest that regulation Linde (1995b) argue for strict, rather than lax regulation, continues to be overly prescriptive. An alternative on the grounds that incremental tightening of regulatory possibility is that “end-of-pipe” solutions are more coststandards will only lead to incremental responses from effective, given the currently available technologies. If it industry. They argue that if regulation is to spur is the case that end-of-pipe solutions are cheapest, then innovation, it must be stringent, so that incremental or again, a number of conflicting interpretations are marginal changes to current techniques are not feasible possible. It may be that regulations are too ambitious, or ways of complying. This appears to be rather a high-risk that they do not give companies enough time to adapt. approach to regulatory design, as any such proposal Alternatively, in line with the arguments of Porter and could hardly pass an up-front cost-benefit analysis. van der Linde, it could be that regulation is not Minimum conditions necessary for such “leap in the ambitious enough, so that it fails to encourage more dark” policy approaches must surely be a relatively long innovative approaches to pollution abatement. A further timeframe for meeting the ultimate policy objective, that possible explanation is that “end-of-pipe” solutions may is, a gradual but credible and predictable tightening of have been an appropriate way to address relatively regulation, and a commitment to review progress straightforward issues such as pollution from large point regularly. This seems to be important for the regulated sources, but that as the problems tackled by sector and for the industries providing abatement environmental policy become more complex and technologies and services. diffuse, greater recourse to market-based instruments

– Cost effectiveness: Policies should be designed and will be necessary.

implemented so that they can achieve their The preference for flexible, market-based approaches environmental aim at least cost. In principle this over traditional regulation arises because the latter implies using market-based approaches or generally is unable to take account of the specificities of differentiated regulation that makes best use of individual firms, and for this reason will generally not information available at the level of enterprises and be the lowest-cost solution. Unlike market-based that takes adequate account of the investment cycle approaches, prescriptive regulation does not give firms and abatement costs that are faced by specific incentives to outperform whatever standard is set for sectors. them. Nevertheless, this may be the preferred choice

when it is necessary to avoid “hot spots” of local including ecological tax reforms, in which pollution, or when it is imperative that a particular environmental tax revenues are used to reduce other, objective be exactly met. more distorting taxes. At Community level, however,

Without compromising on environmental effectiveness, the requirement that fiscal measures be adopted market-based instruments will in many situations be unanimously by the Council is an extra obstacle, making cheaper than alternative regulatory approaches. This is the Commission reluctant even to table such because market-based instruments offer firms greater proposals.

381 These obstacles make it all the more

flexibility, and give them incentives to devise new, important that regulatory proposals are based on a cleaner production techniques that reduce the cost of thorough assessment of their impacts, so that any trademeeting environmental targets. To be effective in offs between competing environmental and economic reducing pollution at low cost relative to other policy objectives can be identified.

possibilities, market-based instruments require pricesensitive markets. However, even when markets are

inelastic, market-based instruments can be expected to 7. Conclusions

be more economically efficient than alternative forms of

regulation as a way to achieve a particular This chapter has examined in how environmental environmental policy target. regulation could enhance the overall efficiency of the

economy and therefore encourage economic growth.

The forthcoming European Community greenhouse gas Explicitly or implicitly, environmental regulation takes emissions trading scheme is a flagship for the use of the form of defining and assigning or re-assigning

market-based approaches to addressing environmental

st property rights. This is comparable to taking away or problems. From January 1 2005, electricity generators reducing a “subsidy” from a sector (polluter) that is

and the more energy-intensive sectors of manufacturing “financed” by others (victims). This corrects a distortion industry will face an aggregate ceiling on their in relative prices, suggesting that, implemented emissions of carbon dioxide, the main greenhouse gas. appropriately, environmental protection can be Emission allowances have been allocated to the beneficial for the environment and the economy.

operators of individual plants, who will have to

surrender a quantity of allowances each year matching However, what might be good for the economy might their actual emissions. Allowances may be traded on a not necessarily show up in higher economic growth but Community-wide market, giving incentives to operators “only” in higher welfare. The benefits of nature to find low cost ways of reducing emissions: operators protection, for example, may or may not show up in who reduce their emissions below the level of their terms of higher levels of economic activity, though the allocation may sell their “spare” allowances to operators costs will certainly fall on the economy. In such cases, who have fewer allowances than they need. although the policy may yield benefits for society as a

whole, there is a trade-off between environmental policy

Recent analysis estimates the annual compliance costs and economic growth as measured in national accounts. for the sectors covered by the Community emissions The aim should then be to ensure that the regulation is trading scheme to be € 2.2 billion in the first Kyoto cost-effective so that it internalises the costs of pollution Protocol commitment period (2008-2012), based on an while minimizing negative economic or social

allowance price of about € 13 per tonne of carbon

379 implications for the regulated sectors and their

dioxide. Some ways to reduce emissions, such as customers.

substituting biofuels for conventional energy sources,

give rise to abatement costs of over € 100 per tonne of The discussion above of the determinants of the carbon dioxide, so it is clear that the emissions trading environmental Kuznets curve provides additional insight scheme has the potential to lower abatement costs by into the relationship between environmental policy and several billion euros compared with some alternatives. 380 economic activity. In the absence of technological

progress and/or changes in the composition of output,

Despite their advantages over “traditional” regulation, economic growth will lead to higher levels of pollution. market-based instruments face obstacles in practice, not As the purpose of environmental policy is presumably least because they make the price of pollution more neither to slow growth, nor to reduce the output of transparent. This makes the costs of implementation particular sectors, it is important that it allows maximum clearer, and draws attention to the changes in income scope for innovative technological solutions to distribution that will result. EU Member states are environmental problems.

increasingly using environmental taxes and charges,

381

379 The 2003 Directive on energy taxation was only adopted European Commission (2003a). In late July 2004, the after many years of negotiations, and did not require market price was less than € 10 per tonne of carbon significant changes to tax levels in several member States. dioxide. Nevertheless, the Directive provides a common framework

380

  Notwithstanding their high cost, these alternatives may for taxing energy products in the EU and in this way may make a contribution to other policy objectives, such as offer a basis for future environmentally-related tax security of energy supplies. adjustments.

As regards the overall impacts of environmental policy on economic growth, an acute lack of data means that no firm conclusions can be drawn. Comparison with the effects of the Clean Air Act in the United States suggests that the impacts to date may have been modest, and in any event substantially outweighed by the wider environmental and social benefits.

Nevertheless, given the aim of the Lisbon strategy to make simultaneous progress towards economic, environmental and social objectives, this lack of information about the interaction between environmental policy and the economy is a serious drawback. Priority should be given to filling this gap in our knowledge by carrying out systematic ex post analyses of the

(economic) impact of Community environmental policies. This will provide much-needed information about the scale of trade-offs that have been made in the past, and will help policy makers to design future interventions so as to maximise the potential for “winwin” outcomes.

References

A CEMOGLU , D. (2003), “Directed Technical Change”, Journal of the European Economic Association 2003.

A NDREONI J. AND A. L EVINSON (1998), “The Simple Analytics of the Environmental Kuznets Curve”, NBER Working Paper No 6739.

B ALMFORD A., ET AL . (2002), “Economic reasons for conserving wild nature”, Science Vol. 297.

B LOOM , D. E., C ANNING , D. AND J. S EVILLA (2001), “The Effects of health on Economic Growth: Theory and Evidence”, NBER Working Paper No 8587.

C OASE , R. (1960), “The Problem of Social Cost”, Journal of Law and Economics.

C OPELAND B. AND M. T AYLOR (2004), “Trade, Growth and the Environment”, Journal of Economic Literature, Vol. XLII.

D ASGUPTA P.S. AND G.M. H EAL (1979), “Economic theory and exhaustible resources”, Cambridge University Press.

D ASGUPTA S., B. L APLANTE , H. W ANG , D. W HEELER (2002), “Confronting the Environmental Kuznets Curve”, Journal of Economic Perspectives, Vol. 16, No 1.

E COTEC FOR THE E UROPEAN C OMMISSION , DG Environment; (2002), “Analysis of the EU Eco-Industries, their Employment and Export Potential”.

E UROPEAN C OMMISSION (1994), “Economic growth and the environment: some implications for economic policy making”, COM(94) 465 final i.

E UROPEAN C OMMISSION (1995), “Towards fair and efficient pricing in transport - Policy options for internalising the external costs of transport in the European Union”, COM (95) 691 i.

E UROPEAN C OMMISSION (2002), “European Competitiveness Report” (2002), Commission staff working paper SEC(2002)528, 2002.

E UROPEAN C OMMISSION (2003), “Proposal for a Directive of the European Parliament and Council amending the Directive establishing a scheme for greenhouse gas emission allowance trading within the Community, in respect of the Kyoto Protocol’s project mechanisms”, COM(2003) 403 final i, 2003.

E UROPEAN C OMMISSION (2003a), “Extended impact assessment on the Directive of the European Parliament and Council amending the Directive establishing a scheme for greenhouse gas emission allowance trading within the Community, in respect of the Kyoto Protocol’s project mechanisms”, SEC(2003) 785, 2003.

E UROPEAN C OMMISSION (2003b), “Towards a Thematic Strategy on the Sustainable Use of Natural Resources”, COM(2003)572 final i.

E UROPEAN C OMMISSION (2004), “The effects of European environmental policy on European business and its competitiveness – a framework for analysis”, SEC(2004) 769.

E UROSTAT (2002), “Environmental protection expenditure by industry in the European Union, Statistics in Focus” 14/2002.

E UROSTAT (2001), “Environmental protection expenditure in Europe”, data 1990-99, detailed tables.

E UROSTAT (2001a), “NAMEAs for air emissions”.

G RAY , W. B. AND R. J. S HADBEGIAN (2002), “Pollution Abatement Costs, Regulation and Plant Level Productivity”, in Gray. W. B. (ed.) “The Economic Costs and Consequences of Environmental Regulation”, Ashgate Publications.

H AQ G., ET AL . (2001), “Determining the costs to Industry of environmental regulation”, European Environment Vol. 11.

H ARRINGTON W., R. M ORGENSTERN , P. N ELSON (1999), “On the Accuracy of Regulatory Cost Estimates”, Resources for the Future Discussion Paper No 99-18.

H OLLAND M.R., H. J ONES , R. B OYD , A. H UNT , A. M ARKANDYA (1999), “Environment, Competitiveness and Health”, AEA Technology/Metroeconomica for the European Commission (Institute for prospective technological studies).

I NTERGOVERNMENTAL P ANEL ON C LIMATE C HANGE (2001a), “Climate Change 2001: The Scientific Basis”, Technical Summary.

I NTERGOVERNMENTAL P ANEL ON C LIMATE C HANGE (2001b), “Climate Change 2001: Impacts, Adaptation and Vulnerability”, Technical Summary.

I NTERGOVERNMENTAL P ANEL ON C LIMATE C HANGE (2001c), “Climate Change 2001: Mitigation, Technical Summary”.

J ACOBS (2004), “An economic assessment of the costs and benefits of Natura 2000 sites in Scotland”, Environment Group Research Report 2004/05, Scottish Executive.

J AFFE A., S. P ETERSON , P. P ORTNEY , R. S TAVINS (1995), “Environmental Regulation and the Competitiveness of U.S. Manufacturing: What Does the Evidence Tell Us?”, Journal of Economic Literature, Vol. XXXIII.

J OUZEL , J., C. L ORIUS , J.R. P ETIT , C. G ENTHON , N.I. B ARKOV , V.M. K OTLYAKOV , AND V.M. P ETROV (1987), “Vostok ice core: a continuous isotope temperature record over the last climatic cycle (160,000 years)”, Nature 329:403-8.

J OUZEL , J., N.I. B ARKOV , J.M. B ARNOLA , M. B ENDER , J. C HAPPELLAZ , C. G ENTHON , V.M. K OTLYAKOV , V. L IPENKOV , C. L ORIUS , J.R. P ETIT , D. R AYNAUD , G. R AISBECK , C. R ITZ , T. S OWERS , M. S TIEVENARD , F. Y IOU , AND P. Y IOU (1993), “Extending the Vostok ice-core record of palaeoclimate to the penultimate glacial period”, Nature 364:407-12.

J OUZEL , J., C. W AELBROECK , B. M ALAIZE , M. B ENDER , J.R. P ETIT , M. S TIEVENARD , N.I. B ARKOV , J.M. B ARNOLA , T. K ING , V.M. K OTLYAKOV , V. L IPENKOV , C. L ORIUS , D. R AYNAUD , C. R ITZ , AND T. S OWERS (1996), “Climatic interpretation of the recently extended Vostok ice records”, Climate Dynamics, Vol. 12, pp. 513-521.

K UZNETS S. (1955), “Economic Growth and Income Inequality”, American Economic Review, Vol. XLV, No 1.

L EGRAIN , P. (2003), “Open World: the truth about globalisation”.

L EIBENSTEIN , H. (1966), “Allocative Efficiency vs. “X-Efficiency””, American Economic Review, Vol. 56.

L EIBENSTEIN , H. (1978), “X-inefficiency Xists – Reply to an Xorcist”, American Economic Review, Vol. 68.

L EONARD , H.J. (1988), “Pollution and the struggle for the world product. Multinational corporation, environment, and international comparative advantage”, Cambridge University press, Cambridge, United States.

L OMBORG B. (2001), “The Skeptical Environmentalist”, Cambridge University Press.

M EADOWS D. H., D. L. M EADOWS , J. R ANDERS , W. B EHRENS III (1972), “The Limits to Growth” , Earth Island Ltd.

M ORGENSTERN R., W. P IZER , J-S S HIH (1998), “The Cost of Environmental Protection”, Resources for the Future Discussion Paper No 98-36.

N ATIONAL G EOGRAPHIC (2004), “The end of cheap oil”, June 2004.

N ORDHAUS , W. D. (2002), “The Health of Nations: The Contribution of improved Health to Living Standards”, NBER Working Paper 2002.

OECD (2003), “Voluntary approaches for environmental policy: effectiveness, efficiency and usage in policy mixes”.

OECD (2004), “Integrating Concerns for sustainable development in the policy process: synthesis of the EDRC peer reviews”.

P ETIT , J.R., J. J OUZEL , D. R AYNAUD , N.I. B ARKOV , J.-M. B ARNOLA , I. B ASILE , M. B ENDER , J. C HAPPELLAZ , M. D AVIS , G. D ELAYQUE , M. D ELMOTTE , V.M. K OTLYAKOV , M. L EGRAND , V.Y. L IPENKOV , C. L ORIUS , L. P EPIN , C. R ITZ , E. S ALTZMAN , AND M. S TIEVENARD (1999), “Climate and atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica”, Nature 399: 429-436.

P ORTER M. (1991), “America’s green strategy”, Scientific American.

P ORTER M. AND C. VAN DER L INDE (1995a), “Toward a new conception of the Environment-Competitiveness Relationship”, The Journal of Economic Perspectives, Vol. 9, No 4.

P ORTER M. AND C. VAN DER L INDE (1995b), “Green and Competitive: Ending the Stalemate”, Harvard Business Review.

P RETTY J.N., C. B RETT , D. G EE , R.E. H INE , C.F. M ASON , J.I.L. M ORISON , H. R AVEN , M.D. R AYMENT , G. VAN DER B IJL (2000), “An assessment of the total external costs of UK agriculture”, Agricultural Systems, Vol. 65.

S CHERP , J, S UARDI , M. (1997), Trade and environment, in : European Commission (Ed.) “The European Union as a World trading partner”, European Economy, Reports and Studies, No 3.

S CHMALENSEE R. (1993), “The costs of environmental protection”, Massachusetts Institute of Technology Centre for Energy and Environmental Policy Research Working Paper No 93/015.

S CHOER K., C. F LACHMANN , A. H EINZE , D. S CHÄFER , B. W ALDMÜLLER (2001), “Bericht des Statistischen Bundesamtes zu den Umweltökonomischen Gesamtrechnungen (UGR)“ [“Environmental-Economic Accounting in Germany 2001”].

S HELL I NTERNATIONAL (2001), “Energy Needs, Choices and Possibilities – Scenarios to 2050”.

S IMON H. (1956), “Rational Choice and the Structure of the Environment”, Psychological Review.

S IMON H. (1979), “Rational Decision Making in Business Organizations”, The American Economic Review, Vol. 69.

S OMMER H., R. S EETHALER , O. C HANEL , M. H ERRY , S. M ASSON , J.-C. V ERGNAUD (1999), “Health costs due to road traffic-related air pollution”, World Health Organisation.

S TIGLER G. (1976), “The Xistence of X-efficiency”, The American Economic Review, Vol. 66, No 1.

S TOCKHOLM E NVIRONMENT I NSTITUTE (1999) for the Swedish Environment Ministry, “Costs and Strategies presented by Industry during the Negotiation of Environmental Regulations”.

U NITED N ATIONS , E UROPEAN C OMMISSION , I NTERNATIONAL M ONETARY F UND , O RGANISATION FOR E CONOMIC C O - OPERATION AND D EVELOPMENT , W ORLD B ANK (2003), “Integrated Environmental and Economic Accounting”.

U NITED S TATES E NVIRONMENTAL P ROTECTION A GENCY (1997), “The Benefits and Costs of the Clean Air Act, 1970 to 1990”.

W ATKISS P., D. F ORSTER , A. H UNT , A. S MITH , T. T AYLOR (2004), “A comparison of EU Air Quality Pollution Policies and Legislation with Other Countries”, AEA Technology Environment and Metroeconomica for the European Commission (DG Enterprise).

W ORLD B ANK (1992), “World Development Report 1992: Development and the Environment”.

W ORLD C OMMISSION ON E NVIRONMENT AND D EVELOPMENT (“The Brundtland Report”) (1987), “Our Common Future”.

• 7. 

  ONGOING ISSUES IN

EU ECONOMIC SURVEILLANCE

Summary

Article 99 of the EC Treaty instructs the Council to monitor economic developments in the Member States and in the Community on the basis of reports submitted by the Commission. In the context of this economic surveillance, DG ECFIN prepares a number background studies that are relevant for a broader audience; the aim of this chapter on current issues in economic surveillance is to present the results of this analysis in an easily accessible format.

