** a - -~210 -~~ ~a 200 European Integration, Regional Policy, and Growth European Integration, Regional Policy, and Growth Edited by Bernard Funck and Lodovico Pizzati THE WORLD BANK Washington, D.C. C) 2003 The Intemational Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 Telephone. 202-473-1000 Intemet: www.worldbank.org E-mail: feedback@worldbank.org All nghts reserved 1 2 3 4 05 04 03 The findings, interpretations, and conclusions expressed herein are those of the author(s) and do not necessarily reflect the views of the Board of Executve Directors of the World Bank or the governments they represent The World Bank does not guarantee the accuracy of the data included in this work The boundanes, colors, denominations, and other information shown on any map in thus work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries Rights and Permissions The matenal in this work is copynghted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The World Bank encourages dissemina- tion of its work and will normally grant permussion promptly For permission to photocopy or repnnt any part of this work, please send a request with com- plete information to the Copyright Clearance Center, Inc., 222 Rosewood Dnve, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, www.copynght.com. All other quenes on nghts and licenses, including subsidiary nghts, should be addressed to the Office of the Publisher, World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubnghts@worldbank org. Cover design by Design Farm, a division of Studio 405. ISBN 0-8213-5395-0 Library of Congress Cataloging-in-Publication Data has been applied for. IContents FOREWORD xi CONTRIBUTORS xv 1. Overview I Bernard Funck, Lodovico Pizzati, and Martin Bruncko 2. Public Policies and Economic Geography 19 Philippe Martin 3. Regional Policies and EU Enlargement 33 Michele Boldrin and Fabio Canova 4. Discussion of "Regional Policies and EU Enlargement" by Boldrin and Canova, and "Public Policies and Economic Geography" by Martin 95 Carole Garnier 5. Issues and Constraints of Regional Convergence 107 Alfred Steinherr 6 European Integration, Regional Policy, and the Nonintervention "Hands- off' Approach: Some Comments on the Boldrin and Canova Study 119 Jz*i Blaiek v vi EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH 7. Regional Policy Experience in Southern Italy 129 Fabrizio Barca 8. European Union Regional Aid and Irish Economic Development 135 Frank Barry 9. Does Cohesion Policy Work? Some General Considerations and Evidence from Spain 153 Angel de la Fuente 10. Do Structural Actions Contribute to Reduced Regional Disparities in the European Union? 167 Antoni Castells and Marta Espasa 11. Impact of Regional Aid on Catalonia 177 Alexandre Muns 12. Real Convergence in the Slovak Republic and European Union Regional Funds 183 Martin Bruncko 13. The Role of EU Regional Aid in Economic Convergence in Slovenia 195 Igor Strmsnik 14. Managing European Union Funds in the Candidate Countries: Administrative Organization and Resource Distnbution 201 Vitalis Nakrosis 15. Managing European Union Regional Aid in Central and Eastern European Countries: Do the Countries Need Development Aid? 219 Jan Szomburg 16. Managing Regional Aid in Latvia 225 Janis Krumins 17. Discussion Notes 227 Luis Madureira Pires 18. What Future for EU Regional Policy? 231 Christian Weise 19. Discussion Notes 241 Vasco Cal CONTENTS vii 20. Conclusions of the Conference 249 Carole Gamier INDEX 255 BOX 14-1 Administrative Capacity Requirements for Structural Funds 208 TABLES 1-1 Regional Policy Funds in EU-15, 2000-06 Fiscal Cycle 14 1-2 Regional Policy Funds in Acceding Countries, 2000-06 16 2-1 Regional Disparities in Per Capita GDP within the Member States, 1994-98 21 2-2 Hypothetical Costs 24 3-1 Growth Accounting 51 3-2 Institutional Indexes 53 3-3 GVA and Employment by Branch, Percentage Share 55 3-4 Sectoral Productivity and Wages by Branch, Relative to EU 57 3-5 Indexes of Similarity 63 3-6 Estimated Steady States 68 5-1 Maximum per Capita Income/Minimum per Capita Income 111 5-2 European versus U S. Mobility 116 6-1 The Main Differences between the Czech Regional Policy and the EU Cohesion Policy 125 8-1 Educational Attainment of the Population Ages 25-64, 1998 138 8-2 Relative Infrastructural Levels in the Cohesion Countries as a Proportion of the EU Average, 1985-86 138 8-3 Business Enterprise Expenditure on R&D as a Percentage of Domestic Product of Industry, Relative to the EU Average 139 8-4 Allocation of Structural Funds in Ireland 139 8-5 Educational Attaimment of the Population Ages 25-34, 1998 143 8-6 BERD as a Percentage of Domestic Product of Industry, Relative to the EU Average 144 8-7 Educational Attainment of the Population Ages 25-34, 1998 144 8-8 Education Indicators, 1995 145 8-9 BERD as a Percentage of Domestic Product of Industry, Relative to the EU Average 146 10-1 Key Data on Regional Aid in the EU: Global Amount and Evolution (Financial Perspectives) 168 10-2 Key Data on Regional Aid in the EU Main Beneficiary Countries, 2000-2006 169 10-3 Regional Redistributive Effects of Structural Actions 170 10-4 Reduction in the Interregional Disparities in GDP per Capita Due to Structural Actions 171 viii EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWrH 10-5 Relationship between Net Fiscal Balance and GDP per Capita 172 10-6 Effects of the Structural Actions on Regional Growth, 1995-97 173 10-7 Macroeconomic Impact of the Structural Funds: Comparison of the Results of the Simulation Macroeconometric Models 174 12-1 Regional GDP per Capita at Purchasing Power Standards, 1996-99 184 12-2 Regional Differences in Productivity and Employment 185 12-3 Relative Level of GDP per Capita in Selected Capital Regions, 1999 187 12-4 Gross and Net Income per Person per Month, 2000 188 12-5 Regional Differences in Productivity and Wages 188 12-6 Regional Unemployment Rates, 1997-2001 189 13-1 Slovenia's GDP, 1995-2001 196 14-1 Alternative Systems for the Management of the Structural Funds 204 14-2 Territorial and Sectoral Concentration of the Structural Funds: Distribution of Resources 212 19-1 Summary Statistics for Regions Fallmg below the 75 Percent Threshold 245 20-1 National Growth and Regional Convergence in Ireland, 1991-99 252 FIGURES 2-1 Growth and Regional Disparities, 1994-98 27 3-1 GDP per Capita and Labor Productivity 38 3-2 Saving and Investment Rates 42 3-3 Foreign Direct Investment (FDI) Rates and Total Factor Productivity 45 3-4 Unit Labor Costs 46 3-5 Unemployment and Participation Rates 48 5-1 Structural Funds 112 5-2 Distribution of the Deviation of Regional per Capita Incomes from the EU Average 113 5-3 Important "withm" Movements 114 7-1 GDP Growth and Structural Funds 131 7-2 Nonfarm Employment 131 7-3 Employment 132 8-1 The Convergence Experiences of Greece, Ireland, Portugal, and Spain 136 8-2 Ireland, Portugal, and Spain Unemployment Rates Minus That of the EU15 141 8-3 Distribution of Employment in Foreign-Owned Industry in Ireland by Technological Level 142 CONTENTS ix 9-1 Beta Convergence/Divergence in Relative Income per Capita Induced by Investment in Productive Infrastructures 158 9-2 Cumulative Impact of the 1994-99 Community Support Framework on Factor Stocks of the Objective I Territory 159 9-3 Cumulative Impact of the 1994-99 Community Support Framework on Output and Employment in the Objective I Territory 160 9-4 Cumulative Impact of the Community Support Framework in 2000 161 9-5 Convergence Ratios Induced by the Community Support Framework 161 9-6 Relative Marginal Product of Infrastructures in the Spanish Regions, 1995 162 13-1 Slovenia's GDP, 1992 Fixed Prices, 1987-2001 196 13-2 Slovenia: Regional Disparities 198 18-1 Allocation of Structural Funds in EU 15, 2000-2006 233 18-2 Structural Funds for New Members, 2013 234 18-3 Objective I Population Post-2007, Alternative Criteria 235 18-4 Structural Funds, Various Scenarios, 2007-13 237 20-1 Hermin Simulation Results on the Impact of Structural Funds Programs, 2000-06 250 20-2 QUEST 1I Simulation Results on the Impact of Structural Funds Programs, 2000-06 251 MAPS 1-1 EU Regions 4 2-1 Change in GDP per Head (PPS), 1995-1999 22 5-1 GDP per Head by Region (PPS), 1996 109 5-2 Objective I Areas 110 19-1 GDP per Head by Region (PPS), 2000 244 Foreword The European Union (EU) stands out internationally as one of the political units that has most explicitly and deliberately attempted to reduce regional disparities among its constituents. How effective this effort has been, and continues to be, is a matter of open debate. The current enlargement of the EU to less affluent new members gives rise to a fresh set of questions First, how can the objective of cohe- sion best be advanced in a context in which initial income disparities among the members of the enlarged EU will be greater? Will the accession cause the income of the poorer regions to converge toward EU standards, or, on the contrary, will pre- vailing disparities be exacerbated? What, if anything, can the new members do, and how should the expected EU structural funds be applied to maximize the cohesion objective? How can a new concept of solidarity for the enlarged Union be devel- oped? Enlargement will present new challenges and raise questions as to whether the existing system of mutual support through structural and cohesion funds can be financed in the future, at least at current levels To elucidate these questions, the World Bank, the Bertelsmann Foundation, and the CIDOB Foundation (Barcelona) brought together leading scholars, senior policymakers, and practitioners from existing and new EU member countries, as well as representatives from the European Commission, to a conference in Barcelona in October 2002. This book presents the results of their discussions. Opinions presented by participants have in most cases been made in their personal capacity and do not necessanly reflect the institutions they represent. Drawing on the experience of existing EU members and the latest developments in growth theory xi xii EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH and economic geography, the authors highlight the potential trade-off between pro- moting national growth and reducing relative disparities within countries rather than within the EU as a whole. This book also emphasizes the role of growth poles on the one hand, and investment climate (at the macro and micro levels) and labor market flexibility on the other, in furthenng regional income convergence at the European level. In addition, it evaluates the potential role of EU structural funds either as mere income transfer mechanisms or as key ingredients of growth- enhancing fiscal strategies. Finally, it discusses the nature of institutional arrange- ments that can help bring about one or the other outcome, as well as the challenges that enlargement (and the ever deeper integration of EU factor and product mar- kets) may bring to EU regional policy. In addition to the papers presented in this volume, the conference drew inspi- ration from speeches and remarks by the following Narcis Serra i Serra President, Fundaci6 CIDOB, Barcelona, Spain Joaquim Llimona Secretary-General, General Secretariat of the Presidential Department, Generalitat of Catalonia, Barcelona, Spain Lluis Riera Director, ISPA and Pre-Accession Measures, DG Regional Policy, European Commission, Brussels, Belgium Jozsef Veress Deputy Commissioner, Office for the National Development Plan and EU Funds, Office of the Pnme Mimster, Budapest, Hungary Much of the success of the conference is due to the effective session chairman- ship provided by the following: Martin Brusis Senior Researcher, Bertclsmann Group for Policy Research, Center for Applied Policy Research, Munich, Germany Bernard Funck Sector Manager, Central and Eastem Europe, World Bank, Washington, D.C., United States Roger Grawe Country Director, Central and Eastem Europe, World Bank, Warsaw, Poland IFOREWORD xiii Franz Kaps Senior Advisor, Aid Coordination and Partnerships, World Bank, Budapest, Hungary Cornelius Ochmann Director, Central and Eastern Europe, Bertelsmann Foundation, Guetersloh, Germany Finally, we would like to acknowledge the World Bank Publications team in man- aging the production of this volume. Johannes Linn Narcis Serra i Serra Werner Weidenfeld Vice President, President, Member of the Board, Europe and Fundaci6 CIDOB Bertelsmann Central Asia, Foundation World Bank Contributors Fabrizio Barca Director General in Charge of Regional Policy, Ministry for Economic and Financial Affairs, Rome, Italy Frank Barry Lecturer, Economics Department, University College, Dublin, Ireland Jiri Blazek Lecturer, Department of Social Geography and Regional Development, Charles University, Prague, Czech Republic Michele Boldrin Professor, Department of Economics, University of Minnesota, Minneapolis, United States; Research Fellow, Centre for Economic Policy Research (CEPR), London, United Kingdom Martin Bnrncko Director, Foreign Policy Studies, Institute for Public Affairs, Bratislava, Slovak Republic Vasco Cal General Coordinator of the Cohesion Report, DG Regional Policy, European Commission, Brussels, Belgium xv xvi EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Fabio Canova Professor, Department of Economics, Universitad Pompeu Fabra, Barcelona, Spain; Research Fellow, Centre for Economic Policy Research (CEPR), London, United Kingdom Antoni Castells Professor, Department of Public Finance, University of Barcelona, Spain; Caries Pi 1 Sunyer Foundation, Barcelona, Spain Angel de la Fuente Researcher, Institute of Economic Analyses, Barcelona, Spain Marta Espasa Professor, Department of Public Finance, University of Barcelona, Spain Bernard Funck Sector Manager, Central and Eastern Europe, World Bank, Washington, D.C., United States Carole Garnier Director, Structural Funds and CAP, DG Economic and Financial Affairs, European Commission, Brussels, Belgium Janis KruminAv Minister of Public Administration Reform of the Republic of Latvia, Riga, Latvia Luis Madureira Pires Calouste Gulbenkian Foundation, Lisbon, Portugal Philippe Martin Researcher, Center of Education and Research in Socio-Economic Analysis (CERAS), Paris, France Alexandre Muns Lecturer of International Economic Institutions and European Integration, Escola Superior de Comerc International, Pompeu Fabra University, Barcelona, Spain Vitalis Nakrosis Deputy Director General, National Regional Development Agency, Vilnius, Lithuania Lodovico Pizzati Economist, Central and Eastern Europe, World Bank, Washington, D.C., United States CONTRIBUTORS xvii Alfred Steznherr Chief Economist, European Investment Bank, Luxembourg Igor Strm§nik Secretary of State for Regional Development, Ministry of the Economy, Ljubljana, Slovenia Jan Szomburg Head of the Board, Gdansk Institute for Market Economics, Gdansk, Poland Christian Weise Senior Expert, German Institute for Economic Research, Berlin, Germany CHAPTER 1 Overview Bernard Funck, Lodovico Pizzati, and Martin Bruncko In the 2000-2006 fiscal framework, the European Union (EU) has allocated E2 13 bil- lion (about one-third of the EU budget) to transfers for regional policy These trans- fers amount to E30 4 billion per year, about 0.35 percent of Europe's gross domestic product (GDP) and roughly 3 2 percent of the beneficiary regions' GDP. The EU regional policy was formally established with the purpose of reducing disparities among European regions Funding under this policy comes from a panoply of EU funds, collectively known as "structural funds," some of which predate the policy itself. Regions with GDP per capita below 75 percent of the EU average are the main beneficiaries. (See the annex to this chapter.) As the European Union enlarges eastward and southward, it will absorb a pop- ulation of more than 100 million people with an average income that is roughly half of the previous EU level This enlargement will test existing regional policy frame- works in two different ways. First, it will raise the question of how to ensure cohe- sion in a Union with such vastly disparate initial conditions between regions and countries and, if this objective is to be maintained, what instruments are available to achieve it. Second, enlargement will raise the question of how EU funding can best be applied in light of a surge in the number of claimants and how funding mecha- nisms should be adjusted to achieve that effect. Will enlargement narrow or widen initial disparities between European regions? What (if any) policy instruments do national govemments, regions, and the EU have at their disposal to achieve regional growth7 And what is the role of the European 2 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH regional policy, alongside other instruments, in stimulating convergence and miti- gating potentially negative effects of integration? Can the existing system of mutual support through structural funds be financed at current levels? If not, how can a new concept of solidarity for the enlarged Union be developed? These are some of the questions that the participants at a conference held in Barcelona in October 2002, "Income Convergence and European Regional Policy," sought to address. It should be remembered that the conference took place prior to the completion of the accession negotiations and raised some topics that have subse- quently been resolved. This publication contains the papers presented at that confer- ence. The volume is organized in four parts. The first two parts discuss the impact of economic integration on the disparity of incomc across regions and the role of regional transfers in that context, from a theoretical and empincal perspective; that is, based on the latest insights coming from economic theories as well as on the recent experience of EU members themselves (in contrast, for instance, to that of the United States). The first part looks at these issues from a global perspective, providing theoret- ical arguments and empirical evidence regarding the effects of economic integration on the geographical dispersion of economic activity and on regional growth perfor- mance at the European level, and the role of EU structural funds in that process. In doing so, this part raises the fundamental questions that will perrneate the rest of the book Are the structural funds developmental in nature (do they serve to enhance pro- ductivity), or are they merely redistributive (do they simply reallocate disposable income between contributing and beneficiary regions)? A consensus emerges around two propositions: (1) that regional policy in the EU sense is to a large extent a second-best substitute for labor mobility (and a potentially self-defeating one, if regional interventions further weaken the incentive for labor to move); and (2) that a trade-off exists between these two objectives (developmental and redistnbutive) both at the EU level and in the circumstances of each particular country. The second part of the volume looks at the same questions on the basis of the specific experience of some current EU members. The examples of Spain, Italy, and Ireland are presented in detail. National officials and academics discuss how they have judged the growth/redistribution arbitrage in their own countries and regions and how the EU framework may have helped make the best of it. The third part of the book focuses on the regional policy prospects in the new EU member countnes. In this part, experts and national officials from the Slovak Repub- lic, Slovenia, Lithuania, Poland, and Latvia discuss how they are approaching the same arbitrage, how they are planning to implement regional policy and the allocation of structural funds in response, and how they are setting up their institutional framework for the task. The final part of the book discusses the future of the EU regional policy. A key question is whether, within a given resource envelope, structural funds should be redirected toward the newer members or spread thinner across new and current ben- eficiaries, or whether the envelope itself should be expanded to accommodate all. OVERVIEW 3 What Is a Region? Before getting into the heart of the matter, however, we must clarify how "regional disparities" are defined. Regional statistics in Europe follow the Nomenclature des Unites Territoriales Statistiques (NUTS) classification (see map 1-1). What level of territorial aggregation should be a matter of concern to the policymaker9 In order to make an assessment of regional inequalities tractable, the European Commission has chosen the NUTS-2 level of disaggregation rather than the more aggregated NUTS-I or the more disaggregated NUTS-3, and it is at this level that the largest amount of structural funds is being allocated. The choice of a particular level of disaggregation has great relevance for the view one can form of regional dispanties. As Alfred Stein- herr points out, at the NUTS-2 level, the United Kingdom presents the largest regional income dispanty among current members, while France shows a very homo- geneous distribution. At the NUTS-3 level, however, it is France that exhibits the widest vanation. Why should we be concerned with one measure of regional inequality (NUTS-2) more than another (NUTS- I or NUTS-3)? One reason, explain Michele Boldrin and Fabio Canova, is that idiosyncratic differences between small, undifferentiated geo- graphical entities are unavoidable, and that the matter of regional disparity becomes significant only when regions are large enough (and thus, presumably, have a mini- mal degree of internal differentiation) to become objects of meaningful macroeco- nomic inquiry. At a higher level of disaggregation, we will find large discrepancies in output per capita. However, they would be quite normal, simply reflecting the dif- ferences in local natural endowments: for instance, a district that, thanks to its ideal soil, specializes in agricultural production would not generate as much output per capita as a neighboring district with a heavy concentration of capital-intensive industries, even though living standards might be identical In this respect, Boldrin and Canova call attention to the large differences that exist even between NUTS-2 regions, which range from 100,000 to 10 million inhabitants They argue that eco- nomic convergence should be targeted for larger areas, if any This would suggest that the NUTS-I level might be the only one that is pertinent for policy and plan- ning purposes, and in the case of the smaller countries (such as Ireland, Portugal, and many Central and Eastern European candidates), it would even correspond to the national boundaries The Impact of Regional Policy on Income Convergence This being posited, we note that the European Union has set itself the goal of pro- moting economic and social cohesion by reducing disparities between the levels of development of its various regions, however defined, and the backwardness of the least-favored regions Why should this require regional transfers? After all, neoclas- sical models predict that increasing trade and economic integration should sponta- 4 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH MAP 1-1 EU REGIONS EU REGIONS - NUTS LEVEL 1 NUTS LEVEL 2 Source Eurostat fl,~~~~~~~ C r,'S 1 KE3 Z l S N~~~~~~~~~~KLOMMR IBRD 32348 OVERVIEW 5 neously fuel convergence. In the presence of diminishing marginal returns on factor inputs, capital investment should flow to less