 The introduction to this year’s chapter offers a definition of economic governance, a term that has assumed a growing importance in recent debates on economic policy in the EU, but which is difficult to pin down conceptually. The main body of the chapter deals with policy issues that are relevant to economic governance and surveillance. It gives a broad overview of the Draft Treaty Establishing a Constitution for Europe and discusses ways in which it strengthens economic governance in the EU. The discussion then moves onto education and its significant but changing impact on economic growth in the coming decades, as the educational profile of the workforce evolves. The chapter concludes with a discussion of structural indicators and macroeconomic statistics in the EU, both of which are critical for the effectiveness and transparency of economic surveillance.

TABLE OF CONTENTS

• 1. 

  I NTRODUCTION ............................................................................................................................................256

• 2. 

  T HE D RAFT T REATY E STABLISHING A C ONSTITUTION FOR E UROPE : STRENGTHENING ECONOMIC

GOVERNANCE IN THE EU .............................................................................................................................258

2.1 General implications of the constitutional treaty ................................................................................................ 258

2.2 Specific implications of the constitutional treaty for economic governance ....................................................... 260

2.3 The Commission’s proposals in the area of economic governance ..................................................................... 263

2.4 Conclusion .......................................................................................................................................................... 264

• 3. 

  E DUCATION AND GROWTH ...........................................................................................................................264

3.1 Constant enrolment ............................................................................................................................................. 264

3.2 Increased enrolment ............................................................................................................................................ 266

3.3 The impact of increased enrolment on average attainment ................................................................................. 269

3.4 Conclusions: the possible impact of increased attainment on growth ................................................................. 272

• 4. 

  D EVELOPMENT OF THE STRUCTURAL INDICATORS .......................................................................................274

4.1 Background ......................................................................................................................................................... 274

4.2 Principles and evolution of the structural indicators database ............................................................................ 275

4.3 Role of the structural indicators in the Lisbon Strategy ...................................................................................... 277

4.4 Critical assessment ............................................................................................................................................. 278

4.5 Conclusion .......................................................................................................................................................... 281

• 5. 

  I MPROVEMENTS IN E UROPEAN ECONOMIC STATISTICS UNDER EMU ...........................................................281

5.1 The need for improved statistics under EMU ..................................................................................................... 282

5.2 Attributes of high quality statistics ..................................................................................................................... 282

5.3 Priority areas for improvement, as identified in 1998 ......................................................................................... 282

5.4 Improvements made in response to the EMU action plan ................................................................................... 283

5.5 Beyond the action plan: the response to the Barcelona European Council ......................................................... 285

5.6 Principal European economic indicators ............................................................................................................. 285

5.7 Code of best practice on budgetary statistics ...................................................................................................... 285

5.8 Progress on implementation of the action plan for the new Member States ....................................................... 286

5.9 Conclusions ........................................................................................................................................................ 286

R EFERENCES ..........................................................................................................................................................290

ONGOING ISSUES IN EU ECONOMIC

SURVEILLANCE

• 1. 

  Introduction The use of the term economic governance in the

European Convention rather than more established terms

In the course of its deliberations on Europe's like economic policy coordination, economic institutional and political framework, the European cooperation and gouvernement économique is Convention set up a working group on economic noteworthy.

382 On the one hand, economic governance governance to discuss the case for further cooperation in has much the same meaning as these others terms. At its

the economic and financial field following the launch of base, economic governance concerns the institutions, the euro in January 1999. The working group rules and procedures that govern the conduct of distinguished between three broad strands of economic economic policies in the EU in the light of the increased governance in the EU. The first is monetary policy, interdependence that comes from the completion of the which in the euro area is the exclusive competence of internal market, the exercise of common policies and the the European Central Bank (ECB). The second strand launch of the euro. concerns the narrowly defined rules that Article 104 of On the other hand, the concept of economic governance the Treaty and the Stability and Growth Pact impose on goes beyond terms like economic policy coordination, net government lending and public debt in EU Member economic cooperation and gouvernement économique by States. The third strand focuses on the wider economic recognising the need for accountability, transparency policy mix including the pursuit of sound public and responsibility in relation to the conduct of economic finances and supply-side reform in factor and product policies in the EU. In this sense, the idea of economic markets. Although Member States retain primary governance reflects the broader debate on EU responsibility for such measures, Article 99 calls on the governance that followed the publication of the Council of Ministers to issue Broad Economic Policy European Commission’s White Paper on this subject in Guidelines (BEPGs) concerning Member States’ 2001. 383

economic policies with a view to achieving, inter alia,

sustainable and non-inflationary growth and the smooth In this White Paper, the European Commission functioning of Economic and Monetary Union (EMU). recognised the need to bring policy making in the EU

‘closer to the European citizens’ and to address the

In March 2000 the European Council introduced an perceived democratic deficit in EU institutions. This ambitious agenda for economic reform – known as the principle has been formally recognised under Article 49 Lisbon Strategy – that is designed to make the EU ‘the of the Treaty Establishing a Constitution for Europe, most competitive and dynamic knowledge-based which calls on the EU’s institutions to ‘promote good economy in the world, capable of sustainable economic governance’.

growth with more and better jobs and greater social

cohesion’ by 2010. At the same time, the European When the idea of good governance is applied to the Council recognised the central role of the BEPGs as an economic domain, it expresses two key ideas. The first instrument of economic governance. In this regard it is that there should be transparency over Member States’

called for the BEPGs to ‘focus increasingly on the medium- and long-term implications of structural

policies and on reforms aimed at promoting economic 382 For a discussion of economic policy coordination and growth potential, employment and social cohesion, as cooperation see Mooslechner and Schürez (2001). See

well as on the transition towards a knowledge-based Dyson (2000) for a discussion of gouvernement

economy’. économique. 383

  See European Commission (2001a).

rights and responsibilities concerning economic policies. meetings and allows members to nominate a Eurogroup From this perspective, the usage of the term economic President for a period of 2.5 years. Third, the governance is designed to allay concerns that EMU will impartiality of multilateral surveillance has been lead to a progressive transfer of competence in the field strengthened by giving the Commission the right to of economic policy to the EU level. It does so by issue direct warnings against Member States whose reinforcing the institutional asymmetry of EMU, economic policies are either inconsistent with the according to which euro area monetary policy is BEPGs, or otherwise risk jeopardising the proper formulated by the European Central Bank while primary functioning of EMU. responsibility for economic policies – including fiscal

policy and supply-side reform – rests with the Member The importance of education for economic surveillance

States. in the EU was recognised at the Lisbon European Council in March 2000 when the Heads of State and

The second aspect of good economic governance Government adopted the goal of halving the number of concerns the need to ensure that economic policy in the 18 to 24 year olds with only lower-secondary level EU is conducted in an accountable manner. The concept education who are not in further education and training of accountability here has a two-fold meaning. On the by 2010. Existing studies suggest that education has one hand, it states that EU institutions should be held been a key driver of economic growth in the past, but accountable for their involvement in the formulation and that its impact will change as the educational profile of implementation of economic policies. On the other hand, the work force evolves. Section 3 presents projections of it refers to the fact that Member States should be held educational attainment – by which is meant average accountable to one another. This follows from Article 99 effective years of schooling – for EU Member States which states that Member States should treat economic over the next fifty years. In so doing it builds on the policy ‘as a matter of common concern’. study of ‘Education, training and growth’ in the EU

This chapter, which is a new addition to the EU E CONOMY R EVIEW 2003. The results of this update are E CONOMY R EVIEW , discusses some special topics that the average years of schooling will increase by related to the surveillance of economic policies in the around 0.6 years in the coming decade as compared with EU. In view of the description above, this discussion a rate of 0.8 over the past eight years. It follows that will touch upon matters relating to economic governance while educational attainment will continue to be a driver

and the wider economic policy mix. of economic growth in the EU, the magnitude of this impact will gradually fall. Moreover, there is likely to be

This inaugural discussion focuses on four topics that are considerable variation between Member States owing to of direct relevance to the EU economy and which reflect the variance in the scope for increased educational the broad scope of economic surveillance. The second attainment and also the different estimated rates of section provides a broad brush overview of the new return. From an economic surveillance perspective, this constitutional treaty and considers its likely impact on changing impact of education on economic growth economic governance in the EU. The third section turns underlines the importance of ensuring quality education to the EU labour market and examines the impact of at both the secondary and tertiary levels, for example by education on economic growth. The fourth and fifth targeting reduced drop out rates from universities. sections focus on the informational requirements of

economic surveillance in the EU. The former At the Lisbon European Council in Lisbon in March concentrates on the development of structural indicators 2000, the Heads of State and Government invited the in the context of the Lisbon Strategy and their Commission to draw up an annual synthesis report on contribution to reform efforts. The latter looks at the the basis of structural indicators. Since this time, the progress achieved in relation to macroeconomic Commission’s database on structural indicators has statistics since the launch of EMU and identifies future grown to 117 indicators, covering five main areas:

challenges. General Economic Background, Employment, Innovation and Research, Economic Reform, Social

The second section of this chapter looks at the Cohesion and Environment. In October 2003, the implications of the Treaty Establishing a Constitution Commission proposed a short-list of 14 indicators, for Europe for economic governance in the EU. In which in keeping with the principle of streamlining will addition to introducing greater transparency, be revised every three years or in the intervening years accountability and democracy in EU policy making in to take new policy priorities into account only. Section 4 general, the Constitution strengthens economic of this chapter argues that the structural indicators are a governance in a number of key respects. First, there will useful tool for monitoring structural reforms undertaken be greater scope for the Council of Ministers for by the Members States and an important guarantee of Economic and Financial Affairs (Ecofin) to take consistency between different policy messages. The decisions relating to the sole votes of the euro area, for contribution of the structural indicators to economic instance in relation to the part of the BEPGs concerning surveillance could be strengthened, it is argued, by the euro area in general. Second, the Constitution placing greater emphasis on country rankings and by includes a Protocol on the informal Eurogroup that promoting a more in depth analysis of reform formalises the Commission’s involvement in Eurogroup implementation and its impact on economic institutional framework, Section 2.1 provides a general performance. overview of the main changes in this area compared

The production of high quality, reliable and timely with the present treaties. The relevance of these new macroeconomic statistics is, it is argued in Section 5, elements for economic governance in the EU is then essential for successful economic surveillance in the EU. briefly discussed.

If this condition does not hold, then it will jeopardise the Section 2.2 focuses on economic governance proper and conduct of monetary policy in the euro area and make it on the various enhancements proposed by the harder both to assess compliance with EMU’s budgetary constitutional treaty. It focuses in particular on measures rules and to identify common challenges for Member which strengthen and streamline decision making in the States’ economies. A report by the Monetary Committee euro area, including those which bolster the in 1998 on information requirements for EMU Commission’s role in multilateral surveillance. concluded that there were a number of deficiencies in

the EU’s macroeconomic statistics in relation to, for Section 2.3 summarises the Commission’s position example, money growth, banking, the financial markets, during the negotiations at the Convention and the the balance of payments and (most importantly) price subsequent IGC and compares the final text of the statistics. This paved the way for the launch of an EMU Constitution with the Commission’s overall stance. The action plan in 2000, which included targets for the final section concludes.

production of national data to permit the timely

compilation of reliable key statistics for the euro area 2.1 General implications of the constitutional and the EU, with at least 80 per cent coverage of treaty

Member States’ data. After four years, it is commonly agreed that euro area statistics have improved

considerably both in scope and timeliness. Key 2.1.1 The constitution becomes the Union’s single achievements in this regard include the production of foundation

improved quarterly national accounts, quarterly Europe has been built in stages and is based on different government statistics and short term business indicators. Treaties that have been concluded over time. This is one This progress notwithstanding, the evolution of of the reasons why the European construction is economic surveillance will require further improvements sometimes difficult to understand. From now on, the in relation to, inter alia, hourly working data, statistical “European Union” will replace the present “European indicators for the services economy and the balance of Communities” and the “European Union”. The three payments. “pillars” will be merged, even though special procedures

in the fields of foreign policy, security and defence are

• 2. 

  The Draft Treaty Establishing a maintained. The EU and EC Treaties, as well as all the

Constitution for Europe: strengthening treaties amending and supplementing them will be replaced by the “Treaty establishing a Constitution for

economic governance in the EU Europe”.

Coming only a few weeks after the Union’s biggest ever The Constitution integrates the Charter for Fundamental enlargement, the agreement of the Heads of State and Rights, becoming Chapter II of the Constitution,

384

  and

Government on the constitutional treaty marked a moreover clearly acknowledges the Union’s values and critical juncture for European integration. The objectives as well as the principles underlying the Constitution – which is now subject to ratification by the relationship between the Union and its Member States. It Member States – is not a revolution, but neither is it a also contains a clearer presentation of the distribution of mere consolidation of the Union’s current legal competences and a simplified set of legal instruments architecture. Above all it is designed as a means to and procedures.

define the competences of the Union, simplify its policy In legal terms, however, the Constitution remains a instruments, and improve the democratic legitimacy, treaty. Therefore, it will enter into force only after all transparency and efficiency of its working methods. Member States have ratified it, which implies popular The elaboration of a Constitution was entrusted to a consultations in some of them. It should be noted that Convention in which representatives of national any modification of the Constitution at a later stage will parliaments, the European Parliament, the national require the unanimous agreement of the Member States governments and the Commission discussed the future and, in principle, ratification by all. For some of the Union. The Intergovernmental Conference (IGC) modifications, however – for example with regard to the

th

that gave its final agreement on June 18 2004 has extension of the scope of qualified majority voting – a largely taken on board the Convention’s proposals. unanimous decision by the European Council will

This section examines the key developments in the

Constitution from the point of view of economic 384 This refers to the text of the Constitution as adopted in June governance. Since the Constitution has considerably 2004 by the IGC and which contains 4 chapters. The final

amended and improved the Union’s overall legal and version of the text will not contain any subdivisions and articles will be continuously numbered.

suffice, although the possibility will exist for just one Eurogroup in a separate Protocol, which notably national parliament to block the decision of the provides for the appointment of a President for a period European Council and thus to prevent the switch to of two and a half years. Both issues will be further qualified-majority voting and/or co-decision. discussed in Section 2.3.

As to the composition of the institutions, the IGC finally 2.1.2 A revised institutional framework decided to raise the maximum number of seats in the

The Convention made a particular effort to reform and European Parliament to 750. These seats will be clarify the EU’s institutional framework, notably as allocated to the Member States according to the regards the respective roles of the European Parliament, principle of “degressive proportionality”, with a the Council and the Commission and this achievement minimum of six and a maximum of ninety-six seats. The

has been largely confirmed by the IGC. precise number of seats attributed to each Member State will be decided before the European elections in 2009.

The Constitution recognises the different competences The IGC decided to maintain the current composition of of the Commission, including its near monopoly of the Commission – one Commissioner per Member State legislative initiative, its executive function and its role as – until 2014. From then on, the Commission will external representative of the Union, except in the field comprise a number of Commissioners corresponding to of common foreign and security policy. It extends very two thirds of the number of Member States. The substantially the scope of the co-decision procedure, members of the Commission will be chosen according to which, significantly, will henceforth be called the a system based on equal rotation among the Member legislative procedure (95 per cent of European laws will States, which had been already decided by the Nice be adopted jointly by the Parliament and the Council). Treaty.

This generalised recourse to co-decision, obviously

constitutes a significant enhancement of the The definition of qualified majority for decision-making Parliament’s involvement in the EU’s legislative in the Council proved to be one of the more vexing

process. questions that the IGC had to deal with. As proposed by the Convention, the Council will henceforth decide on

The main institutional innovation is the creation of the the basis of the double majority of the Member States post of Union Minister of Foreign Affairs, who will be and of the people. The IGC nonetheless decided to raise responsible for the representation of the Union on the the thresholds: instead of the majority of Member States international scene. This function will merge the present representing 60 per cent of the population, the IGC tasks of the High Representative for the Common decided that a qualified majority will require the support Foreign and Security Policy with those of the of 55 per cent of the Member States representing 65 per Commissioner for External Relations. The Minister of cent of the population. This definition is accompanied Foreign Affairs will thus be mandated by the Council for by two further elements. First, in order to avoid a common foreign and security policy and he (or she) will situation in which only three (large) Member States chair the External Relations Council. At the same time, could block a Council decision due to an increase in the the Minister of Foreign Affairs will be a full member of population threshold, a blocking minority needs to the Commission and as such in charge of the comprise at least four Member States. Moreover, a Commission’s responsibilities in the field of external number of Council members representing at least threerelations as well as of the coordination of the other quarters of a blocking minority, whether at the level of aspects of the Union’s external action. Member States or the level of population, can demand

The Constitution establishes the European Council as an that a vote is postponed and that discussions continue for institution, distinct from the Council. The European a reasonable time in order to reach a broader basis of Council will be chaired by a President, with limited consensus within the Council.

powers, appointed for a period of two and a half years.

On the other hand, and in contrast to what had been 2.1.3 A limited number of EU policies have been proposed by the Convention, the system of twice-yearly revised

rotation among the Member States of the presidency of As opposed to, for example, the Single European Act or the different Council formations (with the exception of the Maastricht Treaty, the Constitution does not extend the External Relations Council) will be maintained, the Union’s competences considerably. Neither does it although within a “team presidency” of three countries. modernise all the Union’s policies since the content of This system will be able to evolve in the future since it most provisions that govern these policies remains can be altered by the European Council acting by unchanged.

qualified majority. However, the Constitution significantly updates

In relation to EMU, the Constitution establishes the provisions in the field of Justice and Home Affairs, in European Central Bank as an institution, while order to facilitate and improve the establishment of the preserving its legal personality and independence vis-à- area of freedom, security and justice. In fact, the vis the other institutions and the Member States. It Community method will from now on apply to all the

moreover recognizes the important role of the areas in question. Moreover, they will fall to a large extent within the scope of qualified majority voting. architecture. In the field of economic governance, the Nevertheless, the Constitution retains or introduces introduction of the Eurogroup (and its President) some special features in these areas, namely in the area deserves to be mentioned and forms part of a general of judicial cooperation in criminal matters and in the tendency towards reinforcing the euro area’s area of police cooperation. governance.