developed countries, which tend to be relatively undercapitalized and where, consequently, capital can earn higher profits Increased investment should translate into improved productivity, which should in turn lead to higher income levels and spontaneous real convergence to EU levels Still, we cannot fail to note that progress in EU integration has gone hand in hand with an increasing emphasis on interregional transfers. In fact, the fear that the advent of the Single Market and of the Economic and Monetary Union would exacerbate intra-European disparities contributed to the addition of a new transfer mechanism to the existing panoply of structural funds, the so-called Cohesion Fund. Drawing on the tenets of "economic geography theory," Philippe Martin explains why this might be the case Firms benefit from agglomeration effects (arising from the quality of infrastructures, concentration of human resources, etc.) and economies of scale. When markets open up and transaction costs are reduced, firms naturally tend to congregate in areas where such agglomeration effects are strongest or they can best exploit economies of scale (i.e , closer to main consuming markets). ' If labor markets are ngid, concentration of activity may lead to greater disparities in unem- ployment across regions than is politically tolerable. Hence the role of regional pol- icy to bringjobs to the people, if the people do not go thejobs As Martin indicates, this pattern of activity concentration is not specific to Europe. In the case of the United States, economic activity is highly concentrated in a few regions, although much less than in past decades. Due to American labor mobility, concentration of production can more easily attract the bulk of the pop- ulation. Therefore, when we look at income per capita, regions with high GDP are also very dense in population, and present GDP per capita is similar to that in regions with low GDP and low population. According to Martin, a process of regional concentration is under way in Europe today However, Europe does not have the same labor mobility as the United States, and, therefore, economic inte- gration may lead to increased regional discrepancy in per capita GDP. If there is underutilization of economic potential in the periphery because of labor market rigidity, structural funds may provide the incentive for firms to relocate enterprises to the depressed peripheral regions. Martin adds, however, two important words of caution First, attempting to coun- teract firms' natural tendency to concentrate (at least in the initial stages of integra- tion) necessarily leads to a loss of efficiency, reducing the overall growth potential accordingly. Conversely, one can see that if labor-market concerns could be addressed differently, the scope would exist for maximizing growth by stimulating, rather than dampening, agglomeration effects, including, for instance, through a deliberate concentration of infrastructures within "growth poles." One example of this approach is the deliberate effort to improve transport infrastructure in the North of France, which made this region competitive with neighboring wealthy European regions, spurring a concentration of economic activity there. 6 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Martin's second caveat is that investing, for example, in public infrastructure may have different effects in different regions. For instance, investments aiming to reduce trade costs, such as those in new highways, may have exactly the opposite effects. If economic integration has not advanced sufficiently, decreasing trade costs may not induce firms to establish their production in the economically disadvantaged regions-even if those regions may offer lower labor costs. In contrast to what hap- pened in the North of France, highway construction in southern Italy may well have accentuated its decline by encouraging firms established in this region (or that would have otherwise located in it) to move (or keep) their production outside of it in order to benefit from economies of scale closer to larger markets. Michele Boldrin and Fabio Canova challenge this view, calling into question whether the agglomeration effects Martin envisions exist at the macroeconomic level in the real world. According to their research, such effects are not observed empiri- cally at any reasonable level of spatial disaggregation. They also argue that European structural funds have had no observable impact on regional growth and economic convergence. In the past 20 years, they claim, despite the significant increase in regional funds invested, there was no visible impact on economic convergence within countries. As to convergence across countries, Boldnn and Canova insist, it is national policies that reduce taxes and inefficient public spending, liberalize labor markets, attract foreign direct investments, and minimize income support transfers that have led to increased growth. After all, the authors stress, it is not high invest- ment per se that causes growth. Rather, when policies create appropriate conditions for growth, investment follows. The most striking example the authors cite to sup- port their argument is East Germany, which received E571 billion from the German government between 1991 and 1997. Despite having full access to the EU markets, East Germany has not converged. Other transition economies have fared better in the absence of regional transfers by undergoing the necessary structural change and hav- ing new investment directed toward productive sectors. If this is the case, Boldrin and Canova note, EU regional policy can be rational- ized only as an exercise in income redistribution or by political economy arguments pertaining to the need for EU-level politicians and bureaucrats to secure political sup- port and legitimacy for the European enterprise, albeit at some efficiency cost-a motivation expressed unapologetically by Barca and Cal later in the volume. Whatever the case might be, in the absence of economies of scale or the type of external effects described by Martin, efficiency considerations would argue in favor of reducing the scope of the related transfers as much as possible and targeting them to the lower-income region rather than toward potentially more affluent "growth poles," as earlier suggested. To the extent that they remain in place, income support programs should target job-seekers (as in the Netherlands), and structural funds should create employment opportunities to enlarge labor force participation (as in Ireland). In general, Boldrin and Canova suggest focusing regional funds on infra- structure (transport, communication, power, water, education) and not on subsidies OVERVIEW 7 to small business development, as that may create the wrong incentives. Instead, a favorable national policy framework should facilitate the business environment Carole Gamier retorts that Boldrin and Canova's policy recommendations are (lenved from excessively limited empirical data. Indeed, it is only since the late 1 980s that structural funds have had an efficiency-oriented design. Because supply-side effects take time to materialize (as does investment in infrastructure aiming to spur productivity), it is too early to assess the impact of structural funds on long-term con- vergence. While regional policy based on the small and heterogeneous NUTS-2 is being cnticized as a target for regional convergence, she submits that structural funds can make a positive contnbution to national convergence. In line with Martin's thesis, she attributes this effect to a trade-off between national convergence (national GDP per capita catching up with EU average) and regional-level convergence (regional GDP per capita converging toward national averages). Whether the trade-off identified by Martin and Gamier has actually been prop- erly exploited is a different matter. A more recent paper, Midelfart-Knarvik and Over- rnan (2002) suggests that Boldrin and Canova's findings (no empirical evidence of clivergence) are not necessarily irreconcilable with the view that agglomeration effects exercise a powerful influence on firms' location decisions. Contrary to Boldrin and Canova, however, Midelfart-Knarvik and Overman (2002) finds that regional policy has been effective in shifting firms' location where it is less efficient, thereby poten- tially offsetting the impact of agglomeration effects (and explaining the lack of observ- able divergence) In other words, the regional policy's initial emphasis on dispersing economic activities would have "succeeded," but at the cost of overall growth. This begs the question as to why regional transfers would be needed at all. WVould there be a need to remedy a shortage of private investment due, for example, to national savings being insufficient or to limited access to foreign borrowing? Are there microeconomic constraints to "good projects," public or private, finding financ- ing9 Alfred Steinherr doubts that such macroeconomic or microeconomic constraints still exist now that EU capital markets have opened up-at least as far as the current members are concemed (the accession countries might present a different case). Therefore, instead of a lack of public investment, Steinherr identifies three types of extemalities and labor market rigidity as the main source of regional income dif- ferences First, technological externalities lead firms to locate where there is already economic activity and a developed market (Martin's argument). Second, as a result of excessive movement of capital and labor to more productive areas, pecuniary externalities arise from falling wages and increasing consumer prices that may emerge from labor movements. Third, when a region does not reach a minimum threshold of economic activity, it lacks the required price information and business knowledge, and this leads to a coordination problem for the development of a net- work of services and intermediate goods suppliers Finally, labor market rigidity, understood as low degree of labor mobility and skill differences, exacerbates those three factors of regional imbalances Steinherr points to inflexible housing markets 8 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH (high registration taxes and other transaction costs), lack of cross-regional job infor- mation, and language barriers as the main sources of a low degree of labor mobility. According to Steinherr, these, and not a lack of investable funds, are the problems that need a remedy. That leaves us with only the political economy argument for "doing something" in the name of Europe to justify regional aid. Jin Blazek concurs with Boldrin and Canova on the dangers of creating a "dependency mentality" and agrees with their insight on using larger territorial units for regional fund eligibility However, Blazek wams against their nonintervention approach, given the risk that the potential frus- tration with limited opportunities on the part of the population in post-communist countnes may put into question the market economy and parliamentary democracy Regional "activism" would obviate this risk. Country-Speciflc Experiences In the chapters focusing on specific country experiences, we find our contributors generally more sanguine about the impact of EU regional aid than the previous dis- cussion may have led us to believe. Overall, we can divide the impact of structural funds into demand- and supply-side effects. On the demand side, the increased spend- ing will have a definite relevance for aggregate expenditures. Whether regional funds are serving a purely redistnbution purpose or are efficiently invested, in the short run they will have a positive impact on aggregate demand. On