The provisions regarding external relations have been As far as the Council is concerned, the “double ceiling” re-written, but in essence, the distinction between definition of qualified majority constitutes a major common foreign and security policy and the other improvement over the existing provisions in the Nice aspects of EU external action still determines the Treaty and will contribute towards facilitating effective respective roles of the institutions and the procedures decision-making. that apply. Nevertheless, the creation of the post of

Union Minister of Foreign Affairs, with the task of While Parliament has managed to increase its developing mutual confidence among Member States in involvement in virtually all EU policy areas, its role order to achieve a truly common and European stance in under the EMU chapter, which is characterised by strong external affairs, undoubtedly strengthens the Union’s Member State and Council involvement in the economic role in world affairs, in all areas. Moreover, the domain and sole ECB competence in the monetary field, possibility of providing additional ways for the Member remains broadly unchanged.

States to co-operate more closely in the field of defence

will underpin the credibility of the Union’s foreign 2.2 Specific implications of the constitutional policy. treaty for economic governance

Amendments were also introduced in the area of

economic governance, as further detailed in the next 2.2.1 General

section. It should be noted that unanimity is retained in In view of its mandate extended by the Laeken European the field of taxation and, partially, in the field of social Council, the Convention (and the subsequent IGC) policy and common foreign and security policy. focused on the European Union’s legal and institutional

Although “passerelles” allow a unanimous decision that

henceforth qualified majority will apply in a given area, framework (see Section 2.2), which are described in

it remains to be seen whether the existence of such parts I and IV of the Constitution. The provisions covering the different EU policy areas were mostly

clauses will be sufficient to maintain the Union’s taken over unchanged (cf. part III of the Constitution).

capacity to act. Moreover, the future development of the

Union means that account must be taken of the fact that Economic governance constitutes one of the few EU laws on own resources and the financial perspectives policy areas that was discussed in depth. Shortly after must be adopted unanimously, as must revisions of the the Convention on the future of Europe started its Constitution itself. activities, it decided that one of its eleven working

groups would be in charge of examining economic

2.1.4 A system marked by increased democracy and governance issues with a view to presenting proposals to transparency the Convention’s plenary. The group was composed of

34 Convention members and chaired by Mr Klaus

The Constitution introduces, or confirms in a Hänsch, former President of the European Parliament. fundamental text, an important number of provisions to The Hänsch group recognised the need for strengthening deliver more democratic, transparent and controllable economic policy coordination, while considering that EU institutions that are closer to the citizen. For EMU’s monetary “pillar” functioned in an appropriate example, the Constitution provides citizens with the manner and did therefore not require major revision. The right to invite the Commission to submit an appropriate report adopted in October 2002 by the group helped to proposal to the legislator, if they manage to collect one shape the Convention’s stance as regards EMU-related million signatures in a significant number of Member issues, and many of its recommendations were States. The proceedings of the Council, when exercising eventually taken up by the Convention in the its legislative function, are to be open to the public. constitutional treaty.

National parliaments are to be informed about all new initiatives from the Commission and, if one third of

them consider that a proposal does not comply with the 2.2.2 The role of the Ecofin Council

principle of subsidiarity, the Commission must review

its proposal. New provisions on participatory democracy During both the Convention and the IGC, the Ecofin and good governance have acquired constitutional Council proved to be active in trying to reach consensus

status. concerning most issues under discussion and in attempting to get its views across, first in the

Convention and later in the IGC. The Constitution was

2.1.5 Impact on economic governance frequently on the agenda for lunchtime discussions

All EU policy areas will benefit – to varying extents – between Ecofin members, while the Economic and from the strengthening of the EU’s institutional Financial Committee (EFC) was invited to carry out all

preparatory work and to achieve as broad a consensus as Council opinions on stability programmes, decisions on possible on all open items. Already in May 2002, the the existence of excessive deficits as well as Council EFC was asked to prepare a first issues paper and the recommendations with a view to bringing that situation Committee was subsequently invited to discuss to an end within a given period, should be adopted by outstanding issues on a regular basis and to elaborate the votes of euro area Member States only when they compromise proposals. As Ecofin’s formal positions relate to participating Member States. At the same time, repeatedly met with resistance from the part of Foreign the Constitution establishes a clear asymmetry in the Affairs Ministers who were formally in charge of EU’s decision-making rules, since euro area Member presenting Member States’ positions in the relevant States will continue to vote on any such decision relating committees, it gradually moved towards exercising its to countries that do not belong to the euro area. influence through national channels, although it should

be acknowledged that some countries appeared to be Euro area Member States also receive a more direct say more systematic than others in conveying commonly in decisions on future entries into the euro area. While

agreed Ecofin positions as national positions. the final decision on the abrogation of the derogation of a non-participating country will be taken by all Member 2.2.3 Progress achieved States (as is the case at present), the Council will only be able to adopt such decisions on the basis of a prior

Alongside the general provisions detailed in Section 2, recommendation adopted by the euro area Member the Constitution has managed to achieve meaningful States. The abrogation procedure itself continues to be progress in the area of economic governance. First of all, initiated on the basis of a Commission proposal.

(i) he capacity of the euro area to decide and act

autonomously has been significantly enhanced in most The Constitution includes a specific Protocol on the areas, thereby reflecting the need for close policy informal Eurogroup, which inter alia indicates that the coordination among Member States sharing the same Commission will participate in the Eurogroup as well as currency. In other areas, the Constitution has in the preparatory meetings (the Commission’s current (ii) recognised the need to strengthen the Commission’s status is less clear since the Luxembourg conclusions of involvement in multilateral surveillance. Finally, the December 1997 indicate that it is only “invited” to the Constitution has (iii) updated and simplified a meetings). The Protocol moreover specifies that the significant number of EMU-related provisions contained Eurogroup will nominate a President for a period of 2.5

in the current treaty. years. While the content of the Protocol does not introduce major changes compared to current practices,

(i) Reinforced decision-making within the euro area it marks an important and logical step towards

The present EC Treaty already excludes the voting confirming the role of the Eurogroup as a key player in rights of the Council representatives of the nonthe euro area’s decision making process by embedding it participating Member States in a number of areas, in the Union’s legal and institutional architecture.

notably when decisions are taken which solely concern (ii) Stronger powers for the Commission to monitor euro area Member States, such as the issue of euro the observance of the rules banknotes and coins, the adoption of ECB acts, the

nomination of the members of the ECB’s Executive The Constitution strengthens the Commission’s role as Board, the adoption of decisions relating to the euro’s independent “referee” in relation to economic exchange-rate policy, the imposition of sanctions under governance in several key respects. In the context of

the excessive-deficit procedure, etc. multilateral surveillance, the Commission will have the possibility to issue a “direct” (i.e. without the

The Constitution significantly extends the scope of endorsement of the Council) warning to Member States Ecofin decision-making based on the sole votes of euro whose economic policies are either inconsistent with the area Member States. For example, the part of the BEPGs BEPGs, or otherwise risk jeopardising the proper concerning the euro area in general (as opposed to the functioning of EMU (e.g. a significant budgetary different country sections dedicated to the individual deviation justifying an early warning). euro area Member States) will henceforth be adopted by

the votes of euro area Member States only. The current possibility for the Council to issue similar recommendations (on the basis of a Commission

More importantly, this approach has also been extended recommendation) remains in place. In essence, it will to a significant number of Council decisions which are thus be for the Commission to decide on a case by case applicable to all EU Member States. This evolution basis whether it issues a “direct” warning or whether it reflects the evident need for stronger economic policy prefers to involve the Council. The Constitution coordination between participating Member States, since moreover provides that such Council recommendations the euro area countries are more directly and (including early warnings under the SGP) will be significantly affected by policy deviations (such as adopted without the vote of the Member State excessive deficit situations) arising in other euro area concerned. Under the present rules, this Member State is countries. The Constitution therefore provides that indeed judge and defendant at the same time, and the Council recommendations (or early warnings) in the change of practice introduced by the Constitution will context of the multilateral surveillance framework, help to strengthen the impartiality of multilateral surveillance. In doing so, the Constitution moreover the Commission, this draft text became the object of removes the existing bias in favour of large Member intensive (and eventually successful) lobbying on behalf States, since under current arrangements the latter can of a few countries that strongly argued in favour of constitute a blocking minority more easily than the reverting to wording coming much closer to the sibylline

smaller countries because of their larger voting weight. provisions of the Maastricht Treaty. 386 The practical

Under the excessive-deficit procedure, Council consequences of this fight over words are minimal decisions on the existence of a deficit will henceforth be however since Article I-11(6) provides that the based on a Commission proposal as opposed to a mere provisions of Part III of the Constitution shall determine recommendation. This amendment facilitates the the scope of and arrangements for exercising the

adoption process since the voting threshold will be Union’s competences.

lowered and will thus be easier to reach. In addition, the While monetary policy was left largely unaltered by the Council will take its decision without the vote of the Convention and the IGC, the Convention decided to Member State concerned. amend the ECB’s current status as a sui generis

(iii) Streamlined and simplified decision-making institution (Art. 8 EC) and to include it in the list of EU

procedures institutions listed in the Constitution’s institutional title. The term “Eurosystem” moreover appears for the first

The EMU Chapter is probably one of the few parts of time in the Treaty. These changes are however not the EC Treaty which has not been amended since the expected to have material consequences for the signature of the Maastricht Treaty, notably in order to functioning of the ECB and ESCB. For its part, the ECB avoid the launching of counterproductive discussions judges that the Constitution preserves its ‘special and debates which could have endangered the features’ of independence, legal personality, and

preparations for the introduction of the euro. regulatory powers. 387

As a consequence, the EMU Chapter contains a large The possibility of introducing a more broadly defined

number of provisions that are obsolete. This applies to enabling clause, 388 which would allow for a

most of the transitional provisions (the transition from comprehensive reform of the governance of the ECB in stage II to stage III, the establishment of the ECB and the light of enlargement was raised during the IGC and winding up of the EMI, etc.) which are no longer strongly supported by the Commission. This amendment relevant now that the single currency has been formed part of the compromise package tabled by the introduced. Other provisions are no longer up to date Italian Presidency in December 2003, but was (e.g. during earlier revisions of the Treaty, the eventually not taken up in the Constitution agreed in coordination procedure has been systematically replaced June 2004. The IGC however agreed that the members by the co-decision procedure in all other parts of the of the ECB’s Executive Board should henceforth be Treaty). nominated by qualified majority and no longer

In other areas, the Constitution has simplified decisionunanimously, a useful move which, if implemented making rules by transforming the unanimity requirement earlier, would have helped to avoid protracted and into qualified majority (nomination of the members of acrimonious discussions in May 1998 over the the ECB’s Executive Board) or by establishing a nomination of the first ECB President and the duration

specific legal base providing for decision-making by of his mandate.

QMV (granting of macro-financial assistance to third countries) instead of Article 308 EC which requires unanimity.

Other EMU-related issues 14(2): “Specific provisions shall apply to those Member States which have adopted the euro.”

One of the most controversial issues during the 386 Art. I-11(3): “The Member States shall coordinate their Convention concerned the formulation in Chapter I of economic and employment policies within arrangements as

the Constitution of the EU’s competence in the area of determined by Part III, which the Union shall have

economic policy coordination. While the Convention competence to provide.”

decided to adopt a neutral wording that merely Art. I-14(1): “The Member States shall coordinate their

constitutes a factual description of the present economic policies within the Union. To this end, the

situation, 385 and which was therefore also supported by Council shall adopt measures, in particular broad guidelines for these policies. Specific provisions shall apply

to those Member States whose currency is the euro.”

387 See ECB (2004). 385 Art. I-11(3): “The Union shall have competence to promote 388 The present enabling clause (Article 10(6) ESCB/ECB)

and coordinate the economic and employment policies of already allows for limited revisions to the decision-making the Member States.” Art. I-14(1): “The Union shall adopt rules of the ECB’s Governing Council. It was used as a measures to ensure coordination of the economic policies of legal basis for amending Article 10(2) ESCB/ECB and the Member States, in particular by adopting broad introducing a 3-group “rotation” model of the voting rights guidelines for these policies. The Member States shall coin the Governing Council, once the number of euro area

ordinate their economic policies within the Union.” Art. I- Member States exceeds a certain number.

2.3 The Commission’s proposals in the area of already allows euro area Finance Ministers, the Commission and the ECB to exchange views on an

economic governance informal basis on all issues of common interest. It does In assessing the impact of the Constitution on economic not however have any decision-making powers since all governance, it is useful to recall the Commission’s formal decisions can only be taken at Council level. proposals in relation to this domain, during the Moreover, the functioning of the informal Eurogroup, Convention. The Commission issued two which operates on a more or less inter-governmental communications 389 to the Convention in which its basis, does not properly reflect the European Union’s

position on the different institutional and policy issues, governance principles, notably in terms of inter

390

including economic governance, was set out in detail. institutional interactions. The Eurogroup therefore only While the experience gained with the EMU framework constitutes a partial and temporary response to the need

established by the Maastricht Treaty is still relatively for closer policy coordination. short, the Commission nevertheless considered that The Commission therefore advocated the creation of a some lessons could be drawn and that scope for genuine Ecofin Council of the euro area, in which only improvement exists in several respects. the Ministers of the participating countries would be

represented. This step is all the more relevant following

2.3.1 Strengthening the EU dimension of economic the latest enlargement, since the euro area Ministers governance currently represent less than half (12 out of 25) of the

A key message in the Commission’s contribution to the total number of Ecofin representatives.

Convention was that the Community dimension of the

EU’s economic governance process needs to be 2.3.3 Making sure that the euro area is properly reinforced so as ensure a harmonious interaction of represented outside the EU

national economic policies. The need for further Finally, the Commission argued in the Convention that progress is particularly apparent with respect to the the euro area’s influence on the international scene BEPGs which constitute the Community’s overarching should be commensurate with its economic and instrument for economic policy coordination. In the commercial weight. This is unfortunately not the case present framework, the negotiation and adoption of the under existing arrangements, notably because the euro guidelines are largely in the hands of the Council and area is not properly represented in the relevant the Member States, since the possibility for the international institutions and fora, such as the IMF, the Commission to influence the final outcome is very G7, etc. While the parties most directly concerned limited once it has initiated the adoption process by recognise that the current situation is unsatisfactory, no tabling the draft BEPGs. In view of this fact, the meaningful progress has been achieved since the Commission made the case for basing draft BEPGs and introduction of the euro in 1999. In addition, the associated surveillance measures on Commission conclusions adopted on this issue by the Vienna proposals (as is the case in most other EU policy areas) European Council in December 1998 are not being rather than on Commission recommendations. This applied. The Commission therefore invited the would bring a greater degree of impartiality to Convention to contribute towards unlocking the present multilateral surveillance as well as taking greater stalemate. The Commission moreover considered that it account of the Community interest when preparing is institutionally well placed to be put in charge of the guidelines concerning economic policy. euro area’s external representation, as it already takes on

this role in many other prominent EU policy areas. 2.3.2 Strengthening economic governance within the

euro area 2.3.4 Assessment

Secondly, the Commission proposed to strengthen the Being represented in both the Convention and in the euro area’s decision-making capacity. The Eurogroup IGC, respectively as a member and an observer, the

Commission was in a position to actively defend its views throughout the revision process of the current

389 COM(2002)247 i of 22 May 2002 and COM(2002)728 i of 5 treaties and to influence the final text of the

December 2002. Moreover, a draft Treaty (known as Constitution, albeit with varying degrees of success. In

Penelope) was released in December 2002. This document respect of the necessary strengthening of the EU

was not endorsed by the Commission as such, but issued as

a “feasibility study” under the authority of the President and dimension of the economic governance framework, the Messrs Barnier and Vitorino. Constitution extends the use of Commission proposals

390 as opposed to recommendations only in relation to the The Economic and Social Committee (2002) also presented

a report on economic governance. As regards the European existence of excessive deficits. In spite of this fact,

Parliament, a draft report “on the development of and new several improvements in the surveillance area deserve to

prospects for the European economic union” was prepared be mentioned such as the possibility for the Commission for discussion in Parliament’s EMAC committee, but never to issue a “direct” surveillance warning, or the exclusion

made its way to the plenary since the Committee rejected of the Member State concerned from the decisionthe

report. making process on surveillance decisions.

The euro area’s capacity to decide and to act The present note extends the analysis to the individual autonomously was also significantly enhanced, since Member State level, thus allowing for an investigation

virtually all decisions relating to the euro area or to the of cross-country differences. 392 It presents a simple

participating Member States, particularly in the policy methodology for estimating years of schooling on the surveillance area, will henceforth be in the hands of the basis of Labour Force Survey data and presents

sole participating countries, as already detailed in attainment projections for 10 and 50 years ahead. Section 3 above. The Constitution stops short of

establishing a genuine Ecofin Council for the euro area, The paper restricts itself to education – including prebut it further emphasises the role and importance of the primary, lower-secondary, upper-secondary and tertiary Eurogroup in a separate Protocol, and establishes the education, but not including continuing vocational function of a Eurogroup President to be appointed for a training or workplace training. Training merits a

period of 2.5 years. separate treatment in its own right, but the main reasons for excluding it here are: first, it is difficult to compare

Finally, in respect of the euro area’s external with formal education in terms of years of schooling; representation, neither the Constitution nor the IGC secondly, since employers and individuals pay a large achieved any material progress, despite the fact that both share of the costs, the implications for public finances recognised the importance of this matter. are relatively limited; and, thirdly, the available data

allow at best a very partial coverage. 393

2.4 Conclusion The paper is structured as follows. Section 3.1

Now that a political agreement has been reached on the introduces the methodology employed and presents content of the Constitution, the text is being finalised for projections for educational attainment in 2012 and 2052 official signature in Rome on the 31 st of October 2004. under the assumption that enrolment in secondary and This will also be the starting date for the ratification tertiary education will remain constant. Section 3.2 process in all 25 Member States. relaxes this assumption and considers how enrolment

The Commission has indicated that it wholeheartedly rates might evolve in the coming years. Section 3.3 welcomes the new Constitution, which constitutes a incorporates the assumption of increasing enrolment into

significant improvement of the present treaties. This the projections of educational attainment. Section 3.4 section concludes by considering the likely effects of

conclusion also applies in the field of economic these educational attainment projections on economic

governance, since the various changes discussed above,

while recognising that many of them are relatively growth.

minor when considered individually, constitute a major

improvement when assessed on a collective basis and 3.1 Constant enrolment

pave the way for the future strengthening of economic The first step in this study is to establish a baseline governance in the EU. concerning what would happen to average years of schooling in the 25-64 population if enrolment remained

• 3. 