the other hand, if these funds are successfully invested, they will increase the production potential of the recipient region irrespective of future funding. These are the supply-side effects that will increase income and employment levels in the long run as well, without neces- sitating the inflow of additional funding. Frank Barry makes the additional distinc- tion between "direct" and "indirect" effects of regional funds. Besides directly adding to the productive capacity of the beneficiary economy, Barry suggests that the direct impact could be complemented by indirect supply-side effects if, past their initial impact, the investments lead to a sustained increase in the productivity growth rate The magnitude of supply-side effects is hard to quantify and presumably initially much lower than that of the demand effects, which are close to the amount of actual spending. Without any supply-side effects whatsoever, the funds would de facto amount to a pure income transfer, increasing expenditures in beneficiary regions at the expense of contributing regions, and perhaps also of overall growth, if the related transfers are financed by distortionary taxes Most authors would agree that the demand-side effects are positive. Looking at a wide geographical coverage, Castells and Espasa find clear evidence of the redis- tribution effect of regional aid, as the demand-side effect of these transfers induces a short-term reduction in regional income disparities. Positive as the demand-side effects may have been, has EU regional aid achieved supply-side effects on growth beyond the natural impact of redistribution? De la OVERVIEW 9 Fuente contends that in Spain, direct supply-side effects of structural funds can also be observed Bany's estimate of the impact of EU regional aid on the Irish conver- gence is also positive, though surprisingly modest, particularly on the supply side. Was the potential of the structural funds insufficiently carried out? De la Fuente's analy- sis indicates that, in the case of Spain, regional aid could have had much higher impact had the government attached a greater weight to efficiency considerations rather than to redistribution. It may be argued that this redistnbutive bias is actually preordained in the regional targeting of structural funds to selected NUTS-2 regions. Certainly, when relieved from that constraint and allowed to allocate regional aid irrespective of regional income, the Spanish government chose to priontize very different regions. When it came to allocating the (regionally untargeted) Cohesion Fund resources, the comparatively more affluent Catalonia came first, followed by Andalusia and Madrid. Of course, some projects financed in Catalonia may be justified as having interregional benefits (e g , the construction of a high-speed train route and the expan- sion of the port of Barcelona). But substantiating Steinherr's remarks on the risk of regional aid displacing private market-based financing, Alexandre Muns notes that many projects cofinanced with EU regional aid would have been undertaken even without EU regional funds. This being said, a number of authors stress that the main benefits of the EU regional policy lie elsewhere Fabrizio Barca, for instance, underscores the impor- tance of the peer review/pressure it involves for improving the quality of expendi- ture policies and institutions Regional policy puts a strong emphasis on planning, partnership, monitoring, evaluation, and control requirements in the use of EU funds. In order to draw upon these funds, national administrations must have a cer- tain set of institutions and follow strictly defined policies at both national and regional levels. All these requirements are designed to ensure the effectiveness and efficiency of public interventions funded by EU money. Structural funds have thus been instrumental in increasing efficiency in public administration Barca recounts, for instance, how the discipline created by the EU framework helped the Italian gov- ernment deal with the notoriously inefficient Cassa del Mezzogiorno, a centralized Italian state agency that managed subsidies and infrastructure. Following decades of mismanagement, the agency was terminated in 1993. The results, Barca says, are plain to see: savings in southern Italy have increased, unemployment has dropped, and self-employment is surging. Barry similarly suggests that the indirect impact of regional aid may have been the most important one, particularly its contribution to the quality of Ireland 'sfiscal adjustmnent. Up to the mid-1980s, the country had been struggling with a debt crisis resulting from pro-cyclical fiscal expansions Financing rising public spending with high taxation had proved unsuccessful, as workers responded with higher wage demands. With the increase of structural funds in 1989, Ireland was able to bring down its deficit without compromising its public investment program. EU transfers helped relax the budget constraint and allowed lower tax rates, which in turn attracted 10 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH foreign firms (primarily thanks to lower corporate taxes), with a direct impact on growth. According to Barry, the regional policy framework also proved instrumen- tal in bringing about an improvement in the composition of Ireland's public expen- diture, with greater emphasis on communication infrastructure and education. Favorable labor market conditions, however, proved key to the success of the Irish formula. Barry emphasizes the importance of labor marketflexibility, remind- ing us how Portugal, with a different industrial strategy focused on low-tech manu- facturing (as opposed to the Irish high-tech), is also converging to EU standards. With flexible labor markets, new investment increases the demand for labor, which in turn translates into increased employment and income growth. Conversely, in labor mar- kets more dominated by insiders, such as in Italy or Spain, the same labor demand increase will translate into higher wages (for insiders), and thus into less employment and growth. National policies, of course, "naturally" generate less rigidity in small and comparatively uniform countries like Portugal and Ireland than in larger, more differentiated ones like Spain and Italy. With similar national wage policies, and thus comparable labor costs in both rich and poor regions, it is hardly surprising that firms would locate production in industrialized areas, leaving the poorer regions with higher unemployment. Lessons for Accession Countries How does that experience apply to the countries that are now joining the EU? How should they spend the huge sums that will be coming from the structural funds? Should they target efficiency or redistribution goals in their allocation decisions? How should they manage the structural funds to maximize their intended impact? Two main mes- sages come out of this book: First, regional policy should be used pnmarily accord- ing to efficiency criteria, with growth-oriented domestic policy ensuring the diffusion of convergence across regions. Second, if regional aid is to be disbursed, it should be used to finance essential infrastructure rather than for transfers to private and public enterprises in the name of attempting to support small business development. As for the institutional arrangements for managing structural funds, they may be more or less decentralized, according to the size and degree of decentralization of the recipi- ent country itself, what is important is that they be firmly integrated as part of gov- ernment's overall fiscal strategy (as described above in the case of Ireland) and in the mainstream of public resource management. The first point is made by Martin Bruncko in the context of the Slovak Repub- lic. While recognizing that substantial regional dispanties exist, particularly between the booming capital of Bratislava and the eastern part of the country, Bruncko argues that these disparities are not particularly large by European standards, and the dif- ference in regional output per capita is not widening. The truly worrisome disparity is in unemployment, as the growth pole of Bratislava, prospering with small and medium enterprises, is in direct contrast with the declining agriculture and collapsed OVERVIEW 11 military industry of the eastern part of the country. Bruncko also observes that regional wage variation is less pronounced than productivity variation would war- rant. It is no surprise that wage/productivity ratios lead foreign investments to areas like Bratislava, which also happens to be closer to major European markets and trans- port arteries. It would be a mistake in this context to bypass the emerging Bratislava growth pole as recipient of structural funding (as the Slovak government proposes), since from an efficiency perspective it promises the highest return on investment, at least in the imtial years of accession. What would facilitate the diffusion of investment and growth to more backward regions is not dispersing public spending efforts, Bruncko suggests, but improving wage competitiveness in poorer regions and reforming social welfare to stimulate labor mobility, while reducing trade costs over time in periph- eral regions through a well-calibrated transportation infrastructure. Even so, the magnitude of the investment required may sometimes exceed the financial capacity of the country alone. If the benefits of the proposed investments also accrue largely beyond a country's borders, there might be a good case, Strmsmik submits, for the international community (that is, the EU) to fund the investment, irrespective of narrowly defined national benefits and financial limitations Slove- nia's location makes it an important link for Europcan infrastructure integration. While the interregional benefits might make the development of this infrastructure worthwhile for Slovenia's neighbors in the first place, the magnitude involved would strain budgetary resources and make it difficult for a country of Slovenia's size to meet the Maastricht criteria on fiscal deficits. Hence the need, according to Strminik, for the country to have access to structural funds, irrespective of its income level (which may exceed the eligibility threshold), not so much in consid- eration of Slovema's priorities and requirements per se, but on the basis of larger cross-border returns. In practice, the allocation of structural funds will be determined in part by the nature of the institutional arrangements set up for the tasks. Integrating structural fiunds in the overall management of national budgets would typically allow govern- ments to deploy them better in the framework of a consistent fiscal strategy than an extra-budgetary operation would. On the continuum that links "redistribution-based" and "efficiency-based" investment decisions, a more decentralized setting is more likely to err on the redistributive side, while a more centralized one may be more adept at marshaling resources toward maximizing national growth. Conversely, a more decentralized approach might be more effective at mobilizing local partner- ships. Vitalis Nakrosis outlines the pros and cons of the institutional choices involved. integrated versus unintegrated, centralized versus decentralized arrangements. After an initial hesitation, countries have generally adopted the integrated approach. As to the second alternative, Nakrosis concludes, in line with the argument made at the out- set about the existence of a minimum size threshold for regional targeting, that an integrated-centralized system would be the most cost-effective option for a country the size of Lithuania. 12 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH In contrast, Janis Krumins defends the decentralized approach. Krumins argues that in the case of Latvia's national growth can currently be seen only in the biggest cities, while the rural areas are in continuous decline. In areas with no foreign invest- ment and nojob creation, the low income levels do not permit a large enough tax base to support the maintenance of essential infrastructure, much less its development. It is to these lagging and depressed territories that, in his view, regional aid should be directed. And the best way to make sure that this happens is to decentralize the funds away from the capital. After considering the idea of decentralized management, Jan Szomburg explains, Poland has chosen a mixed approach, with about 80 percent of funds being managed under national programs and the remainder being allocated as follows: 80 percent allocated to all regions in proportion to their population, 10 percent to regions with GDP per capita below 80 percent of national average, and 10 percent to provinces with an unemployment rate 50 percent above the national average. Szomburg antic- ipates, however, that Poland will shift to a more fully decentralized approach in the next budget cycle (2007-2012). Luis Madureira Pires observes that, although the range options available for managing regional aid varies a lot between large regionalized countnes such as Poland, and smaller, more centralized ones, the European Commission has typically favored more centralized approaches-sometimes, in his view, to the detriment of local efforts to build more modem and more responsive govemance frameworks. Similarly, while countries may take different views of the arbitrage they want to make between redistribution and efficiency considerations in utilizing regional aid, the Structural Funds philosophy has, since 1988, increasingly stressed the need to reduce the development gap between poor member states and the EU average rather than to decrease intemal regional disparities, the latter being regarded more and more as a domestic issue. The Future of EU Regionall Policy Arguing that structural funds have had no impact on growth but generate distorting effects, Boldrin and Canova, as we have seen, called for their termination at the end of the current EU budget cycle (in 2006), or at least for lowering eligibility to regions with 50 percent of EU average GDP per capita (which would leave out virtually all current beneficiaries in Westem Europe, as well as Slovenia, the Czech Republic, and perhaps Hungary), with funding to be phased out by the next six-year budget cycle. Christian Weise makes the case for reform, instead. In his view, reform will be needed to resolve two principal-agent problems embedded in the current regional policy framework. First, net paying countries do not directly control the allocation of transfers, relying on the European Commission to ensure the appropnate implemen- tation of structural funds. Second, the European Commission is itself subject to a pnncipal-agent problem, as it cannot implement the desired policy itself and must OVERVIEW 13 rely on the receiving countries, which may or may not pursue the same ultimate goals. With enlargement, Weise warns, these problems will worsen, as there will be more receiving countnes to potentially take advantage of such pnncipal-agent problem and the regional administrative structures in new member countries are likely to be weak. Weise recommends, therefore, heightening the incentives for recipient countries to adhere to the spirit of a growth-oriented cohesion policy; for instance, through a higher degree of national cofinancing, a greater reliance on loans instead of grants, and perhaps even policy conditionality. Moreover, Weise favors leaving the current eligibility threshold for regional aid (expressed as a percentage of the average EU income) untouched after enlargement, even if that means weaning many current ben- eficiary regions in Western Europe from EU support (because the absolute level of the threshold would decline as new lower-income entrants bring down the average EU income). This would be in line, he says, with the fundamental rule of the EU cohesion policy, which is to concentrate support on the needy. In his reply, Cal argues that this approach is politically impractical. Limiting finding to the extent proposed by Weise would undermine the political support for the entire idea of regional policy (as fewer would benefit from it), with the ultimate result that even the money for less developed areas would disappear. As part of the enlargement process, Cal highlights how EU regions will basically be divided into three groups: (1) per capita GDP below 45 percent of average (most of the Central European accession countries); (2) per capita GDP above 120 percent of average (rnost of the current member states); and (3) per capita GDP around 80 percent of average-previous cohesion member states (Spain, Portugal, and Greece) as well as some new entrants (Czech Republic, Slovenia, and Cyprus). With enlargement, EU regional policy will thus automatically focus more on less developed areas. But also as a result, regions encompassing as many as 18 million people may lose their pre- vious eligibility, thereby potentially eroding the political constituency for the system. The best way to obviate this risk, Cal suggests, is for EU regional aid to continue to flow to the not-so-underdeveloped regions of the enlarged EU. Conclusions Summng up the proceedings, Carole Gamier stresses how limited our understand- ing of the determinants of long-term growth remains, particularly at the regional level Therefore, our understanding of the potential and actual impact of EU regional policy in this process, particularly in the presence of such unquantifiable benefits as the improvement in public administration efficiency noted by some of the speakers, is also limited What is clear, however, is that simply injecting resources is not suf- ficient to ftuel sustained regional growth. What counts is the policy framework into which those resources are injected and the strategy applied in deploying them. Gamier lists three essential ingredients to such a successful framework/strategy for regional convergence at the EU level. First, financial stability is a prerequisite for 14 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH growth, and national-level macroeconomic policies need to be supportive of that objective. Second, convergence policies need to be pursued at the appropriate terri- torial level. A trade-off appears to exist between maximizing growth at the national level and reducing disparities between smaller territorial units within individual countries. This is not necessarily a zero-sum game, however. In the case of Ireland, for example, regional disparities actually increased during the economic takeoff, as some regions grew faster than others; even the lagging regions, however, converged to European averages. The third ingredient for a successful strategy concerns the spe- cific types of investment to be supported. There is no formula for this. Investment strategies need to be grounded in a careful analysis of regional specificities, bottle- necks, and potential competitive advantages. Annex: What is EUJ Regionall PoRcy? The European Union's regional policy is based on financial solidarity, inasmuch as part of member states' contributions to the Community budget goes to the less pros- perous regions and social groups.2 For the 2000-2006 period, these transfers will account for one-third of the Community budget, or E213 billion (see table 1-1 below): TABLE 1-1 REGIONAL POLICY FUNDS IN EU-1 5, 2000-06 FISCAL CYCLE (millions of E) Regional policyfunds Cohestonfund Structuralfunds EU-15 Objective I Objective 2 Objective 3 Spain 56,205 11,160 45,045 38,096 2,651 140 Germany 29,764 29,764 19,958 3,510 4,581 Italy 29,656 29,656 22,122 2,522 3,744 Greece 25,000 3,060 21,940 21,000 0 0 Portugal 22,760 3,060 19,700 19,029 0 0 United Kingdom 16,596 16,596 6,251 4,695 4,568 France 15,666 15,666 3,805 6,050 4,540 Ireland 4,200 720 3,480 3,482 0 0 Netherlands 3,286 3,286 123 1,752 1,686 Sweden 2,186 2,186 722 406 720 Finland 2,090 2,090 913 489 403 Belgium 2,038 2,038 625 433 737 Austna 1,831 1,831 261 680 528 OVERVIEW 15 * E195 billon will be spent by the four structural funds (the European Regional Development Fund, the European Social Fund, the Financial Instrument for Fisheries Guidance, and the Guidance Section of the Euro- pean Agncultural Guidance and Guarantee Fund); * E 18 billion will be spent by the Cohesion Fund. The structural funds concentrate on clearly defined priorities: * 70 percent of the funding goes to regions whose development is lagging behind. They are home to 22 percent of the population of the Union (Objective 1). * 11.5 percent of the funding assists economic and social conversion in areas experiencing structural difficulties. 18 percent of the population of the Union lives in such areas (Objective 2). * 12.3 percent of the funding promotes the modernization of training sys- tems and the creation of employment (Objective 3) outside the Objective I regions, where such measures form part of the strategies for catching up. There are also four Community Initiatives seeking common solutions to specific problems. They spend 5.35 percent of the funding for the Structural Funds on * cross-border, transnational, and interregional cooperation (Interreg III); * sustainable development of cities and declining urban areas (Urban Il); * rural development through local initiatives (Leader +); and * combating inequalities and discrimination in access to the labor market (Equal). T here is a special allocation of funds for the adjustment of fisheries structures out- side the Objective I regions (0.5 percent). There are also provisions for innovative actions to promote and experiment with new ideas on development (0.51 percent). The structural funds finance multiyear programs that constitute development strategies drawn up in a partnership associating the regions, the member states, and 16 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH the European Commission, taking into account guidelines laid down by the Commis- sion that apply throughout the Union. They act on economic and social structures to o develop infrastructure, such as transport and energy; o extend telecommunications services; o help firms and provide training to workers; o disseminate the tools and know-how of the information society. Development initiatives financed by the structural funds must meet the specific needs identified on the ground by regions or member states. They form part of an approach to development that respects the environment and promotes equal opportunities. Implementation is decentralized, which means that it is mainly the responsibility of the national and regional authorities. One particular fund, the Cohesion lFund, provides direct financing for specific projects relating to environmental and transport infrastructure in Spain, Greece, Ire- land, and Portugal, as these are still inadequate. The instrument for Structural Policies for Pre-Accession (ISPA) provides assistance along the same lines to the 10 Central and Eastem European countries that have applied for Union membership. (See table 1-2 below.) Irrespective of the type of assistance, these instruments complement but do not replace national efforts. TABLE 1-2 REGIONAL POLICY FUNDS IN ACCEDING COUNTRIES, 2004-06 (millions of E) Regional polic) funds Cohesion fund Structuralfunds Acceding countries Objective I Objective 2 Objective 3 Poland 11,369 3,733 7,635 7,321 0 0 Hungary 2,847 994 1,853 1,765 0 0 Czech Republic 2,328 836 1,491 1,286 63 52 Slovak Republic 1,560 510 1,050 921 33 40 Lithuania 1,366 543 823 792 0 0 Latvia 1,036 461 575 554 0 0 Estonia 618 276 342 329 0 0 Slovenia 406 169 237 210 0 0 OVERVIEW 17 Notes 1. With high transaction costs (bamers to trade, transportation costs), it is more convenient to disperse production in each country (and be closer to consumers). With low transaction costs, it is preferable to produce in areas with access to larger markets, leading to a concen- trated agglomeration of production :2. The information in the annex is from the European Commission. ]Bibliography Midelfart-Knarvik, Karen H., and Henry G. Overman. 