  Education and growth fixed at 2002 levels, given the expected demographic developments. Under constant enrolment rates, the

Education is attracting growing interest from economic current age profile of attainment largely determines any policy-makers, including at the EU level for two key future increase in attainment. In other words, in

reasons. First, the best available economic evidence countries where younger workers are much bettersuggests that rising educational attainment is an educated than older workers, average attainment will important influence on economic growth. 391 Secondly, automatically increase as older workers retire. In education accounts for a sizeable share – around 11 per countries where older workers are almost as wellcent in the EU as a whole – of public expenditure. educated as their younger counterparts, this effect will

be much smaller. This note builds on the analysis of the EU E CONOMY

2003 R EVIEW that was presented in a chapter on ‘Education, training and growth’. The chapter reviewed

the impact of education on growth, and examined the 392 The Member States that joined the European Union in May likely evolution of educational attainment. Attainment is 2004 are not covered in this paper mainly because the defined as the successful completion of a given level of available data do not extend far back enough for these education and is usually measured in effective years of countries. However, it should be straightforward, given the

schooling – the sum of the standard lengths of studies necessary information, to extend the exercise to 25 Member successfully completed. The chapter concluded that States. average years of schooling in the EU were set to rise by 393 A recent study by Coulombe et al. (2004) suggests that the a baseline rate of around 0.65 over the next ten years. causal links between investment in education and growth

may have been weakened by the use of diplomas and degrees as proxies for skills that increase productivity. These results underline the importance to the knowledge economy of promoting "competences" (functional literacy)

391 See de la Fuente and Ciccone (2002); de la Fuente (2003). and of lifelong and lifewide learning.

Table 1: Years of schooling by age group, 2002

15-24 25-34 35-44 45-54 55-64 25-64

Belgium 10.2 12.2 11.5 10.8 10.0 11.2 Denmark 9.3 13.0 12.6 12.6 12.1 12.6 Germany 9.4 12.7 12.9 12.7 12.2 12.7 Greece 10.1 11.8 11.1 10.1 9.2 10.6 Spain 9.7 11.0 9.9 8.8 7.8 9.6 France 10.0 12.0 11.1 10.5 9.8 10.9 Ireland 10.4 12.1 11.2 10.4 9.6 11.0 Italy 9.3 10.8 10.1 9.4 8.3 9.7 Netherlands 10.0 12.4 12.1 11.6 11.0 11.8 Austria 9.9 12.5 12.4 11.8 11.1 12.0 Portugal 8.5 9.2 8.3 7.9 7.5 8.3 Finland 9.8 12.8 12.6 11.8 10.7 12.0 Sweden 10.2 12.5 12.2 11.8 11.2 12.0 UK 11.7 12.5 12.3 11.8 11.2 12.1 EU-14 9.9 11.9 11.5 10.9 10.2 11.2

Note: EU-14 excludes Luxembourg.

Source: Commission services.

The approach here is, first, to use Labour Force Survey Average years of schooling are estimated by multiplying (LFS) data to estimate average years of schooling in 10- the highest level of education achieved by the standard

year age groups. These estimates are then used to project number of years it takes to reach that level. 395 The broad

the future increase in years of schooling due to classification of educational attainment in the LFS –

replacement of older workers by better-educated low, medium and high, corresponding to ISCED 396 1997

younger ones, and, secondly, due to current enrolment. levels 0-2, 3-4 and 5-6 respectively – is employed in Even assuming that enrolment rates are fixed, years of order to obtain estimates for most EU countries going

schooling will still increase when people currently back 10 years. 397 In a few cases where the data do not

enrolled receive their qualification. extend back to 1992, linear extrapolation is used.

In principle, an easier approach would be to use Table 1 shows estimated years of schooling by age available data on enrolment by age, and to infer future group, Average attainment is highest in the 25-34 age stocks of attainment from current flows of enrolment. In group and, as would be expected, declines thereafter practice, however, it is difficult to establish a clear link with age. Country differences are striking: attainment between the data on enrolment (from administrative ranges from just over 8 years in Portugal to almost 13 sources) and the data on attainment (usually from LFS years in Germany. Here a word of caution is in order,

or censuses). 394 since education systems in different countries are not

The advantage of using the LFS data is that they are fully comparable. The tables cannot take into account available on a comparable basis for all countries. The the fact that attainment is higher in some countries in disadvantage is that the estimates, particularly since they part because courses last longer, while it is debatable rely on splitting the sample up into 10-year age groups, whether the quality of outcomes increases in proportion are subject to sample error. In addition, the allowance with the length of studies. The age profile of attainment made for increased attainment due to current enrolment ranges from a steep incline in the case of Spain – where is very imprecise. Thus, estimates based on detailed attainment of 25-34 year-olds is over three years higher national sources may produce slightly different results than that of 55-64 year-olds – to almost a plateau in

and, accordingly, the results presented here should be Germany.

seen as indicative of cross-country differences rather

than precise estimates of the situation in each country. 395 The latter is obtained from de la Fuente and

Doménech (2001).

396

394 Complications include: the time lag between enrolment and International Standard Classification of Education, attainment (the latter being measured by the highest developed by the UN.

qualification achieved, not the number of years actually 397 LFS data on attainment by the finer ISCED 1997

spent in school or college); drop-outs, repeat years and partclassification are available for a few recent years only. In time studies, which mean that the average year of enrolment any case, as noted in European Commission (2003a), the results in less than a year of average attainment; and estimate of average years of schooling for 2002, at least for inconsistencies between the (administrative) enrolment data the EU as a whole, is similar using either the broader or

and estimates of attainment based on surveys or censuses. finer classification.

Based on these data, the following rough projections of Table 2: Projected years of schooling in the 25-64 average years of schooling in 2012 are made. population with constant enrolment

• For a lower bound, it is assumed that 15-24 year 2002 2012 2052 olds in 2002 will reach the same level of attainment BE 11.2 11.7 12.3

as 25-34 year-olds in 2002, and that the older

groups will remain at 2002 levels of attainment. DK 12.6 12.8 13.0

Thus, 45-54 year-olds in 2012, for example, would DE 12.7 12.8 12.7

have the same level of attainment as 35-44 year EL 10.6 11.3 12.1

olds in 2002. This misses the impact of, for ES 9.6 10.6 11.9

example, a 26 year-old who is currently enrolled for

a university degree but is yet to graduate. FR 10.9 11.4 12.1

• For an upper bound it is assumed that attainment IE 11.0 11.8 12.8 IT 9.7 10.4 11.1

will rise in each age group in the same proportion as

it did between 1992 and 2002. Thus, for example, NL 11.8 12.1 12.4

attainment of 35-44 year-olds in 2012 is estimated AT 12.0 12.6 13.1

by: attainment of 35-44 year-olds in 2002 × PT 8.3 8.8 9.4

(attainment of 45-54 year-olds in 2002 / attainment FI 12.0 12.8 13.7

of 35-44 year-olds in 1992). 398 This includes not

only the impact of current enrolment but also the SE 12.0 12.3 12.8

increase in enrolment rates between 1992 and 2002. UK 12.1 12.4 12.9

• The rough projections reported below are the mid EU-14 11.2 11.7 12.2

points between the lower- and upper-bound Note: EU-14 excludes Luxembourg. estimates. The difference between the mid-point and

the lower bound is thus, in effect, taken to reflect Source: Commission services.

the impact of current enrolment on attainment. The

long-run estimates (for 2052 and beyond) take the 3.2 Increased enrolment

2012 result for 25-34 year-olds and add to this the Of course, the assumption that enrolment patterns will estimated increase in attainment due to current remain constant is moot. Table 2 provides indicators of

enrolment for older groups. Thus the long-run

attainment profile under constant enrolment is current enrolment in upper-secondary and tertiary education. As discussed in the EU E CONOMY R EVIEW

slightly increasing with age. Table 2 presents 2003 there are a number of reasons to expect enrolment

projected years of schooling under constant to increase in the coming years.

enrolment. Population figures are taken from

Eurostat’s baseline population projections for 2010 First, average years of schooling in the 25-64 population and 2050. These results differ slightly from those have been growing at a roughly constant rate of 0.8 per

presented in the EU E CONOMY R EVIEW 2003, decade in the EU as a whole since the 1960s.

mainly because of a correction to the data for the

UK. 399 Second, policy-makers (at both EU and national levels) have established explicit targets for increased enrolment

and/or attainment at upper-secondary and tertiary levels.

398 It turns out that, for a few combinations of countries and Nevertheless, one should not necessarily expect age-groups (especially the 15-24 age group), estimated enrolment to continue to grow indefinitely. In some

attainment in 2002 is actually slightly lower than estimated

attainment in 1992, which means that the ‘upper bound’ is countries, upper-secondary education (up to the age of below the ‘lower bound’. This is puzzling given the 18) is already compulsory and therefore near-universal. available evidence on enrolment, which suggests that In tertiary education, the position of the United States

participation especially in tertiary education rose during the suggests some scope for further increases in

1990s. Moreover, educational reforms in recent years have, participation in most EU countries. Beyond this, if anything, aimed to reduce course durations and drop-out however, it is unclear whether tertiary participation will

rates. In these cases, the upper bound estimate was become saturated, or whether it can continue to grow. constrained to the lower bound one.

399 The data for the UK are corrected for a break in the series

around 1997, which appears to be due to the UK’s decision to count success in the GCSE exams (General Certificate of Secondary Education, which pupils usually sit at age 16 shortly after the end of compulsory schooling) as uppersecondary attainment. The data reported respect that

decision, but adjust the earlier part of the series accordingly. reducing the apparent 1992-2002 increase in years of This means that the increase in attainment over the decade schooling in the UK by 1.2 years and in the EU as a whole

1992-2002 in the UK is considerably less than it first by 0.17.

appears from the raw data. This has a significant impact, Table 3: Indicators of participation in upper-secondary and tertiary education

Upper-secondary (2002)

%18-24 year-olds qualified or in Tertiary (2001)

further training Enrolment as % of 20-29 population

Belgium 87.6 27.4

Denmark 91.6 27.2

Germany 87.4 21.8

Greece 83.9 30.1

Spain 71.0 28.0

France 86.6 26.0

Ireland 85.3 26.0

Italy 75.7 22.5

Netherlands 85.0 24.4

Luxembourg 83.0 4.04

Austria 90.5 26.2

Portugal 54.5 24.2

Finland 90.1 44.0

Sweden 89.6 32.5

UK 82.3 26.7

EU-15 81.5 25.4

USA n.a. 36.6

Note: Tertiary participation is low in Luxembourg because most students study abroad.

Source: Commission services.

This section focuses on upper-secondary and tertiary a similar path to that of enrolment in primary and education, since this is where most of the scope for secondary education, approaching universality in the increased enrolment lies. We do not include pre-school long run. On the other hand, some commentators have education or most of adult education and training here. raised concerns about ‘over-education’, some even Although empirical evidence suggests that early child predicting that tertiary enrolment may fall towards what

care and education has a positive impact on cognitive they consider more reasonable levels. 401 This would

abilities, this may be regarded as an influence on the suggest saturation of tertiary enrolment well below quality of learning rather than something to classify as universality. part of formal education. Clearly, adult education and

training merits a separate discussion in its own right. Another issue is how to specify the enrolment rate and However, adult education that leads to a formal in particular its denominator. Here, we express (gross) educational qualification (in the ISCED classification), tertiary enrolment as total enrolment in tertiary studies is included in the data on years of schooling and, divided by the population aged 20-29 in all countries. It

implicitly, in the benchmarks discussed below. should then be clear that, if two countries have the same tertiary enrolment rate but degree courses are longer in

Third, even if these areas were included, the impact on country A than in country B, then the share of people average years of schooling would be relatively small who graduate from tertiary education is lower in country compared to increased upper-secondary and tertiary A than in country B.

enrolment. 400 In the case of preschool education, it takes

more than 20 years for this to have any impact on Over long periods of time, enrolment rates may be average schooling in the 25-64 workforce. The impact of observed to follow an S-shaped adoption curve similar reaching the EU Education Council benchmark for to that which characterises the diffusion of many goods increased adult education and training (12.5 per cent of and services. A curve for US data is estimated (since this 25-64 year-olds participating at any given time by 2010) is the country with the highest tertiary enrolment rate for would be larger, though progress towards this target has which a sufficiently long series is available), imposing a

been slow thus far.

Projecting enrolment in tertiary education is inevitably a tentative exercise. One might imagine that it will follow

400 See EU E CONOMY 2003 R EVIEW , Chapter 3, Table 7. 401 See, for instance, Krugman (1996).

Graph 1: US enrolment in degree-granting institutions, In upper-secondary education, the natural assumption is

1900-2000 100 per cent attainment in the long term. Uppersecondary

 education is already compulsory and near

0.5 universal in some EU Member States. For simplicity, a

similar curve to that used for tertiary education is

actual assumed, with the implication that enrolment rates n 0.4 estimated follow a concave path (increasing at a decreasing rate) ul at

io

towards the long-run maximum.

op

 p 0.3

0- 29 The age range 18-24 is used for comparison with the

 2 Lisbon target to halve the number of 18-24 year-olds

n of 0.2

rt io with below upper-secondary level education who are not in further education or training by 2010. In 2000, around

P ro

po

0.1 19.4 per cent of 18-24 year-olds were in this position. 403

A liberal interpretation would be for an additional 9.7 0 per cent of 18-24 year-olds to reach upper-secondary 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 attainment by 2010 (in practice, it could be less than 9.7

per cent, since the target refers to enrolment and not all

• 1) 

  Normalised. of those enrolled will necessarily graduate). In that case,

the EU-15 enrolment rate would reach 90.3 per cent in

Notes: A logistic specification was used to capture the S-shape: 2010, which is significantly above the 87.8 per cent

y = y max shown in the table below. Thus, again, the present +

e at +

where y is the enrolment rate, y max is the maximum

1 b scenario may be regarded as slightly on the conservative

enrolment rate (in this case 0.5, or 50% of the 20-29 year-old side compared to publicly announced targets.

population), t is time in years (0 = 1900) and a and b are parameters.

This was estimated by OLS with dependent variable  The table below summarises the benchmarks for y ˆ =  ln  y max − 1 .

  y  increased enrolment in both upper-secondary and

The estimated curve was then used to project future enrolment rates tertiary education. These are not to be regarded as

given today’s enrolment rate. forecasts, but rather as ‘plausible benchmarks’, with a view to judging what might happen to economic growth

Source: Commission services. US Department of education, Census

Bureau and Commission services. and public spending on education if enrolment continued to increase. What actually happens in individual

Member States will depend on precise policy measures.

maximum enrolment rate of 50 per cent (compared to an For example, a country might see a sharper increase in

402

actual rate of 38.6 per cent in 2000). The estimated upper-secondary enrolment if it decided to make upperrelationship between enrolment rate and time (shown in secondary education compulsory; or tertiary enrolment Graph 1) is then used to project tertiary enrolment in the might rise by much less if spending on higher education EU, given data on enrolment rates in 2001. was capped.

This implies that tertiary enrolment for the EU as a whole would rise from 25.4 per cent of the 20-29 population in 2000 to 30.7 per cent in 2010. It could be argued that, given widespread recognition of the importance of tertiary education for the knowledgebased economy, and the presence of national-level targets for more substantial increases in enrolment in a number of Member States, a more ambitious benchmark for 2010 would be appropriate. If such targets are considered plausible, then the estimates of increased attainment presented here should be regarded as conservative.

402 A 50 per cent rate would imply that, if only 20-29 year-olds

attended university and degrees lasted for 5 years, then in the long run 100 per cent of the population aged 30 and above would graduate (assuming no drop-outs, repeat years or part-time studies). In practice, since some under-20s and

over-29s are enrolled, and many people study for longer 403 By 2003, this had fallen to an estimated 18.0 per cent. See

than 5 years, the long-run share of graduates will be well Eurostat Structural Indicators (described in this chapter’s

below 100 per cent. Section 4).

Table 4: Benchmarks for increased enrolment, 2010 and 2050

Tertiary level Upper-secondary level Enrolment as a % of the population aged % of 18-24 year-olds with upper 20-29 secondary level or in further studies 2000 2010 2030 2050 2000 2010 2030 2050

Belgium 27.1 32.8 41.6 46.4 87.5 91.9 96.7 98.7 Denmark 26.6 32.4 41.4 46.3 88.4 92.5 97.0 98.8 Germany 21.5 27.4 38.0 44.6 85.1 90.2 96.0 98.4 Greece 26.6 32.4 41.4 46.3 82.9 88.7 95.3 98.2 Spain 27.7 33.4 42.0 46.6 71.2 80.0 91.2 96.5 France 25.7 31.5 40.8 46.0 86.7 91.3 96.5 98.6 Ireland 24.9 30.8 40.3 45.8 83.6 89.2 95.6 98.3 Italy 22.2 28.2 38.6 44.9 74.7 82.7 92.6 97.0 Lux. 4.2 6.5 14.0 25.2 83.2 88.9 95.4 98.2 Netherlands 23.4 29.4 39.4 45.3 84.5 89.8 95.8 98.4 Austria 25.3 31.1 40.6 45.9 89.8 93.4 97.4 99.0 Portugal 23.2 29.1 39.2 45.2 57.1 68.3 84.9 93.6 Finland 42.7 45.2 48.0 49.2 91.1 94.3 97.7 99.1 Sweden 31.3 36.5 43.8 47.4 92.3 95.1 98.1 99.2 UK 26.2 32.0 41.1 46.2 81.7 87.8 95.0 98.0 EU-15 24.9 30.7 40.2 45.7 80.6 87.8 94.8 97.9

Note: Tertiary enrolment projections based on convergence to a long-run maximum of 50% along a curve estimated on US data; uppersecondary projections based on convergence to 100% along a similar logistic curve. Source: Commission services.

3.3 The impact of increased enrolment on a year of attainment, which is the variable of interest for present purposes.)

average attainment If enrolment is growing, then a relatively large share of

In order to determine the potential impact of increased students is in the earlier years of study. In this case, l in

enrolment on economic growth the benchmarks must be the above formula may be replaced by:

expressed in terms of increased years of schooling. This

is straightforward in the case of upper-secondary l − 1

education, since the chosen benchmark is already in ∑ ( 1 + g ) i ,

terms of attainment. In the case of tertiary education, i = 1 allowance must be made for high drop-out rates and

study durations beyond (or below) the standard length, where g is the (constant) annual rate of growth of

which mean that years of enrolment are significantly enrolment.

higher than years of attainment in some cases. Applying this formula to the 2001 Eurostat figures on

The available data suggest that drop-out rates in tertiary enrolment and graduates in tertiary education as a education vary considerably among countries and in whole, and taking de la Fuente and Doménech’s figures some cases are very high. Figures reported in the on duration of full-length tertiary courses for l, the OECD’s Education at a Glance publication, for following results for d can be obtained (Figures from the

instance, suggest that in a couple of countries more than previous three years are used to estimate g).

405

half of those who begin a tertiary programme fail to

graduate. 404

405

If enrolment were constant over time and all students It must be stressed that d here is not the same as the OECD

remained enrolled for the same number of years, the indicator, which compares the number of graduates to the

relationship between the number of graduates and the number of entrants in the typical year of entry. A high level of d may result not only from drop-outs in this sense, but

number enrolled in a given year could be expressed as: also from studies lasting longer than the standard number of

years. For example, if a student takes 6 years to complete a

( degree, but the standard length – for the purpose of G = 1

l E 1

−

t t d

) , estimating years of schooling – is 4 years, then this

represents a ‘drop-out’ of 2 years out of 6, or 33 per cent.

where G The negative estimates of d in some countries may be partly t is the number of graduates in year t, l is the length of the course, E explained by average degree courses being shorter than the t is the number of students enrolled in year t and d is the “drop-out rate” in years standard lengths (taken from de la Fuente and Doménech,

(i.e. the share of years of enrolment that do not result in 2003) used to estimate years of schooling. It should also be noted that, although these figures are from the same

statistical source, there may still be inconsistencies between numbers enrolled and numbers graduating, as well as cross

404 OECD (2003), p. 52. country differences in data collection and so forth.

“Drop-out rates” for tertiary education thus inferred are then multiplied by the increase in enrolment to give increased attainment in tertiary education.

Table 5: Tertiary duration and “drop-outs”

d years

Belgium 20% 4 Denmark 17% 4 Germany n/a n/a Greece 14% 4 Spain 10% 5 France -7% 4 Ireland -10% 4 Italy 33% 5 Netherlands -2% 5 Austria 57% 4 Portugal 35% 4 Finland 30% 5 Sweden 50% 4 UK -21% 4 EU average 14% 4.4

Note: EU average is weighted by enrolment.

Source: Commission services.

The estimated increases in attainment in each age group must then be translated into increases in average attainment in the 25-64 labour force. For the purpose of this calculation, we assume that attainment or enrolment rates rise gradually along the path assumed in the S- shaped projections.

We now turn to the main results, presented in the

following table. 406

These results suggest that average educational attainment among 25-64 year-olds is set to continue increasing in the EU as a whole, though at a declining rate compared to previous decades. The intuition that increased average attainment over the next 10 years is dominated by the replacement of older workers with better-educated younger counterparts is confirmed. The impact of further increases in upper-secondary and tertiary enrolment on average attainment among 25-64 year-olds is limited in the first decade (also partly because most of those in the relevant age groups are below the age of 25). In the longer term, however, the potential for further increases in average educational attainment clearly depends on increasing enrolment, especially in tertiary education.

406 The results for the EU as a whole differ slightly from those

presented in the 2003 EU E CONOMY R EVIEW on account of the correction to the UK data (described above) and the fact that the benchmarks for increased enrolment here are less ambitious. The estimated increase in attainment with constant enrolment between 2002 and 2012 or 2052 derived in section 3.1 is taken to be equal to the increase between 2000 and 2010 or 2050.

Table 6: Projected effective years of schooling in the 25-64 population

increase with increase due to increase due to total attainment (increase since constant enrolment upper-secondary tertiary 2000)

2000 2010 2050 2010 2050 2010 2050 2010 2050

Belgium 11.1 0.5 1.1 0.01 0.26 0.06 1.06 11.6 (0.6) 13.5 (2.4) Denmark 12.5 0.2 0.4 0.02 0.32 0.06 1.12 12.9 (0.3) 14.4 (1.9) Germany 12.6 0.1 0.1 0.02 0.30 0.04 0.87 12.8 (0.2) 13.9 (1.3)

Greece 10.5 0.8 1.5 0.02 0.35 0.07 1.16 11.3 (0.9) 13.5 (3.0) Spain 9.4 1.0 2.2 0.04 0.74 0.07 1.16 10.5 (1.1) 13.5 (4.1) France 10.8 0.5 1.2 0.02 0.27 0.08 1.49 11.4 (0.6) 13.7 (2.9)

Ireland 10.7 0.8 1.8 0.02 0.34 0.11 1.57 11.7 (0.9) 14.4 (3.7) Italy 9.8 0.6 1.3 0.04 0.82 0.05 0.98 10.5 (0.7) 12.9 (3.1) Luxembourg 0.02 0.85

Netherlands 11.7 0.3 0.6 0.01 0.21 0.07 1.49 12.1 (0.4) 14.1 (2.3) Austria 11.9 0.6 1.1 0.02 0.28 0.03 0.59 12.5 (0.6) 13.9 (2.0)

Portugal 8.3 0.5 1.1 0.05 1.05 0.05 0.97 8.8 (0.6) 11.4 (3.1) Finland 11.9 0.8 1.7 0.01 0.19 0.02 0.34 12.8 (0.8) 14.2 (2.3) Sweden 11.9 0.4 0.8 0.01 0.16 0.03 0.57 12.3 (0.4) 13.4 (1.5) UK 12.0 0.4 0.8 0.02 0.37 0.09 1.64 12.5 (0.5) 14.8 (2.9)

EU-15 11.1 0.5 1.1 0.02 0.43 0.06 1.19 11.7 (0.6) 13.8 (2.7)

Source: Commission services.

Cross-country differences are striking. Over the next As regards the different areas of the education system, decade, the projected increase in average attainment in the results suggest that increased upper-secondary Germany is less than one fifth what it is in Spain. It may enrolment may still have a significant contribution to be worth recalling the main reasons for these make to raising average educational attainment. In most differences: cases, however, the potential contribution of tertiary

• Cohort effects: If a country has experienced a rapid education far outweighs that of upper-secondary, with

increase in enrolment in recent decades, so that the notable exception of Portugal.

young people’s attainment is much higher than that

of older working-age people, then the 3.4 Conclusions: the possible impact of increased

predetermined increase in average attainment is attainment on growth

correspondingly high. This is mostly the case in If the findings of recent research on the link between countries where attainment is relatively low (though education and growth were taken at face value, then the the same is true for Finland where, despite high results presented here would have significant average attainment, enrolment of young people has implications for growth potential in the EU-15 as a increased rapidly). whole and for cross-country differences. If one extra

• Scope for further increases in enrolment: The year of schooling in the labour force aged 25-64 leads to methodology for the projections assumes that an increase in GDP of around 6 per cent and if the countries converge to long-run (i.e. beyond 2050) assumptions behind the attainment projections hold, then

maximum enrolment rates, so that those with the main results could be summarised as follows.

relatively low rates to begin with have greater scope In the EU as a whole the contribution of education to for further increases. growth looks set to decline. The projections suggest that

• Length of upper-secondary studies: The benchmarks average years of schooling will increase by around 0.6 here refer to the number of people completing years in the coming decade, compared to 0.8 per decade upper-secondary education (long-run maximum of over the past 40 years. This implies that the contribution 100 per cent), so the impact on effective years of of education to rising GDP in the EU as a whole would schooling is higher in countries where the standard fall from almost 0.5 percentage points of GDP per year length of upper-secondary studies is longer. 407 The in recent decades to 0.35 percentage points up to 2010, standard length is 3 or 4 years in most countries, and falling slightly further thereafter. This varies a great

except Italy (5 years) and Netherlands (2 years). deal between countries, owing mainly to variance in the scope for increased attainment, but also to different

• Enrolment/attainment ratio in tertiary education: In estimated rates of return. The following table sums up

tertiary education, the benchmark refers to the share the implications for growth, using the de la Fuente of people enrolled. This implies a trade-off between (2003) estimates of raw macroeconomic returns to

the length of studies and the number of graduates. schooling in individual EU countries. The effect of increased enrolment on effective years of schooling is lower in countries where the number of years enrolled is significantly higher than the standard length of studies needed to achieve a degree (owing to a high drop-out rate or repeat years, for example). This effect also partly explains the relatively large impact of increased tertiary enrolment in France, Ireland, the Netherlands and the UK, where the available data indicate that the average time taken to successfully complete studies is shorter than the standard length given in de la Fuente and Doménech (2001).

The pure demographic effect of a falling share of young people in the population has a small impact for the EU as a whole during this period. For example, the increase in average attainment due to higher tertiary enrolment by 2050 would be 1.22 years instead of 1.19 years if, all other things being equal, the structure of the population remained as it was in 2000. The demographic effect makes very little difference to cross-country comparisons.

407 The standard length of upper-secondary studies is taken

from de la Fuente and Doménech (2001).

Table 7: Possible impact of increased attainment on GDP

Projected increase in attainment Macro return

(years) Implied annual % increase in GDP

2010 2050 (1990 data) 2010 2050

Belgium 0.57 2.40 5.82 0.33 0.26 Denmark 0.32 1.88 5.00 0.16 0.18 Germany 0.19 1.27 4.53 0.08 0.11 Greece 0.86 2.98 7.42 0.62 0.40 Spain 1.07 4.14 8.27 0.85 0.59 France 0.60 2.93 5.62 0.33 0.31 Ireland 0.92 3.69 6.24 0.56 0.42 Italy 0.71 3.13 7.30 0.51 0.41 Netherlands 0.38 2.32 5.36 0.20 0.24 Austria 0.64 1.98 5.19 0.33 0.20 Portugal 0.57 3.12 9.16 0.51 0.50 Finland 0.83 2.26 5.35 0.44 0.23 Sweden 0.41 1.53 5.53 0.22 0.16 UK 0.49 2.85 5.58 0.27 0.30 EU 0.58 2.68 6.17 0.35 0.31

Note: Implied annual increase in GDP is calculated as the compound annual growth rate required to yield the implied increase in the level of GDP by 2010 or 2050.

Source: Commission services and de la Fuente (2003) for macro returns.

There is a negative correlation between the projected Since the ‘ifs’ mentioned at the beginning of this section increase in attainment and GDP per capita, which is are big ones, it is instructive to recap the essential consistent with the presumed role of education in caveats which suggest that these results should be economic catch-up. This is not surprising since current interpreted with caution, and in any event as projections attainment is clearly linked to GDP (with causality based on strong assumptions rather than forecasts: almost certainly running in both directions), while longterm

upper bounds on average attainment are imposed in • The future impact of education on growth depends

the projections. Apparent outliers include Ireland (a on quality and efficiency, as well as quantity. Evidence greater projected increase in average attainment than suggests that an improvement in the quality of school expected given current GDP), Greece and Portugal (both education of one standard deviation could in fact have a with a lower projected increase in average attainment larger effect than an extra year of schooling.

408

  At

than would be expected on the basis of cross-country tertiary level, effective years of schooling could be

differences). increased without even raising enrolment by reducing the number of drop-outs and excess years of study. 409

Graph 2: Projected increase in attainment versus current • The absolute level of average attainment (as GDP per capita opposed to changes in average attainment) may be 150 relevant to growth, perhaps especially when it

) 140 comes to extending the frontier of technical

00

 1 130 progress. In that case, countries like Germany may

5 = still enjoy advantages. -1 120 AT DK NL IE • On the other hand, there are some reasons to fear

 ( E

U 110

P S BE

UK

FR that a macroeconomic return of around 6 per cent – P 100 DE SE IT

 in FI i.e. an extra year of schooling raises aggregate

ita 90

ap productivity by 6 per cent – may be optimistic for c 80 ES PT the future. These include the possibility of slower

P p

er

70 technical progress, and the possibility of

G D 60 EL diminishing returns to further increases in tertiary

50 enrolment.

1 1.5 2 2.5 3 3.5 4 4.5 Projected increase in attainment to 2050 (years)

Source: Commission services. 408 See Hanushek and Kimko (2000).

409 For a discussion of the importance of quality and an

introduction to the non-monetary outcomes of investing in education, including the issue of equity, see OECD (2002).

• The difference between estimated rates of return in 4. Development of the structural indicators

different countries depends on the assumed form of the aggregate production function – in the case of de 4.1 Background la Fuente’s estimates, returns are in fact constrained

to diminish as years of schooling increase, so that As requested by the Lisbon European Council, the countries with high current attainment have assessment of progress towards the Lisbon objectives in relatively low returns by assumption. the annual Spring Report is based on a list of structural • The attainment projections for individual Member indicators to be agreed between the Commission and

States should be seen as tentative, given the Council.

411 Since the first proposal by the Commission

underlying data and methodology, and because of in 2000, the indicators database has evolved the inevitable degree of arbitrariness in establishing considerably. In order to make it easier to present the benchmarks for increased enrolment. policy messages and the Member States’ positions

• The projection methodology effectively sets an relative to the key Lisbon targets, the Commission

412

upper bound on tertiary enrolment that may not proposed a shortlist of 14 headline indicators in October strictly apply in practice. Some countries (Finland 2003.

413 The comprehensive database (117 indicators),

in particular) have increased enrolment by more which covers five main areas: General Economic than might be expected according to this Background, Employment, Innovation and Research, methodology. Adult education and training could Economic Reform, Social Cohesion and Environment, also contribute to raising effective attainment. continues to be released on the publicly accessible

• In the shorter term, the projections might be seen as Eurostat structural indicators website.

414

slightly conservative compared to publicly This section aims at presenting the structural indicators announced targets for increased upper-secondary (SI) as an instrument of economic surveillance and, in some countries, tertiary participation. contributing to the assessment of progress of EU

Despite all these caveats, the basic results have a ring of countries towards the Lisbon objectives. The remainder truth about them. There is quite strong evidence that the of this section is organized as follows. The principles change in educational attainment over time is important and evolution of the SI database is first reviewed. The for growth, and the scope for further increases in most recent development is the creation of the shortlist average years of schooling clearly varies a good deal of structural indicators in 2003. A robustness analysis of among countries. Nevertheless, some of the caveats may the progress assessments based on both the shortlist and

be important in the context of education and training the comprehensive database is also included.

415

  The third part is dedicated to the role of the SI in the Lisbon policies: for example, greater attention to quality and

efficiency may be required in some countries, or the strategy. Finally, the current use of the SI is critically design of policies and reforms may need to take into assessed from different angles: the method used to select

account a country’s position relative to the forefront of the indicators; the relevance of the list; the effectiveness of the indicators as part of the governance system; and

technical progress. 410 the use of indicators for country ranking. As far as the

latter is concerned, possible methods for constructing rankings are briefly discussed as well.

411

  The Presidency conclusions of the Lisbon European Council (23-24 March 2000) indeed states (paragraph 36): “The European Council invites the Commission to draw up an annual synthesis report progress on the basis of structural indicators to be agreed relating to employment, innovation, economic reform and social cohesion.” The sustainable development objectives have been added at the Gothenburg Council in June 2001.

412 European Commission (2003). 413 This shortlist was slightly amended following discussions

with the Council (2003a).

414

  http://europa.eu.int/comm/eurostat/structuralindicators.

415

  The robustness analysis is conducted in collaboration with

410 See Aghion and Cohen (2004). the JRC of Ispra.

Box 1: The shortlist of structural indicators

The list of indicators is balanced to reflect the importance that European Councils at Lisbon and Gothenburg placed on the domains of employment, innovation and research, economic reform, social cohesion and the environment.

• 1. 

  GDP per capita in PPS (General Economic Background) 2. Labour productivity (General Economic Background) 3. Employment rate* (Employment) 4. Employment rate of older workers* (Employment) 5. Educational attainment (20-24)* (Innovation and Research ) 6. Research and Development expenditure (Innovation and Research) 7. Comparative price levels (Economic Reform) 8. Business investment (Economic Reform) 9. At risk-of-poverty rate* (Social Cohesion) 10. Long-term unemployment rate* (Social Cohesion) 11. Dispersion of regional employment rates* (Social Cohesion) 12. Greenhouse gas emissions (Environment) 13. Energy intensity of the economy (Environment) 14. Volume of freight transport (Environment)

• * 

  Indicators disaggregated by gender.

4.2 Principles and evolution of the structural Over the years, the number of indicators has tended to indicators database increase thus making it more difficult to draw a clear

picture of progress towards the Lisbon objectives. In

The choice of indicators reflects the overall objective of order to re-focus the policy debate, the Commission the Lisbon strategy, which is for the EU to become: “the proposed in 2003 (COM(2003b) 585 final) a shortlist of most competitive and dynamic knowledge-based only 14 SI, in combination with a publicly-accessible economy in the world capable of sustainable economic database and website containing the longer list. The final growth with more and better jobs and greater social shortlist approved by the European Council in December cohesion”. This overall objective can be represented by 2003 settles the current framework for the SI analysis. In a welfare function reflecting the idea that aggregate accordance with the principle of streamlining documents welfare in the EU depends on economic, environmental and policies, it was agreed that this shortlist would be and social factors. The different structural indicators revised every three years only, although it could be selected reflect policy action or performance in these modified in intermediate years in order to take new

three domains. policy priorities into account.

In addition, the structural indicators in the list have to The shortlist of indicators has several advantages. First, fulfil a number of more specific selection criteria. The it makes it easier to present a clear picture of the indicators have to be: (1) easy to read and understand; Member States’ relative positions with respect to the (2) policy relevant; (3) mutually consistent; (4) available most important Lisbon targets. Second, the shortlist

in a timely fashion; (5) available for most, if not all includes well-known and easy-to-understand indicators. Member States; (6) comparable between these countries, Third, the shortlist of indicators has a better logic, thus as far as possible with other countries; (7) selected from reinforcing the economic foundations of the policy reliable sources; and (8) do not impose too large a messages drawn from the progress assessment. Fourth, burden on statistical institutes and respondents agreeing the list of indicators every three years fits with In practice, the development of the Structural Indicators the streamlined procedure for the Broad Economic database (and shortlist) is the result of interactions Policy Guidelines, the Employment Guidelines and the between the Commission and the Council. The Internal Market Strategy. Hence, the stability of the Commission proposes a set of indicators which is shortlist is of crucial importance to make comparisons agreed, in turn, by the Council. In the past, the selection over time possible. As structural issues develop only method relied on compromises that expanded the slowly over time and as several of the indicators are key database often at the expense of the underlying Lisbon targets, it is wise not to revise the list too

economic rationale. 416 frequently.