2002 "Delocation and European Integration: Is Structured Spending Justified?" Economic Policy 35(October):322-59. CHAPTER 2 Public Policies and Economic Geography Philippe Martin Regional inequalities are an inescapable fact of Europe's economic geography. Close to one-fifth of Europeans live in regions eligible for aid under Objective I of the Structural Funds, the principal regional policy instrument of the European Union (EU). The criterion for receiving this aid is a per capita income 75 percent lower than the European average. If such a criterion were applied to the United States, only two states (Mississippi and West Virginia), representing less than 2 percent of the Amer- ican population, would qualify. The disparities in the average unemployment rate are also very wide in Europe but practically nonexistent in the United States. They rep- iesent disparities not only between countries but also between regions within each country: for the five "big" countries (France, Germany, Great Britain, Italy, Spain), Ihe unemployment rate in the most affected region is at least 7 points higher than in the least affected one. As a reaction to these disparities, the EU has been devoting an increasing share of its budget to regional policies. The Structural Funds and the Cohesion Fund today represent over one-third of the community budget. These transfers are of consider- able macroeconomic significance over the period 1991-94 they represented 3.5 per- cent, 3.3 percent, 2.4 percent, and 1 5 percent of gross domestic product (GDP) for Greece, Portugal, Ireland, and Spain, respectively. For the period 2000-2006, their impact will be less, in particular for Ireland. In terms of investment, they play a deci- sive role: respectively for those four countries, 15 percent, 14 percent, 10 percent, and 6 percent of total investment is financed out of community resources. 19 20 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Such intervention has had contrasting effects. Since 1989, there has certainly been convergence among the European countries, Ireland being the most successful and the most spectacular example. However, convergence among the regions within each country has not come about; indeed, in most countries, regional disparities have increased. Regional Convergence and Divergence in Europe A comparison between the economic geography of the EU and of the United States is instructive on more than one count. From the standpoint of regional income gaps, we have already noted that these are much wider in Europe than in the United States. But this larger disparity does not reflect a greater spatial concentration of economic activ- ities: on the contrary, in Europe close to half of industrial employment is concentrated in 27 regions representing 17 percent of the continent's territory and 45 percent of its population. In the United States, half of industnal employment is concentrated in 14 states representing only 13 percent of the entire country and 21 percent of its pop- ulation. Midelfart-Knarvik and others (2000) show that, even taking into account the fact that industry in Europe is less concentrated than in the United States, most of the sectors are more geographically dispersed in Europe than in the United States. How- ever, over the past 20 years the European countries have become increasingly spe- cialized, and the structures of their industries have become more and more different. The empirical studies perforned by Brulhart (1998), Brulhart and Torstensson (1996), and Amiti (1998) strongly suggest that a process of spatial concentration is under way in Europe today, going back in particular to the 1 980s, when good progress was made with trade integration. Why has the greater spatial concentration of economic activities in the United States not engendered a larger income gap among the states? The first reason is the marked mobility of economic agents in the United States, a phenomenon that does not exist in Europe In the United States, mobility among states is much greater than mobility not only among European countnes themselves but also among the regions within each European country. This strong American mobility explains why the phe- nomenon of spatial concentration of economic activities in the United States has not been accompanied by a process of per capita income divergence among its states. The fact is, when workers follow mobile capital (physical or human) from regions in decline to regions experiencing growth, the problem of spatial equity becomes much less acute. Although certain regions are being emptied of their economic activities, per capita income does not diverge among the different regions. The emigration of work- ers from declining regions to growth regions makes it possible to reduce competition among workers in the former and to increase it in the latter. Such migrations are thus the principal force for adjusting regional inequalities in the United States. This cer- tainly explains why, in the United States, the question of regional or spatial planning policies has never become as significant an issue as it is in Europe. To the contrary, it PUBLIC POLICIES AND ECONOMIC GEOGRAPHY 21 is because the workers (often the most disadvantaged) are also less mobile in Europe that economic geography has taken on a political dimension only in the old continent. Another phenomenon noted in the introduction is the different way inequalities have developed among the European countries and among the countries' own inter- nal regions Table 2-1 illustrates the development of those disparitie,s measured by the standard per capita GDP deviation for the Nomenclature des Umtes Territoriales Statistiques (NUTS 2) regions. Only two countries, France and Germany, showed a lessening of inequalities over the period. In three countries-Austria, Belgium, and lPortugal-they remained somewhat constant. In the seven other countries for which regional data are available, inequalities increased The two last lines of the table also show that while inequalities among countries diminished, those among the countries' own internal regions on average increased. Map 2-1 shows GDP growth rates per region and gives a clear picture: it is easy to r ecognize the areas that have "triggered" their country's growth. For Spain, Catalonia and Madrid; for Portugal, the Lisbon region; for England, the southeast, for Sweden, Stockholm, for Finland, Helsinki; and for the Netherlands, the Amsterdam region N4ore detailed studies (Duro 2001) have shown that up to the mid- I 980s, income in- equalities among member states represented half of the inequalities among the Euro- pean regions, and inequalities among regions within each state represented the other half Since then, inequalities among states have diminished by 25 percent, but regional inequalities within the states have increased by 10 percent. As a result, the majority TABLE 2-1 REGIONAL DISPARITIES IN PER CAPITA GDP WITHIN THE MEMBER STATES, 1994-98 Member state 1994 1995 1996 1997 1998 Austria 28 1 30 8 30 2 29 2 27 8 Belgium 25 9 25 3 25 7 25 7 25 7 Finland 17 1 18 3 21.2 22 0 246 France 30 8 28 2 28 2 27 0 26 5 Germany 313 26 7 26 7 26 5 26 8 Great Bntain 18 3 314 317 33 4 33 9 Greece 7 8 104 102 101 10 2 Italy 25 5 286 287 27 8 27 6 Netherlands 108 134 14 3 15 4 15 8 Portugal 138 135 133 140 142 Spain 15 9 17 1 177 184 19 1 Sweden 110 13 1 14.0 162 17 1 EU15 (by member state) 12 7 125 119 115 112 EU 15 (within member states) 23 0 24 5 24 7 24 8 25 0 Note Standard deviation of index EU 15 = 100 Source European Commission 2001 22 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH MAP 2-1 CHANGE IN GDP PER HEAD (PPS), 1995-1999 CHANGE IN GDP PER HEAD (PPS), 1995-1999 Change in Index, (EU-27 = 100) score W-50--25 D10-50 W Ž100 Source Eurostat X O S ssLOMETERs IBRD 32279 9 Canary Is (Sp) French . ypRUS Azores (Port) Guiana (Fr) t-Cire GuadeIups Mtrbnrqua Runton Madeira (Fr) (Fr) (Fr) lM dolra (Port) F'UBLIC POLICIES AND ECONOMIC GEOGRAPHY 23 of regional inequalities in Europe are explained by inequalities within the countries. Thus, Europe is experiencing a process of convergence among countries at the same time as a process of divergence among the countries' own regions all of the conver- gence among the regions in Europe at the European level Is thus explained by the convergence among countries. A similar development in spatial polarization may be described for unemploy- ment. Overman and Puga (2002) show that since the mid- 1 980s, regions starting out with a low or high unemployment rate have not shown much change in their relative situations. Regions with intermediate unemployment rates, on the other hand, have moved toward the extremes The authors interpret this result as an effect of the spatial polarization of economic activities due to economic integration. They show that the fate of the regions in terms of unemployment is linked much more closely to the results of the neighboring regions (whether or not they belong to the same country) than to those of the respective country itself New Economic Geography The introduction of economies of scale and of transaction costs may explain why regions with no obvious comparative advantage in certain activities can become cen- ters of production of those activities. A model of the underlying mechanisms was introduced by Krugman (1991), who led the way for the so-called new economic geography. The central finding of this literature is that the diminution of transaction costs on trade may engender a concentration of economic activities in certain regions that have better access to the large markets even if they do not have the lowest pro- duction costs This spatial concentration is advantageous because of the existence of economies of scale conducive to limiting production locations, and it is made possi- ble by commercial integration, which, while reducing transaction costs, does not oblige enterprises to be located close to all their consumers. The interaction of economies of scale and transaction costs may be understood on the basis of a numerical example. Let us assume that an industry can locate in three regions: the Ruhr, a rich and central region with high wages and hence high labor costs; Catalonia, a middle-income region close to the large European markets, and Andalusia, a peripheral region with low wages and hence low labor costs. Economies of scale play a major role in the sense that unit production costs increase with the number of locations. The firm can produce in all three regions, in two of them, or in only one The choice of location is simply a minimization of the sum of production and transaction costs. The numerical example in table 2-2 assumes that (a) produc- tion costs are higher in the Ruhr than in Catalonia, and higher in Catalonia than in Andalusia; (b) it is less expensive to concentrate production in one location because of the economies of scale; and (c) ease of market access means that production in three locations minimizes transaction costs and that Andalusia is farther from the large markets than Catalonia, and Catalonia than the Ruhr. 24 