416

  The first Commission (2000) Communication following the Lisbon Council, put forward a list 27 key indicators, which became 35 after agreement with the Council. The Stockholm response to this request, the Commission proposed a new Council in March 2001 and the Gothenburg Council in June list of 36 indicators (European Commission, 2001b), which 2001 called for new indicators in the fields of social became 42 after approval by the Council (expanding to 107 inclusion and sustainable development, respectively. In indicators when all sub-indicators are taken into account). Box 2: Consistency analysis between the shortlist and structural indicators database

This analysis demonstrates that inferences based on the shortlist of 14 headline indicators corroborate assessment that is based on

the comprehensive database. This analysis of robustness is done in terms of average country rankings across the indicators. 1 The

study is carried out both for levels and growth rates. Levels are analysed for the years 1999, 2000, 2001, 2002 and 2003. Growth rates are considered over the period 1999-2003. Two statistical tests are carried out to establish whether the average country rankings obtained with the shortlist are statistically equivalent to the average country rankings obtained with the full set. For more detail see Tarantola, Liska and Saltelli (2004).

Based on the average rankings for country i , obtained from the short list (denoted by X i ) and full list (denoted by Y i ), the following

F-test is performed: a linear relationship Y i = a + b X i + e t between X i and Y i is assumed and the hypothesis H 0 : a=0, b =1 is tested by F-statistics, which under the null hypothesis follows an F 2,n-2 distribution. The F-statistics are presented in the table below together with their p-values and information whether the hypothesis H 0 can be rejected or accepted at 5 per cent level. The graphs

illustrate the result for both the level in 2003 and the average growth 1999-2003. 2

Table: Consistency analysis between the headline indicators and full set of indicators

Indicators y = a + b x H_0 that a = 0 and b = 1

for full set Coefficients (a, b) R 2 Fstat p-value H_0 at 5% level 1999 68 1.83 + 0.77 x 0.83 3.11 0.082 accepted level 2000 70 2.15 + 0.73 x 0.88 6.35 0.012 rejected level 2001 74 1.54 + 0.81 x 0.89 2.98 0.086 accepted level 2002 78 1.63 + 0.80 x 0.85 2.49 0.121 accepted level 2003 78 1.72 + 0.79 x 0.84 2.58 0.114 accepted average growth 1999-2003 77 5.29 + 0.29 x 0.36 19.7 <0.001 rejected

Note: Critical value for the level estimates is F(0.05, 2.13) = 3.81, for the growth estimate F(0.05, 2.12) = 3.89.

Graph : Consistency between full set of indicators and headline indicators

Average country ranking - level 2003 Average country ranking - growth 1999-2003

9 y = bx + a PT

10 y = 0.7853x + 1.7178 IT PT

R2 = 0.8414 EL DE

9 H_o: b = 1, a = 0 FR DE ES NL

Fstat = 2.58 ) 8 SE

FI AT

) 8 LU F(0.05,2,13) = 3.81 UK BE DK y = bx + a

 p-value = 0.114 IE 7 y = 0.294x + 5.2947 H_o accepted at 5 %

F S ( 78 AT NL FI F S ( 77

BE IT R2 = 0.3629

6 DK 7 EL UK ES H_o: b = 1, a = 0

Fstat = 19.7

5 SE FR IE F(0.05,2,12) = 3.89

 p-value < 0.001

4 H_o rejected at 5 %

4 5 6 7 8 9 10 11 6

5 6 7 8 9 10

HI (13) HI (13)

Note: HI and FS represent the average raking for, respectively headline indicators and full set of indicators. The number in brackets represents the number of indicators used for the analysis. HI (13) includes all headline indicators except the dispersion of regional employment rates, due to missing data. The lower the value on the axes, the higher the ranking. Source: Commission services.

The test described above is used for all countries jointly. The hypothesis about equality of the average rankings is accepted at 5 per cent except for the growth analysis and for the year 2000. On the whole, it is possible to identify clusters of countries that are robust. The assessment based on the full set of indicators and the shortlist, in terms of average ranking, is therefore quite consistent and robust for the countries of the EU-15.

A further test is to identify countries for which the average ranking is statistically different across the two sets of indicators. 3 The

shaded cells in the table below indicate that the average rankings obtained on the basis of the shortlist and on the basis of the full set of indicators are significantly different from each other. This test confirms the previous graphical analysis. The robustness of the shortlist is however lower for countries such as Denmark, Spain, France and Greece.

Table: T-test: Level in 5 Years and Growth over 1999-2003 EU-15 4

T i statistics for two sided test (at 5% critical value )

Levels Growth

1999 2000 2001 2002 2003 1999-2003

AT -0.92 -0.60 -0.23 -0.10 0.07 1.52

BE -0.55 -0.47 -0.56 -0.34 -0.35 -1.10

DE -1.14 -0.76 -0.14 0.04 -0.20 1.39

DK -0.60 -0.81 -0.80 -1.08 -1.13 -0.94

ES 0.77 0.89 0.60 0.85 0.94 0.71

FI 0.87 0.51 0.48 0.63 0.48 -0.83

FR -0.78 -0.41 -0.72 -1.54 -1.49 -0.19

EL 0.98 1.14 0.71 0.64 0.87 -1.28

IE 0.04 0.02 -0.32 0.11 -0.11 0.82

IT 0.06 0.32 0.28 -0.05 -0.21 0.58

LU -0.63 -0.70 -0.32 -0.15

NL 0.44 0.24 0.39 -0.06 -0.08 1.04

PT 0.15 0.18 0.36 0.31 0.26 0.07

SE 0.15 0.18 0.54 0.62 0.55 -1.08

UK 0.13 -0.08 -0.14 -0.26 -0.15 -0.68

5% critical value 0.83 0.83 0.83 0.83 0.83 0.83

• 1) 

  The average rankings are computed by, for each country, averaging the rankings obtained for each indicator.

  N

• 2) 

  This F-statistic is defined as F = ( N

  − 2 )( R −

1 R ) , where, R ∑ = ( Y − ˆ − ˆ )

2 =

N

2 R i

a b X i and R ∑ ( Y − X ) 2 1 i i , a ˆ , ˆ b are OLS

i = 1 i = 1

estimates based on sample (X i , Y i ), i = 1, .. , N.

• 3) 

  Such identification is possible via the two-sample t-test, which is more severe that the previous one. The test is conducted for each country independently. Let X i represents the average ranking of country i over the short list and Y i represents the average ranking of country i over the full set of indicators. Denote further by S x j , S y j their corresponding standard deviations and by x and y the number of indicators in short set and the full

list. Under the null hypothesis X i = Y i , the test is defined as follows:

T = X

−

i Y i xy ( x

+ y − 2 ) , which follows t distribution with (x+y-2) degrees of freedom.

i

( x − 1 ) S 2 + , ( y − 1 ) S 2 , x + x i y i y

• 4) 

  Luxembourg is not included.

Moreover and probably more importantly, the 4.3 Role of the structural indicators in the Lisbon assessment based on the shortlist has been shown to be Strategy

relatively robust. Robustness analysis conducted by the

Joint Research Centre (Ispra) for the EU-15 confirms the The economic policy coordination serving the Lisbon overall consistency between the ranking obtained with Strategy is organised in three stages. First, the main the shortlist and the database; thereby reducing the risk decisions are taken and economic policy orientations are of a partial or biased analysis when displaying the agreed at the annual Spring Council. A report prepared performances (levels) and progress assessment for the by the Commission, the so-called Spring Report, is used 14 headline indicators. 417 In particular, leaders and to guide the decisions of the Heads of States. Second, laggards can be identified in a robust way. those decisions are translated into policy

recommendations in the Broad Economic Policy Guidelines (BEPGs), the Employment Guidelines (EG) and the Internal Market Strategy (IMS). Finally, the Implementation Package (three Implementation Reports (IR), one for each of the BEPGs, EG and IMS) and the

417 The same conclusion holds to some extent for the new Spring Report assesses whether these policy

Members States. For some of these countries discrepancies recommendations have been implemented. Two

exist due to different reasons (problems of data availability

and quality, as well as strongly changing economies). This instruments of governance contribute to the

makes the use of the shortlist for these countries a bit less implementation of the Lisbon strategy: the BEPGs and effective as a summary of the long list, at present. the Open Method of Coordination (OMC).

The structural indicators are mainly used in the Spring 4.4.1 Method used to select the indicators

Report to provide an assessment of the Member States performance towards the main Lisbon targets. They are The basic rationale underlying the choice of indicators also used in other Commission and Council reports such has become less clear as the requirement to reach as the BEPGs. The structural indicators are a useful tool agreement between the Commission and Council on the of the policy coordination, as they provide valuable structural indicators necessitated the finding of a information about the steps already taken by national compromise solution. The choice of indicators has been governments to achieve the Lisbon targets. based on discussions on the relative merit of individual

The success of this policy coordination instrument in indicators rather than on the consistency of the indicator assessing Member States’ policies and in measuring set as a whole. The yearly revision of the database also

their performance depends on two main factors: leads to continuous addition of new indicators that dilute the potential influence of each indicator on monitoring

• The structural indicators should increase the and policy analysis.

capacity to correctly monitor national performances

in the most relevant areas and establish a There is an agreement that the shortlist established in benchmark for the Member States that is as accurate 2003 would be revised every three years only. The 14

as possible. headline indicators provide a good and manageable set of indicators that should help to attract and increase

• The assessment exercise should help to identify the public awareness of structural reforms. In the face of the areas where Member States lag behind and where considerable size of the comprehensive database and further reform effort is necessary; eventually it according to the principle of streamlining, it is important should aim at encouraging Member States to to resist the temptation for more frequent changes or

significantly step up the pace of structural reforms. extension. Following completion of the mid-term review of the Lisbon strategy, a revision of the shortlist could

In addition, it is important to take into account the be envisaged, in order to better reflect the priorities for different starting positions across Member States to action during the second half of the decade. However, if provide encouragement to those who have undertaken it is decided to leave the overall objectives of the Lisbon difficult reforms, when using, presenting and strategy in place, then, according to principle of interpreting the structural indicators. Therefore, there streamlining, the shortlist should remain constant.

has been an increasing focus on measuring both the level as well as the progress (growth rates) in Member States’

performance. The Statistical Annex of the Spring Report 4.4.2 Relevance of the list: policy or performance illustrates the use made of the structural indicators. In indicators?

particular, Table 15 and Table 16 of the Spring Report One can take a rather critical stance with respect to the 2004 present an assessment of the EU and Member indicators’ reflection of reality. Are the indicators States performance in terms of levels and progress made successful in measuring the progress in terms of reforms since Lisbon. implemented? There has been a gradual shift in

emphasis from an ‘input indicators’ approach (i.e. policy

4.4 Critical assessment measures taken, progress with respect to the reform

The structural indicators have been successful in several agenda) to an ‘output indicators’ approach (i.e. picture ways. They have been used in the Commission’s Spring of the actual performances). While the latter is more in Reports, in the BEPGs, as well as in other Commission line with the conclusions of the Lisbon European documents to provide statistical support for policy Council, the former may be more effective in bringing messages and to measure progress towards the Lisbon about policy reform. As such, the ‘input indicator’ could objectives. Being used in different processes, the help to improve the peer pressure and the efficiency of structural indicators database is a guarantee for the structural indicators as an economic policy consistency between policy messages. The structural coordination instrument. A drawback of the ‘input indicators have also attracted a lot of outside attention indicator’ is that the link between policies and

being one of Eurostat’s most popular websites. performances is not always straightforward. Many external factors, including the business cycle, intervene.

However, the indicators have also been subject to a lot This criticism also applies to the ‘output indicator’ of criticism focusing, on, amongst other issues, the basic where the performances measured are not necessarily rationale underlying the choice and the selection method due to efficient economic reforms but could be due to of indicators, the relevance of the list and its focus on favourable economic conditions.

policy versus performance indicators, the lack of

effectiveness of the indicators as a tool to bring about The choice of the benchmarks and the reference to the policy change, and the use of the indicators for country US model can also be questioned. Some of the structural

rankings (including a brief presentation of country rankings methods in Box 3). These different points of criticism are considered in somewhat more detail below.

differences observed between the USA and Europe may 4.4.4 The use of the indicators for country rankings

result from strongly different social preferences. 418 The success of Lisbon depends on the effectiveness of

peer pressure and the benchmarking of best practices 4.4.3 Effectiveness of the indicators as part of the amongst countries. So far, it has not proven to be very

governance system successful. Along the lines of the previous section, a

One major criticism is related to the effectiveness of possible candidate for communication instrument that structural indicators in assessing the implementation of could further simplify and highlight the policy messages policy measures and as a tool to bring about policy is the country ranking.

change. There are two (complementary) approaches to The communication potential of the ranking method is this issue. illustrated by the numerous country ranking produced by

First, the structural indicators, particularly the shortlist, other national of international institutions, think-tank are used to give a synthetic view of the performance of and the like (Financial Times, World Economic Forum, the Member States. But the rather descriptive analysis Unice, OECD, The Economist). There is an obvious based on the 14 headline indicators needs to be demand for this kind of popularization instrument. supplemented by a more in-depth analysis in order to However, although the indicators have not been used in provide a reasonable and more comprehensive picture of a mechanical way, there has been fierce opposition reform implementation and their effectiveness in against the inclusion of country rankings in the improving performances. More sophisticated analytical Commission and Council reports, particularly from the tools may be needed to better understand the effects of countries that perform relatively poorly.

reform as well as the interactions and tradeoffs between There are pros and cons to the use of country rankings. different reforms undertaken. Provided the structural As mentioned above, the major advantage of the reform assessment is underpinned by a sound and rankings is its communication potential. Moreover, since thorough analysis, possibly including a specific the rankings are computed anyway by others actors, analytical framework that would have unanimous there is a strong case for the Commission and the support, the credibility (and hence the desirability) of the Member States to do it themselves. However, arguably, whole Lisbon Strategy would increase. Eventually, the rankings only display a factual view limited to the 14 support for (sometimes painful) structural reform would headline indicators without guarantee that the rankings rise. would convince Member States to undertake the

Second, producing and clearly conveying convincing necessary reforms. Nevertheless, there exist several evidence to the public is a major challenge. Against the methodologies to construct country rankings and, in background of falling public support for structural general, they have proven to be quite robust; especially reforms, some reflection is necessary on a new for leaders and laggards, whose rankings tend to be dissemination instrument that would increase public more robust than that of the countries in the middle. It is awareness of the potential benefits of such reforms. also interesting to note that the results obtained by Improving the communication of the Lisbon Strategy calculating the average ranking of countries are and increasing public awareness is a first step in corroborated by those of these more sophisticated bringing about policy change. The shortlist is a good methods. Some of these methodologies are illustrated candidate as a ‘marketing instrument’ to draw the below in Box 3. In particular, the results of the Benefitattention of the public. As a corollary, the shortlist could of-the-doubt method are worth highlighting. In those even be shorter than the current one and, accordingly, cases where some countries could argue that the weights only reflect main priorities. Next to an in-depth analysis, used penalize their revealed performance, the method the shortlist of indicators could therefore be designed as shows that using the most favourable weights for a given a powerful communication device, which would country does not change significantly the results. This is

contribute to the transparency of the Lisbon Strategy. particularly relevant for the poor performers who are more likely to complain about country rankings. The

results prove that the poorest performers remain poor performers even if the most favourable weights for them

are used.

418

  For recent contributions to this debate, see Blanchard (2004) and Gordon (2004).

Box 3: Illustration of alternative methods for country rankings

1

The Joint Research Centre (Ispra) has computed country rankings based on the shortlist of structural indicators by using different tried and tested statistical techniques, among which two of composite indices. Overall, the country rankings appear quite consistent regardless of the techniques used. As far as the charts below are concerned, the rankings corroborate the synthesis analysis made in the 2004 Spring Report (Section 2.5). This analysis enabled a distinction to be made between Member States with relatively overall achievement to date (among others the Nordic countries) and those that are performing relatively poorly (Southern Europe). Similarly, this approach allowed an identification of the countries that have made rather good progress towards the Lisbon objectives (Greece) from the ones that have been rather disappointing (Germany, Austria, The Netherlands and Portugal).

For all charts included below, the performances in terms of levels (year 2003) are displayed on the horizontal axis while the progress (measured in growth rates) are plotted vertically. The clusters of the leaders and laggards can be observed in the top right hand and bottom left hand corner respectively. The chart on top displays the average rankings of 2003 level versus 1999- 2003 growth. The composite indices as given by the Equal Weighting method (EW) and the Benefit-Of-the-Doubt method (BoD) are presented in the two bottom graphs.

The EW method computes a synthetic index by using equal weights for each of the indicators included in the index. The BoD procedure calculates a synthetic index for a given country by using the best set of weights, which maximizes the index for that country with respect to the best performing country using the same set of weights. The same procedure is followed for each country. Weights are therefore country-dependent. To put it simply, this weighting scheme provides the best possible ranking for each country individually. Nevertheless, in general, even though this method provides the best weights combination possible, some countries clusters (leaders and laggards) still appear.

Graph : Country ranking based on different computing methods

Average country ranking Benefit-of-the-doubt

11 1.1 BE

10 AT DE 1.0 EL FR IE UK DK

NL ES SE 3 9

PT FI

00 IT 03 0.9

9 - 2 8 IT

IE

99 ES 9-

20

 1 7 FI 99 0.8 PT NL w th SE

FR

6 DK UK w th

 1 0.7

G ro BE EL

5 G

ro

0.6 DE AT

4

4 5 6 7 8 9 10 11 0.5 Level 2003 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Level 2003

Equal weights

1.0 EL

0.9 BE UK SE

3 FR

0.8 IE FI DK

-2 00 ES IT

0.7

 1 9

99

w th 0.6 PT

NL

G ro 0.5

DE AT

0.4

0.3

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Level 2003

Source: Commission services.

• 1) 

  See Tarantola, Liska and Saltelli, (2004).

4.5 Conclusion This section firstly reviews progress made in response to

This section reviews the structural indicators and their a report by the Economic and Financial Committee role in the progress assessment of EU countries towards (EFC) on information requirements in EMU endorsed by the Lisbon objectives. the Ecofin Council in January 1999,

419 and the joint Action Plan on EMU Statistics 420 which responded to The structural indicators are a useful tool of policy this report and was endorsed by the Ecofin Council in coordination, which are used in various Commission and September 2000.

Council reports. They contribute to the monitoring of

structural reforms undertaken by the Members States. The note then considers the response to concerns put The structural indicators database is therefore a forward at the Barcelona European Council in March guarantee for consistency between policy messages. In 2002 regarding the need to improve European Economic addition, the regular use of a shortlist of headline Statistics. This includes the development of a set of indicators, as a complement to the entire database, has Principal European Economic Indicators and agreement proven to be a reliable tool. Robustness analysis on a Code of Best Practice on the compilation and ascertains that the assessment of countries performances reporting of budgetary data. It concludes by taking a with respect to the Lisbon objectives made on the basis forward look at initiatives in place to delivery further of the shortlist and of the entire database is relatively quality improvements.

 421

consistent and robust.

However, against the background of the EU-15 (an EU- Table 8: Summary of initiatives

• 25) 

  falling short of the Lisbon agenda, some reflection

remains necessary on the structural indicators as a peer 11/1998

Monetary Committee produces first report on information requirements in EMU.

pressure device. This section has considered some of Transition to third stage of EMU, Ecofin

these shortcomings and provided some suggestions for 01/1999 Council endorses report of the Monetary improvements. One possibility is the technique of the Committee (now EFC). country rankings. Although they constitute an attractive Ecofin Council endorses a second progress

communication device, the rankings cannot guarantee 06/2000 report and invites the ECB and the Commission that Member States will carry out the necessary reforms. (Eurostat) to establish an EMU Action Plan

Nevertheless, it is clear that the rankings alone are not where urgent progress should be made. sufficient and that they should be accompanied by in 09/2000 Ecofin Council endorses EMU Action Plan. depth analysis. 11/2001 Ecofin Council endorses third and fourth

progress reports. Barcelona Council invites Commission and

• 5. 

  Improvements in European economic 03/2002 Council to present a comprehensive report on

statistics under EMU euro area statistics for Spring European Council 2003.

Commission publishes communications on ‘the

It is widely acknowledged that modern democracies can need and means to upgrade the quality of only function efficiently if policy makers and the public 11/2002 budget statistics’ and ‘Towards improved at large are well informed about economic and social methodologies for euro zone statistics and developments. High quality, reliable and timely statistics indicators’. are clearly required for the development and evaluation Ecofin Council endorses fifth progress report. of appropriate economic policies. In recent years, the Approves a code of best practice on the

creation of monetary union, with a single and 02/2003 reporting of budgetary statistics. Approves the independent monetary policy and decentralised but cocomprehensive report on euro area statistics,

ordinated fiscal policies, has increased the need to make and transmits it to the Spring European Council 2003.

greater progress in the availability of statistics, Ecofin Council endorses final EFC progress

improving and harmonising the methodologies used to 06/2004 report and makes recommendations for further draw up euro area statistics and indicators. This is improvements. alongside the ongoing, more general need to continue to

improve EU statistics. Source: Commission services.

This section reviews the progress that has been made in improving the quality of euro area macroeconomic statistics in response to a number of Commission and

Council initiatives (a full timetable is shown in Table 1).

The proceeding section provides a separate review of the structural indicators which are used to measure progress 419

against the Lisbon economic reform goals for all EU See Monetary Committee (1998). 420

Member States. See European Commission (1999). 421

  For an alternative account of developments in economic statistics for the euro area between 1991 and 1993, see ECB (2003).

5.1 The need for improved statistics under EMU ECB and Eurostat (the Statistical Office of the European Communities).

The availability of high quality statistics in the euro area

is essential to the conduct of monetary policy and the Thus for the purposes of economic and monetary policy coordination of economic and, in particular, fiscal making, the EFC report argued that it is important to policies. The EFC’s original report in December 1998 have a methodologically sound and consistent recognised that statistics for the euro area were information system that produces relevant information somewhat underdeveloped and that the operation of the in a timely fashion. The system needs to facilitate the single monetary policy would require extensive and interpretation of data as well as the assessment of the improved information on areas including money growth, significance of any new information that comes banking, financial market statistics, financial accounts available. Sound and comprehensive statistical and balance of payments statistics, including the information reduces uncertainties and thereby the risk of international investment position. Most importantly, confusion and instability in the markets.

price statistics, including the Harmonised Index on Nevertheless, it was recognised that requirements put on Consumer Prices would be required for the successful statistical data, notably as to detail, frequency, and operation of the policy. The report acknowledged that timeliness, vary between users, depending on the the ECB’s main interests would lie with developments specific purpose. This points to an obvious trade off in in the euro area as a whole, but also noted that detailed data production between the speed and level of detail complementary information on developments at and accuracy.

Member-State level would be valuable if developments

in the euro area and the effects of the single monetary 5.3 Priority areas for improvement, as identified policy were to be adequately assessed. in 1998

Member States retain responsibility for economic policy The EFC report recognised that major improvements in including budgetary policy in EMU. However, for the statistics had already taken place during the 1990s. In proper functioning of EMU, it is important that their particular, the Council decision in June 1996 to replace policies be mutually consistent and appropriate in the the European system of integrated economic accounts light of the single monetary policy. The Stability and (ESA79) with the new and more comprehensive Growth Pact (SGP) and the Excessive Deficit Procedure European System of National and Regional Accounts (EDP) ensure that budgetary discipline in the EU is (ESA 95), 422 had introduced major improvements and examined on the basis of two quantitative criteria for the extended coverage. It obliged Member States to report government deficit and debt level. Therefore the national accounts data from April 1999 onwards within availability of quality statistical data is crucial to ensure a well-defined time schedule on both an annual, and for adequate implementation of the agreed budgetary some main aggregates, a quarterly basis. In addition, surveillance framework and effective coordination of Council and Commission regulations on the HICP had budgetary policies. Non reliable statistics may lead to led to the production of high quality indices since 1997 the wrong policy decision at national and European and since autumn 2001, the provision of an advanced level, to an inefficient coordination of economic estimate for total HICP growth for the euro area at the policies, and put the credibility of procedures and end of the current month. Agreements had also been institutions at risk. made that were expected to lead to improvements in the

quality of short-term economic statistics and of constant 5.2 Attributes of high quality statistics price GDP estimates.

It is important that the compilation of the statistics be However, the report noted that for the purposes of both free, and perceived to be free, from political economic policy, quarterly national accounts are the interference. The general public needs to be able to trust core statistical information system and should therefore that official statistics do not serve any other interest than be developed further. Given the long lead in times in providing the best possible information on the state of improving the production of statistics it was also clear affairs. that EU members that are not participating in the third

The EFC report also noted the need for co-operation stage of EMU should fully engage in steps to improve between different agencies involved in gathering euro-area statistics, in anticipation of their eventual statistics. At the national level, there are two main joining.

groups of data producers. National central banks are mainly in charge of statistics on monetary aggregates, financial transaction and balance sheets, and balance of payments statistics, whilst national statistical offices generally deal with most other statistics. In several countries, Ministries of Finance also have major

responsibilities in the compilation of budgetary data. At 422 Council Regulation (EC) No 2223/96 i of 25 June 1996 on

the European level, their respective counterparts are the the European system of national and regional accounts in the Community.

The report urged for rapid progress in a number of areas: Balance of payments and trade

Public Finances The EFC report considered that the creation of EMU

As discussed, it has always been clear that the would add considerably to the difficulty in measuring achievement and maintenance of public finances is of trade and financial flows within the euro area. The central importance in all Member States and to the creation of the single market had already led to a success of EMU. Comprehensive information on public lowering of quality, regarding intra-trade statistics, for finances decisions are therefore crucial to implementing example, huge statistical discrepancies had emerged the provisions of the SGP and EDP. between dispatches and arrivals.

The report stressed the importance of a shift to However, the Committee considered that intra-EU trade producing quarterly national accounts for the general data contained more detail than was necessary for the government sector. Comparable and timely information coordination of economic policy. For policy on budgetary indicators, including tax revenue, social coordination purposes it was considered sufficient to security contributions and the borrowing requirement collect only quarterly information and for a smaller would also provide useful indications for the monitoring number of product categories. The Committee

of budgetary developments. recommended that the present intrastate system be simplified and other methods explored to arrive at a

Labour market suitable output, while reducing costs.

The EFC report observed that cross-country

comparisons of the labour market and of labour costs 5.4 Improvements made in response to the EMU

will demand more attention in EMU. This is because in action plan

EMU the link between labour cost developments and Having endorsed the EFC report on information employment is strengthened, since there is no longer any requirements in EMU, the Ecofin Council invited the possibility of regaining competitiveness between Commission, in close co-operation with the ECB, to Member States through the adjustment of nominal establish an Action Plan on EMU statistical exchange rates. requirements in order to address the deficiencies

Given the importance of wage inflation within overall outlined in the report. The first action plan was endorsed inflation, information on labour market developments by the Ecofin Council in September 2000. It set a target and labour costs should, ideally, be consistent with, or for the production of national data, to permit the timely even integrated in, the national accounts. The report compilation of reliable key statistics for the euro area argued that further progress was required on producing and the EU, with at least 80 per cent coverage of quarterly statistics on the compensation of employees Member States’ data.

and costs per unit of labour or per unit of production. In The EFC made an assessment of progress made since addition it was considered important to implement the their 1999 report in May 2004, 423 a summary of which is labour force survey providing quarterly results on provided in Table 9. It concluded that overall, employment and unemployment by 2000 at the latest. substantial improvements have been made to economic

Short term business indicators statistics, both for euro-area aggregates as well as for the national components. Particularly strong progress has

The EFC report stressed that short-term business been made regarding the availability, coverage and indicators are essential for assessing developments in the timeliness of quarterly national accounts, quarterly economy, notably its position in the cycle. Short-term government finance statistics and short-term business business indicators can supplement national accounts by indicators. However, they also concluded that more providing early data on output, demand and prices. Such progress is still needed, particularly regarding labour data will clearly be important for the surveillance tasks statistics, where the process has not yet yielded the of the Commission and for the ECB in assessing expected results and more efforts are needed. The EFC economic developments and in deciding on the made more detailed comments on the specific areas of monetary policy stance to be taken in response. the Action Plan:

The Council’s adoption of a regulation in May 1998 Quarterly national accounts (main aggregates)

regarding short-term indicators was viewed as a major

step in improving the quality and speeding up the The EFC considered the provision of national data to production of short-term quantitative indicators; have improved, with most countries now complying although a number of Member States had negotiated with the 70 days deadline set by the EMU Action Plan. derogations that threatened to reduce the regulation’s Nevertheless, there are still gaps for several countries impact in the short to medium term. It was also hoped and variables that affect euro-area aggregates. that that qualitative short-term indicators, regarding Since May 2003 Eurostat has published GDP flash business and consumer confidence would be produced to estimates for the EU and euro area after 45 days. At

complement the quantitative indicators.

423

  Economic and Financial Committee (2004).

around the same time flash estimates are available for Labour Market Statistics Germany, France, Italy, the Netherlands, Greece,

Finland and, before that date, for Belgium and the The EMU Action Plan required improvements to data on United Kingdom. Spain and Sweden are planning to employment, hours worked, the labour force and labour

move to the 45 days target. cost statistics. This remains the area where least progress has been made in terms of European aggregates. Yet the

Quarterly euro-area aggregates are published in two legal instruments requested by the Action Plan have main releases with a coverage of well above 80 per cent been adopted and are expected to yield significant

of the euro area. The output and expenditure side is results by 2006. released after about 65 days. The release of the

remaining variables, in particular all income variables, From 2004, quarterly ESA 95 data for hours worked however, is still only possible after around 105 days, have to be transmitted by all countries except Austria

over a month after the Action Plan target. and Portugal (who have a derogation). The EFC urged all countries, in particular Spain and Italy, to meet the

Quarterly public finance statistics legal obligations for the crucially important hours

The EFC concluded that the EMU Action Plan has worked data.

stimulated the process of compiling quarterly public Progress has been made with regard to euro-area shortfinance

statistics covering comprehensive data on term labour cost data. This follows the adoption of a

government revenue and expenditure, financial regulation 427 in February 2003 governing the

transactions and balance sheets as well as EDP debt. harmonisation of the labour cost index. First estimates,

A regulation 424 covering quarterly data on taxes, social with coverage of more than 90 per cent can now be contributions and social benefits, has been successfully published within 80 days after the quarter, close to the implemented; with a set of back data starting in 1991 target of the EMU Action Plan of 75 days. Several

provided in July 2002. countries have improved the industry coverage and the measures for labour costs and hours worked. The new

425

Similarly, a regulation covering the remaining legislation also requires the coverage of the services categories of revenue, expenditure and net lending/net sector and improving the timeliness of the first release to borrowing, was enacted in June 2002 and data became 70 days. However, due to extensive derogations the available in September 2002. However, in line with an objectives of the regulation will not be fully achieved

agreement laid down in the Council minutes to ensure before 2005. quality, the national data are subject to a trial period,

ending in 2005 at the latest. Until then, they are Short-term Business Statistics

confidential unless made public by the Member States. The objective of the Action Plan to produce indicators Most of the countries are submitting the entire set of by the end of 2001 (new orders by end 2002) was variables within 90-100 days. achieved for a number of indicators, including industrial

Quarterly financial transactions and balance sheets for production, output prices and retail trade turnover. central government and social security funds have been Progress has been made more recently regarding the provided by all Member States, although the coverage release of industrial new order statistics, and early needs to be further improved. Several Member States estimates for euro-area retail trade turnover. Qualitative have also voluntarily transmitted quarterly data for other surveys on service industries are already published on a sub-sectors of general government. The next step monthly basis and the Commission plans to further

towards a full set of quarterly financial accounts for the extend its coverage within the service industries.

government sector is the implementation of the recently For other variables, though, the objectives of the Action adopted regulation on quarterly financial accounts for Plan have not been achieved. For euro-area aggregates,

general government. 426 In particular, Germany, Greece the situation is still not satisfactory for indicators in the

and France have to make an effort in order to comply construction sector and for services.

with the Action Plan targets set in this area. A number of countries have still to comply with Council

Regulation 1165/98 i aimed at ensuring good quality short-term statistics. More emphasis also needs to be placed on improving the comparability of statistical

methods and adjusting for seasonal and calendar effects.

424 Short-term Public Finance Statistics Regulation. Regulation

External Trade Statistics (EC) 264/2000 of 3 February 2000, OJ No. L 29 of 4

February 2000, pp.4. The EMU Action Plan set Member States the target of

425 Regulation on quarterly non-financial accounts of transmitting first euro-area aggregates after 40 days.

government. Regulation (EC) 1221/2002 i of 10 June 2002, First aggregates are presently transmitted within 42 days

OJ No. L 179 of 9 July 2002, pp. 1. by most Members States on the basis of a gentlemen’s 426 European Parliament and Council Regulation (EC)

501/2004 of 10 March 2004, OJ No. L81 of 19 March 2004,

pp.1. 427 Regulation No. 450/2003 i.

agreement. However, regulations to be implemented in monthly and quarterly euro-area and EU indicators 2005 will ensure that the 40 day deadline is met. implicitly changing almost every day as new or revised

Revisions in balance of payments statistics have been data is published by national statistical institutes. The progressively reduced, although the bias in its errors and full set of indicators with existing and target compliance

omissions continues to raise concerns. In order to is included in the annex to this section.

preserve the quality of the data, Member States have

started to elaborate national action plans on future 5.7 Code of best practice on budgetary statistics

compilation systems. Future collection systems will rely The February 2003 Ecofin Council also gave its support more on multiple sources, in particular direct reporting to a code of best practice on the compilation and by enterprises. In many countries work in this direction reporting of data in the context of the Excessive Deficit has already been completed or is underway. Procedure, building on an earlier communication from the Commission. 429 In putting forward the code, the

5.5 Beyond the action plan: the response to the Commission recognised that although considerable

Barcelona European Council progress had been made in the compilation and reporting

The Barcelona European Council of March 2002 also of budgetary statistics, experience had revealed some recognised the importance of the availability of high weaknesses in terms of reliability, transparency and quality statistics in a monetary union. It added fresh timeliness of budgetary statistics. The code of practise impetus to the process by inviting the Commission and thus signalled a strong commitment by all parties to

the Council to present a comprehensive report on euroimproving performance in these areas.

area statistics in time for the Spring European Council The Commission communication argued that 2003. government accounts were not as reliable as they should

The final report was approved by the Ecofin Council in be and subject to large revisions. This had been February 2003 before being transmitted to the 2003 highlighted by the late identification in 2002 of an Spring European Council. 428 The report recognised the excessive deficit in Portugal for 2001 and large upward considerable progress that had been made in the revisions in the government deficit and debt levels of

improvements of EU and euro-area macroeconomic other Member States.

statistics as a result of the implementation of the Action The communication also noted that the government Plan. But it also noted that for the European Statistical accounts of several countries are not transparent enough. System to produce macro-economic statistics reaching For example, for several countries, the government quality standards comparable in terms of availability and deficit and the change in debt level are not easily timeliness to those of the United States, another reconcilable with other indicators, for example the cash quantum leap was needed. based balance of the government sub sectors. There had

also been problems in the transmission of government

5.6 Principal European economic indicators data by some Member States in terms of both timeliness

and the completion of reporting tables. In recognition of weaknesses identified in Barcelona, the

Ecofin Council gave its support to a list of key European Finally, the communication noted that in some Member indicators (Principal European Economic Indicators States, the reporting tables are prepared by the Ministry (PEEIs, see Table 10)) that had been proposed by the of Finance and the national statistical institute (NSI) has Commission, in agreement with the main European a relatively minor role in the process. This raises policy users. For this set of key indicators it was agreed concerns about independence given that the statistics are that focus would be directed to the provision of a more the basis for assessing the budgetary performance of complete range of variables, higher timeliness and each country. In contrast, although the Commission higher frequency of the time series than foreseen by the fulfils the role of statistical authority regarding the EDP EMU Action Plan on Statistical Requirements. It was and the SGP, in the internal organisation of the additionally agreed that the Principal European Commission, the tasks of scrutinising the reported Economic Indicators will be produced for the euro area accounts and interpreting the accounting rules are on the basis of a sufficient – but not necessarily carried out by Eurostat. By delegating this task to complete – coverage provided by the Member States, Eurostat, the commission seeks to ensure that the with later releases having a broader coverage. Such a accounting and statistical issues are treated release schedule is in line with the ‘First for Europe’ independently by an impartial and technically competent

principle. This principle means that the release calendars body according to objective criteria. (of both first releases and subsequent revisions) for the The code of best practice has addressed a number of PEEIs and for respective national contributions are these issues, in particular, reinforcing and reaffirming aligned and take into account European policy needs. the independence of statistical authorities based on

Aligning release calendars reduces the problem of

428 See Council (2003b). 429 See European Commission (2002).

scientific methods. The code outlines best practise • A statistical framework on long-term interest rates

regarding: has now been established. The long-term interest

• the compilation and reporting of budgetary data by rate statistics used for convergence assessment

Member States; purposes for the new Member States was released for the first time at the end of April 2004.