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH TABLE 2-2 HYPOTHETICAL COSTS Transaction costs Assumption Production costs High Low Production in three locations 16 0 0 Production in Ruhr + Catalonia 1 5 3 1 5 ProductioninRuhr+Andalusia 14 6 3 Production in Ruhr 13 4 2 Production in Catalonia 12 5 5 Production in Andalusia 11 16 8 What happens with the integration process in this numerical example when the transaction costs are cut by half? At first, it is easy to see that when transaction costs are high, it is profitable to produce in the three regions in order to minimize trans- action costs. We thus have the situation of a geographically dispersed industry, since companies want to be located close to all their markets, both big and small. When transaction costs are reduced, for example because of commercial and monetary integration or the integration of transport infrastructures, the firm may exploit the economies of scale and produce in only a single location. Which region is going to benefit from this concentration? In this example, we see that it is not the region with the lowest costs that benefits from relocation. Catalonia has higher costs than Andalusia but has better access to the large markets, not only its own but also those within the core of Europe Thus, in this example integration has a convergence effect, since Spain benefits from the relocation within its territory of activities that were at first partially located in a richer country. However, in Spain itself, activities are relocating from the poor and peripheral region to the wealthy region. Thus we have the phenomenon of local divergence. It is clear, however, that if transaction costs were lower (for example, if they were close to zero) the region with the lowest costs would benefit from relocation. In theory, we have here a bell-shaped relationship between spatial concentration and transaction costs. This means that the poor and peripheral regions will attract enterprises for which the regional costs are relatively higher than the transaction costs, and for which the economies of scale are relatively small On the other hand, the "core" regions will attract firms for which economies of scale and transaction costs are of prime importance. The study by Forslid, Haaland, and Midelfart- Knarvik (2002) shows that in Europe the location of industries seems to follow this pattern and that the relationship between concentration and commercial inte- gration follows a different bell curve for each industry. Concentration is particu- larly strong when the firms involved belong to sectors with increasing returns, PlUBLIC POLICIES AND ECONOMIC GEOGRAPHY 25 which seems to provide empirical confirmation of the studies on the new eco- nomic geography. Recent literature using the term "new economic geography" exploits this inter- action between economies of scale and transaction costs but goes further, analyzing the cumulative phenomena that can come about. These can culminate in agglomera- tions based on worker migration (American model) or on relocation of the enterprises (European model). Let us take the example of the second agglomeration mode. An industry setting up in a region with few other businesses will need to import most of its inputs from other regions and will also have to sell most of its production in regions other than tlhe one in which it is located. From that standpoint, it will be at a disadvantage com- pared with firms in other regions with high levels of industrial concentration. This will apply in particular to industries with strong economues of scale, namely those to which market size is decisive for their profits. These are in general firms from sec- tors where the fixed cost (for example, the amount assigned to research and devel- cpment before the start of the production phase) is high. On the other hand, businesses lDcating in a low-concentration region will experience less competition in terms of attracting workers (and hence have lower wage costs). We therefore see that industries with strong economies of scale, belonging to sectors with a high profit margin, will automatically be attracted to regions where other enterpnses (both customers and suppliers) are already installed In such regions ai cumulative process is likely to be established, since the more a given region has been able to attract enterpnses of one type, the more it will attract others. Conversely, the regions from which those businesses have fled will find it increasingly difficult to attract others. Such regions will be able to attract firms whose wage cost represents a large share of total costs, in sectors that are highly competitive and thus post low profit margins. But this cumulative process will mean a high level of specialization in regions with a heavy concentration of industries with high profit margins and wages as compared with regions that are able to attract only firms with low profit margins and wages. The former will specialize in high-technology goods that do not suffer much competition from the opening up of the low-wage countnes to trade. The latter will be directly affected by trade globalization. This process of spatial concen- iration is not linear, given that the drop in transaction costs may have no impact at iFirst. It is only when those costs have reached a critical level that the cumulative process takes shape. Once that process has started (once the former spatial equilib- rium has been destabilized), it becomes self-maintaining and self-reinforcing. Regional Policies-Trade-off between Equity and Efficiency Equity is one of the traditional motivations of regional or spatial planning policies. Certain economic agents, be they workers or consumers, are not mobile and are therefore condemned to live in poor or declining regions from which the mobile 26 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH factors (capital and highly skilled workers) have departed. Because of the lower labor demand in such regions, real wages will adjust downward, or, if real wages do not adjust because of labor market rigidities, unemployment will increase. In both cases, the welfare of the inhabitants will deteriorate. As consumers, those agents will also see their welfare deteriorate since certain goods and services will no longer be produced locally (the businesses will have left for more wealthy regions). In certain cases, in particular for certain services, the transaction cost will become so high that the agents will no longer be able to afford the services. Thus the diversity of consumable goods and services in the poor region will decline. Moreover, the most mobile agents are in general those with the highest level of human capital (education, experience, and so forth). Such agents, thanks to the pos- session of "positive externalities" in the form of localized social interactions, have a positive impact on the productivity and thus on the real wages of other workers. By leaving a region in decline, the most productive workers thus also have a neg- ative impact on the productivity of the remaining workers-that is, those who are the most disadvantaged. There is, therefore, an absence of market coordination, given that when certain agents decide on their location, they do not take into account the effect of their choice on the other agents. From that standpoint, there is a real market failure, with the consequent increase in inequalities that is specific to the spatial dimension of the economy and may thus serve as motivation for pub- lic intervention. There are several ways to analyze the impact of the agglomeration phenomenon on the least mobile agents. The first would be to refuse to see it as a problem of equity but to interpret it as coming from a specific market failure. This approach would find its origin exclusively in the lack of mobility of the most disadvantaged agents, some- thing we have already noted as characteristic of the European countries relative to the United States. In Europe, promoting the spatial mobility of workers is not considered a solu- tion to the problems of regional inequality. This is legitimate, but only partially, since because of cultural and sociological obstacles, there will always be a substantial set of workers who will be harmed by geographic inequalities. Having regions empty both of inhabitants and of economic activities (such as the Dakotas in the United States) is unacceptable in Europe. A further motivation for public intervention at the regional level, put forward by the Commission, is that of efficiency. It sees in geographic disequilibria "an under- utilization of economic and social potentials and an inability to take advantage of opportumities that could be beneficial to the Union as a whole" (European Commis- sion 1999). This motivation is much less clear than the equity-based motivation. If the phe- nomena of spatial concentration are explained by the existence of economies of scale, this means that the spatial agglomeration is at the ongin of economic gains. This will be the case if firms can benefit from the proximity of other enterprises in the same PUBLIC POLICIES AND ECONOMIC GEOGRAPHY 27 sector to diminish their costs (transport costs or fixed costs). It will also be the case i F such concentration makes it possible to increase the firms' productivity through localized spillover effects-that is, if the firms can receive transfers of knowledge from other neighboring businesses. These localized spillovers have been documented in numerous studies (see, for example, Jaffe, Trajtenberg, and Henderson 1993) The example of Silicon Valley shows the advantage a country can obtain from a very heavy spatial concentration of activities with positive technological externalities. The stronger spatial concentration of innovation-based activities in relation to production activities thus has an economic rationale, and the benefits of this spatial concentra- tion go beyond pnvate gains. The objective of policies promoting a greater dispersal of economic activities is based on the assumption that the economic geography produced by market forces alone is too concentrated. However, the efficiency argument may demand more or less spatial concentration: on the one hand the economic gains of spatial agglom- eration, and on the other the effects of congestion (pollution, for example). The fact that in Europe the convergence of countries is accompanied by national divergence makes one think that the former type of argument, efficiency gains with spatial con- centration, has pride of place. In this case, a trade-off between equity and spatial efficiency appears inevitable (see Martin 1999a, 1999b). Figure 2-1 shows the pos- itive correlation between per capita income growth and growth in regional dis- parities over the five-year period 1994-98 and suggests the existence of such a trade-off. FIGURE 2-1 GROWTH AND REGIONAL DISPARITIES, 1994-98 Annual GDP growth rate 5 Finland 4 Portugal Netherlands 3r* Spain Nt* United 3 - Belgium * Kingdom France Greece Sweden 2 -*+ Austria* Italy Germany 1 -10 -5 0 5 10 15 20 Growth rate of regional disparities in per capita GDP 28 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH DiMffeullty of Assessing Regionnall IPoRlies One of the principal lessons from the studies on the effects of public policies on eco- nomic geography (see, for example, Puga 2002 and Baldwin and others 2002) is that because of the cumulative, nonlinear processes under way it is very difficult to eval- uate and anticipate the effect of public policies intended to offset tendencies toward concentration. Thus, regional policies that, for example, finance transport infra- structure, have effects in terms of both supply and demand. The demand effects