• the securing of quality of budget data, including • The new Member States regularly transmit national

through NSIs providing to Eurostat a detailed

inventory of methods, procedures and sources used balance of payments data on a monthly, quarterly for the compilation of government deficit and debt and annual basis and statistics on their international

data; and investment position on an annual basis, following methodological standards agreed at the European

• publication of the budgetary data by the level. However, additional efforts are needed by

Commission within a few weeks of the reporting several new Member States to achieve compliance

deadline. with the requirements of the Action Plan.

The code also provides for increased powers to ensure • As regards the timely transmission of detailed

the quality of reported data. In addition to acting as the statistics on extra-trade considerable progress has final authority on the interpretation of accounting rules, been made. All but two new Member States are able Eurostat has the authority to examine in depth to meet the transmission deadline of 42 days. government accounts for each Member State and where Cyprus and Malta need to further adapt their it deems necessary to make appropriate amendments to national collection systems in order to comply with data reported by Member States prior to final the timeliness requirements.

publication. • The new Member State also need to continue to

5.8 Progress on implementation of the action make progress regarding the relevant infra-annual

plan for the new Member States data. • Concerning quarterly national accounts, priority

The high level meeting with the (then) Candidate

Countries in May 2003 in Athens endorsed the Action should be given to the provision of seasonally Plan on economic, monetary and financial statistics for adjusted data and to compliance with the new the Candidate Countries. The meeting identified six transmission deadline of 70 days laid down in priority areas where the countries would have to Regulation 1267/2003 i.

concentrate their efforts in the run-up to the accession. • Only half of the new Member States have started

The final progress report on information requirements in transmitting quarterly public finance statistics. It is EMU concluded that good progress has been achieved in important to speed up work in this area. In addition, relation to the action plan, in particular as regards annual quarterly financial accounts should be taken up with national accounts as well as the primary convergence high priority.

indicators (government deficit and debt, HICP and long • As regards labour market statistics, the situation is

term interest rates). But it also noted that the length of

the time-series is not yet satisfactory for many countries. generally satisfactory concerning unemployment On specific priority areas, the EFC noted that the and labour cost data. Only one country has failed to following progress had been made (the EFC report implement infra-annual Labour Force Surveys. A annex contains detailed progress on the Action Plan in primary goal must be the timely transmission of a

each new Member State and Candidate Country): complete set of quarterly employment data under ESA 95.

• Regarding annual national accounts, all new • In the area of short-term business statistics every

Member States have reached an appropriate level of

compliance with ESA 95. Data availability is effort must be made to ensure compliance with

considered satisfactory and the key variables are existing legislation.

well covered. In some countries, efforts are needed

to fill existing gaps, with particular care required 5.9 Conclusions

regarding the revision of back data. The new A process of reform, which began in 1999 and was Member States must now make efforts to meet the accelerated by the EMU Action Plan in 2000, has 70 day deadline for the transmission of the main succeeded in stimulating substantial improvements in aggregates and provide data on hours worked. economic statistics in the EU. As the 2004 OECD

• The new Member States provide Harmonised Economic Survey of the Euro Area noted, euro-area Indexes of Consumer Prices which are up to the statistics have improved considerably since 1999, both standards of the HICPs compiled by the old in scope and timeliness. Availability, coverage and Member States in terms of timeliness, comparability timeliness of quarterly national accounts, quarterly

and compliance. government statistics and short term business indicators are now significantly better than they were in 1999. This

in turn will reinforce the credibility and implementation The Ecofin Council on 2 nd June 2004 agreed a number

of the budgetary surveillance and the effective of measures to further improve euro-area statistics.

coordination of budgetary policies. Regarding budgetary statistics, the Council made

However, greater commitment is required from national particular note of the fact that on several occasions, authorities if the target to compile and disseminate a set these had been revised by Member States after a new of Principal European Economic Indicators is to be government took office. With this in mind, the Council achieved by 2005. More improvements are also required invited the Commission to make, by June 2005, a if the EU is to bridge the gaps with the most developed proposal for minimum standards for the institutional set statistical systems, particularly that of the United States. up of statistical authorities that reinforces the While the EU has had to concentrate on harmonisation independence, integrity and accountability of Member issues, the USA has been able to develop and refine new States’ national statistical institutes. In addition, having indicators such as hedonic price indices. observed that the requirements of the code of best

More effort is needed to improve the quality of labour practise remain to be fulfilled in many Member States, market statistics and in particular to ensure that hours the Council invited the Commission to strengthen the worked data are available for all Member States. In monitoring of the quality of reported fiscal data and addition, for short-term business statistics, only six report back by the end of 2004.

countries had fully or almost implemented the relevant Given the increasing requirements for high quality regulations by May 2004. Moreover, more work is statistics at both national and European level, the required to develop the statistical basis for the services Council also agreed that it is important to review economy and to minimise balance of payments statistical priorities and to reduce (legal) requirements asymmetries. for areas that are now considered to be of less

Agreement on the code of best practise marks a importance. The Council therefore invited the EFC, with significant step forward in the quest to ensure budgetary the assistance of Eurostat and the ECB, to produce a surveillance is based on high quality data. The code has report on ’negative’ priorities in statistics, which may already led to an improvement in the reporting of help to free resources for the implementation and budgetary statistics, with the EDP notification from continuous production of high-priority statistics and to March 2004 showing improved compliance regarding reduce regulatory burden. A preliminary discussion of reporting deadlines. There was also considerable priorities should take place by the end of 2004.

improvement in the availability of detailed data on the Finally the Ecofin Council invited the EFC to continue government sub-sectors, even thought they remain to monitor at regular intervals the quality and incomplete. However, compliance was not satisfactory availability of statistics needs for EMU and the EU, as regards the institutional arrangements in Member covering both euro-area/EU aggregates and the key States and the submission of their respective inventories. indicators, in particular the PEEIs, and government In this and other respects the requirements of the Code finance statistics covered by the Code of Best Practise. of Best Practise need to be fulfilled in a number of A follow-up report will be submitted in 2005. Member States in the coming years.

Table 9: Summary of improvements in European Economic Statistics since the start of EMU

Action Plan target Assessment of progress

Quarterly national accounts

• First reliable estimates within 70 • Almost complete availability of euro-area and EU aggregates based on a

days coverage well above 80 per cent of Member States’ data.

• Second estimates within 90 days • Most of the EU-15 (B, DK, D, F, I, GR, E, NL, P, FIN, S and UK) supply GDP • Back data compiled from 1980 and a full or partial set of main aggregates within the 70 days deadline. • Limited set of quarterly sector • Euro-area and EU flash estimates have been regularly released at 45 days since

accounts May 2003.

• Flash estimates available at 45 days for D, F, I, NL, GR, FI and UK. B,E and S

are planning to meet the 45 days target.

• Adoption by the Commission of the proposal for a European Parliament and

Council Regulation on quarterly accounts for institutional sectors. Quarterly public finance statistics

• Complete implementation of the • Quarterly non-financial statistics are regularly transmitted within the deadline of

short term public finance statistics 3 months (taxes, social contributions and social benefits).

regulation • All countries except IRL regularly transmit data for quarterly government • Quarterly non-financial statistics expenditure and revenue variables.

(taxes and social contributions) • Quarterly financial statistics for central government and social security funds

and financial statistics (where the sector exist) are supplied by all Member States on a voluntary basis. (expenditure and revenue) for Several Member States have also voluntarily transmitted quarterly data for the general government available other sub-sectors of general government. after 90 days. Labour market statistics

• Full and quick transmission • Currently 8 countries (D, F, I, E, NL, FIN, DK and S) supply employment data

(within 70 days) of data under (under ESA 95) within the deadline. Work is still necessary for IRL and UK to ESA 95, including employment comply with the deadline. GR and P should start supplying data.

and hours worked • Hours worked are transmitted by D, NL, FIN and S. Efforts have to be made by • Quick implementation of the the remaining Member States in 2004 (except derogations) to ensure regular

Continuous Labour Force Survey transmission.

(availability within 91days) • Several countries supplied quarterly results or proxies within the 3 months • Improve the quality of the labour deadline for the continuous labour force survey.

cost index (availability within 75 • Most Member States comply with the labour cost index target. European

days) estimates are available at 80 days with more than 90 per cent Member States’ coverage.

• Monthly unemployment data continues to be steadily calculated. Improvements

are to be expected according to the progress in LFS and continuous surveys. Short-term business statistics

• Quick implementation of the • Action Plan objectives achieved for industrial production, output prices and retail

short-term regulation for trade turnover.

manufacturing, construction and • Progress has been achieved concerning the release of new order statistics and

retail trade timeliness of retail trade turnover.

• Development of qualitative • The short-term statistics regulation has been fully or almost fully implemented

business survey only by DK, D, F, FIN, P and S. Euro-area aggregates of specific indicators (industrial output prices and some construction variables) suffer mainly because of the delays in meeting compliance of I and E. Major efforts to comply with the requirements are required by B, A, IRL and GR.

• The Action Plan objective of a regular and timely monthly publication of

qualitative business surveys (DG ECFIN) has been achieved. Coverage will be extended in the near future.

External trade statistics

• First estimates of extra-EU and • Regular release of first estimates for the EMU around 50 days. extra-euro area trade within 40 • The proportion of fully harmonised data in the first estimate is above 80 per cent.

days (80 per cent coverage)

• Detailed extra-EU trade (within

42 days)

• Detailed intra-EU trade (within 70

days)

Table 10: Principal European Economic Indicators, Target compliance (delay and coverage) and not fully

committed countries 430

Principal European Economic Indicators Current release delay Target Expected PEEI compliance (coverage in brackets) release for 2005

European US delay euro area euro area countries

aggregates indicators coverage not fully committed

Consumer Price Indicators

Harmonised Consumer Price Index: MUICP flash 0 na 0 ~65% FR, IE, LU, NL, AT,

estimate (~55%) PT, FI

Harmonised Consumer Price Index: actual indices 17 14 17 100% none

(100%)

Q UARTERLY N ATIONAL A CCOUNTS

First GDP estimate 45 30 45 ~90% IE, LU, AT, PT

(~90%)

GDP release with more breakdowns 65 30 60 ~90% IE, LU AT, PT, FI

(~70%)

Household and Company Accounts n/a 60 90 ~80% EL, ES, IE, LU, AT,

(n/a) PT, FI

Government Finance Statistics 100 60 90 100 none

(100%)

B USINESS I NDICATORS Industrial production index 47 14 40 ~95% EL, LU, AT

(~95%)

Industrial output price index for domestic markets 34 14 35 ~95% LU, AT

(~90%)

Industrial new orders index 54 28 50 (40) ~95% EL, LU, AT

(~85%)

Industrial import price index na 7 45 ~60% BE, ES, IE, IT, LU,

AT, PT

Production in construction 77 16 45 ~90% El, LU, AT, FI

(~95%)

Turnover index for retail trade and repair 35 15 30 ~90% BE, EL, IE, LU, AT

(~80%)

Turnover index for other services n/a n/a 60 ~95% EL, IE, LU

Corporate output price index for services n/a n/a 60 ~20% BE, DE, EL, ES, IE,

IT, LU, AT, NL, PT,

L ABOUR M ARKET I NDICATORS Unemployment rate (monthly) 34 5 30 <80% EL, IT

(~65%)

Job vacancy rate (quarterly) n/a 5 45 ~70% BE, GR, ES, IE AT,

(n/a) PT, FI

Employment (quarterly) 105 5 45 ~95% EL, LU

(~90%) (monthly)

Labour cost index (quarterly) 80 30 70 ~90% BE, EL, IE, LU

(~60%)

E XTERNAL T RADE I NDICATORS

External trade balance: intra- and extra-MU; intra- and 49 44 45 100 none

extra-EU (~95%)

Explanatory notes and comments:

The target dates for the release of European aggregates (Eurozone, EU-15) have been set in the Communication of the Commission to the European Parliament and the Council on Eurozone statistics (Com (2002) 661 final). The commitments have been made by National Statistical Institutes in autumn 2002 and updated in spring 2004. The current release of European aggregates is described by the number of calendar days after the end of the reporting period. Their coverage by Member States data was calculated with 2002 GDP weights for all indicators except HICP (2004 consumption weights) and labour market indicators (LFS employment weights 2002). Member States are classified as missing if they do not compile the respective indicator within the target time. In some cases Member States are not committed to the objective, but nevertheless delivering their data already as timely as required. For the indicators marked in dark/red the commitments are insufficient for achieving the objectives set in the Communication, for those in light/yellow adequate commitments have been made, but substantial progress has to materialise in the months to come.

430 This table draws on that provided at Annex VI of the EFC’s June 2004 status report on information requirements in EMU.

References

A GHION , P. AND E. C OHEN (2004), “Education et Croissance”, Les rapports du Conseil d'analyse économique, No 46, Paris: La Documentation française.

B LANCHARD , O. (2004), “The Economic Future of Europe”, NBER Working Paper No 10310.

C OULOMBE , S., T REMBLAY , J.F. AND M ARCHAND , S. (2004), “Literacy scores, human capital and growth across 14 OECD countries”, Statistics Canada, Catalogue No 89-552 MIE.

C OUNCIL OF M INISTERS (2003a), “Council Conclusions”, (15875/03), 08.12.2003.

C OUNCIL OF M INISTERS (2003b), “Comprehensive report of the Council and Commission on euro area statistics to

the 2003 Spring European Council”. Approved by the Ecofin Council, 18 th February 2003. Council document

number 7237/03.

D E L A F UENTE , A. (2003), “Human capital in a global and knowledge-based economy. Part II: assessment at the EU country level”, Final report for DG Employment and Social Affairs, European Commission, April.

D E L A F UENTE , A. AND A. C ICCONE (2002), “Human capital in a global and knowledge-based economy”, Final report for DG Employment and Social Affairs, European Commission, May.

D E L A F UENTE , A. AND R. D OMÉNECH (2001), “Educational attainment in the OECD, 1960-1995”, Working Paper D-2001-01, Dirección General de Presupuestos, Ministerio de Hacienda, Madrid

D YSON , K. (2000), “The Politics of the Euro-Zone. Stability or Breakdown?”, (Oxford: Oxford University Press).

E CONOMIC AND F INANCIAL C OMMITTEE (2004), “Status reports on information requirements in EMU”, Economic

and Financial Committee, 25 th May 2004.

E CONOMIC AND S OCIAL C OMMITTEE (2002), “Resolution addressed to the European Convention”, (CES 1069/2002), 19.09.2002).

E UROPEAN C ENTRAL B ANK (2003), “Developments in general economic statistics for the euro area”, ECB Monthly Bulletin, April 2003.

E UROPEAN C ENTRAL B ANK (2004), “The European Constitution and the ECB”, Monthly Bulletin, August, (ECB: Frankfurt am Main), pp. 61.

E UROPEAN C OMMISSION (1999), “Action plan on EMU statistical requirements’, Report by European Commission (Eurostat) in close collaboration with the ECB”.

E UROPEAN C OMMISSION (2000), “Communication Structural Indicators”, (COM (2003) 594 final i).

EUROPEAN C OMMISSION (2001a), “European Governance. A White Paper”, COM(2001) 428final i).

E UROPEAN C OMMISSION (2001b), “Communication Structural Indicators”, (COM (2001) 619 final i).

E UROPEAN C OMMISSION (2002), “Communication to the European Parliament and the Council on the need and the

means to upgrade the quality of budgetary statistics”. Com(2002) 670, 27 th November 2002.

E UROPEAN C OMMISSION (2003a), “Education, Training and Growth”, Chapter 3 in the EU Economy 2003 Review, European Economy, No 6, Luxembourg: Office for Official Publications of the EU.

E UROPEAN C OMMISSION (2003b), “Communication on Structural Indicators”, (COM (2003) 585 final i).

G ORDON , R., (2004), “Why was Europe left at the station when America’s productivity locomotive departed?”, CEPR Discussion Paper No 4416.

H ANUSHEK , E. AND D. K IMKO (2000), “Schooling, labour-force quality and the growth of nations”, American Economic Review, Vol. 90, No 5, pp. 1184-208.

K RUGMAN , P. (1996), “Getting Ahead; White Collars Turn Blue”, New York Times, 29 September 1996.

OECD (2002), “Education at a Glance – 2002”, (Paris: OECD).

OECD (2003). “Education at a Glance – 2003”, (Paris: OECD).

M ONETARY C OMMITTEE (1998), “Report on information requirements in Economic and Monetary Union”, Monetary Committee 16th November 1998.

M OOSLECHNER , P. AND M. S CHÜRZ . (1999), “International Macro-Economic Policy-Coordination. Any lessons for EMU? A selective survey of the literature”, Empirica, Vol. 26, pp. 171-199.

M C M ORROW , D.K. AND W . R ÖGER (2002), “Production function approach to calculating potential growth and output gaps – estimates for the EU Member States and the US”, DG ECFIN Economic Papers, No 176, September 2002.

T ARANTOLA , S., R. L ISKA AND A. S ALTELLI (2004), “Structural Indicators of the Lisbon Agenda: Robustness Analysis and Construction of Composite Indicators”, Joint Research Centre of the European Commission (Ispra Italy).

Met de EU Monitor volgt u alle Europese dossiers die voor u van belang zijn en bent u op de hoogte van alles wat er speelt in die dossiers. Helaas kunnen wij geen nieuwe gebruikers aansluiten, deze dienst zal over enige tijd de werkzaamheden staken.

De EU Monitor is ook beschikbaar in het Engels.