are principally of a short-term, typically Keynesian nature: the construction of a high- way increases the local demand for goods and labor and hence increases the incomes available in the region. In turn, this increase generates expenditure on local goods (in particular nonexchangeable services), and so on. In actuality, macroeconomic stud- ies, such as those of the European Commission (1999), have found a positive effect on regional growth over the short term-that is, over a five- or six-year period of regional policies. But those macroeconomic studies look at the macroeconomic effect at the country level only, not the regional level. This is a problem, since we have seen that convergence exists among the European countries but not among regions in the same country. Nothing is said about regional policies benefitng the poorest regions of countries converging at a global level, but it is well known that the poor and rich regions of those countries did not converge over the same penod. The pertinent study level is the region, not the country. Other empincal studies have not been very encouraging for regional policies. This is particularly the case for Boldrin and Canova (2001), who find no effect of regional policies on regional convergence. Midelfart-Knarvik and Overnan (2002) find that regional policies have encouraged research and development-intensive industries to locate in countries and regions that have low endowments of skilled labor. From the standpoint of efficiency, it is not obvious whether this is a great achievement. More basically, it is essential to study the effects of these long-term policies- that is to say, the effects on supply. Those effects, particularly as regards the location choices of industries, are more complex and may even be the exact opposite of the effects on short-term demand. This may be the case, for example, with transport infrastructure, which has received preferential treatment within the framework of European regional policies. If we look at the case of industries with economies of scale, a policy to open up peripheral regions may have a paradoxical effect. By reducing transaction costs for interregional trade, such a policy may encourage firms to exploit their economies of scale by concentrating production in a single location. For firms whose wage cost is relatively low, this will mean concentratmg production in the wealthy region, even if it means exporting part of that production to the poor region at low cost, thanks to the new transport infrastructures (see Martin 1999a; Martin and Rogers 1995). The numerical example given in table 2-1 may be reinterpreted as an experiment to set up PUBLIC POLICIES AND ECONOMIC GEOGRAPHY 29 transport infrastructure between the three regions, and it is easy to see that it culmi- nates in the concentration of activities in one single region The recent study by Combes and Lafourcade (2001), suggesting that the drop im transportation costs in France over the past 20 years has resulted in an increased concentration of industries with strong economies of scale, shows that this result is not only theoretical. The example of Italy is also interesting the construction of highways between north and south to open up the south does not appear to have had the expected effect and has perhaps even accentuated the decline of the south (see Faini 1983). Of course, this outcome is not a widely generalized one. Martin and Rogers (1995) show that a transportation infrastructure that reduces transaction costs within a disadvantaged region will have a positive effect on that region, since it makes it possible to increase the effective size of the market.in that region The public infra- structure that has enabled the northern region of France to come close (in terms of transportation costs) to the most wealthy European regions is at least partially at the root of that region's renewed growth. In this case, transportation infrastructure has had the effect of making this a "core" region and has not simply served to open up a peripheral region Thus, industries have been able to concentrate their production in a region that has grown close to the large markets and also benefits from relatively low wages or at least a labor market that is very favorable to businesses. This exam- ple shows that the same public infrastructure will have very different effects from cne region to another. While part of the phenomenon of concentration and regional inequality is due to processes of localized "technological spillovers" (that is, to the fact that enterprises located in the same region benefit, through social interaction, from a higher level of technological performance), the recommendations for public policies are very dif- ferent. It is no longer a matter of reducing transaction costs on trade in goods among regions but of reducing transaction costs on the trading of "ideas " This involves a change in pnorities: rather than financing highways (with positive Keynesian effects), it will be necessary to promote technological convergence among regions, which will involve public programs for telecommunications, the Internet, and training of human capital as well as policies aimed at increasing the productivity of poor regions and facilitating the transportation of people rather than goods Martin (1 999b) argues that such policies make it possible to achieve gains in both regional efficiency and regional equity, in contrast to the regional policies financing goods transportation infrastructure or transfers to the poor regions. Another difficulty with regional policies is that the goal of reducing regional inequalities is often confused with that of reducing inequalities among individuals The fact is that regional policies consisting of subsidizing industnes to enable them Io relocate in disadvantaged regions may have exactly the opposite effect. If the cap- ital is mobile, subsidizing the return on that capital in one region amounts to increas- ing its return in all regions. Regional policies for subsidizing capital, even if they 30 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH succeed in reducing regional inequalities, can thus end up increasing inequalities among individuals, since the first beneficiaries may be those who hold the capital. Conclusions The low mobility of Europeans-not only among countnes, but, and more important, among regions of the same country-helps explain why the location of economic activities is a major political and social issue in Europe. On the one hand, it weakens the agglomeration process. Greater mobility would certainly involve the multiplica- tion of regions devoid of both inhabitants and businesses. On the other hand, it is this low mobility that is at the root of the human cost of spatial concentration, in particu- lar in terms of unemployment, since the departure of companies does not involve the departure of workers. In the United States there is more or less a pattern where the workers passively follow the businesses, while in Europe there is a demand for poli- cies designed to encourage firms to follow the workers wherever they choose to live. Regional policies then appear as an alternative to human mobility: it is socially prefer- able to move activities to where the people are than to move people to activities. In this chapter, I have reviewed some of the challenges faced by regional poli- cies. I have done this using some of the insights provided by the new economic geog- raphy literature. I identified several main challenges. First, a trade-off may exist between spatial equity and spatial efficiency: the same economic forces, namely increasing returns, that explain spatial agglomeration are at the ongins of economic gains. Second, policies that aim to reduce the peripherality of poor regions by reduc- ing trade costs among poor and rich regions may lead economic activities to leave the poor regions. The long-term supply effect of these policies may be the opposite of the short-term demand effect. In theory, policies that promote technological con- vergence seem more appropnate. Third, reducing regional inequalities does not auto- matically reduce individual inequalities. The identification of these challenges is not meant to disqualify the legitimacy of regional policies in Europe but to underline sim- ple rules of public economics that should apply to regional policies: it is essential to identify the market failures that legitimate public intervention, to pinpoint objectives (that is, which inequalities to reduce), and to recognize possible trade-offs between these objectives. Bibliography The word processed describes informally produced works that may not be commonly available through libraries. Amiti, M. 1998. "New Trade Theory and Industry Location in the EU: A Survey of the Evidence." Oxford Review of Economic Policy 14(2). F'UBLIC POLICIES AND ECONOMIC GEOGRAPHY 31 B3aldwin, Richard, Richard Forslid, Philippe Martin, Gianmarco Ottaviano, and Frederic Robert-Nicoud 2002 Economic Geography and Public Policy. Pnnceton, N.J.: Princeton University Press. 'Boldrin, Michele, and Fabio Canova. 2001 "Inequality and Convergence: Recon- sidenng European Regional Policies." Economic Policy 32 205-53. Briulhart, Marius. 1998 "Trading Places. Industry Specialization in the EU." Jour- nal of Common Market Studies 36(3):319-46. Bruilhart, Marius, and Johan Torstensson. 1996. Regional Integration, Scale Economies and Industry Location in the EU. CEPR Discussion Paper 1435. London: Centre for Economic Policy Research. Combes, Pierre-Philippe, and Miren Lafourcade. 2001. Transportation Costs Decline and Regional Inequalities Evidence from France, 1978-1993. CEPR Discussion Paper 2894. London: Centre for Economic Policy Research. Duro, Juan Antonio. 2001 "Regional Income Inequalities in Europe: An Updated Measurement and Some Decomposition Results." Madrid Instituto de Analisis Econ6mico Consejo Superior de Investigaciones Sientificas. Processed. European Commission. 1999. The European Regions Sixth Periodic Report on Socio-Economic Situation in the Regions of the EU Luxembourg: Official Publi- cations Office. .2001. Second Report on Social and Economic Cohesion. Luxembourg: Offi- cial Publications Office. Faini, Riccardo. 1983 "Cumulative Process of Deindustrialization in an Open Region: The Case of Southern Italy, 1951-1973." Journal of Development Eco- nomics 12(3) 277-301. Forslid, Richard, Jan Haaland, and Karen H Midelfart-Knarvik. 2002. "A U-Shaped Europe? A Simulation Study of Industrial Location." Journal of International Economics 57:273-97. Jaffe, Adam, Manuel Trajtenberg, and Rebecca Henderson 1993. "Geographic Localization of Knowledge Spillovers as Evidenced by Patent Citations." Quar- terlyJournalofEconomics 108(3).557-98. Krugman, Paul. 1991 Geography and Trade. Cambridge, Mass.: MIT Press. 32 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Martin, Philippe. 1998. "Can Regional Policies Affect Growth and Geography in Europe?" World Economy 21.757-74. -. 1999a. "Are European Regional Policies Delivenng?" European Investment Bank Papers 4(2):10-23. . 1 999b "Public Policies, Regional Inequalities and Growth." Journal of Pub- lic Economics 73:85-105. Martin, Phillipe, and Carol A. Rogers. 1995. "Industrial Location and Public Infra- structure." Journal of International Economics 39:335-51. Midelfart-Knarvik, Karen H., and Henry Overman. 2002. "Delocation and European Integration: Is Structural Spending Justified?" Economic Policy (October). Midelfart-Knarvik, K. H., H. Overman, S. Reading, and A. Venables. 2000. The Location of European Industry. Economic Papers 142. Luxembourg: European Commission. Neven, Damien, and Claudine Gouyette. 1995. "Regional Convergence in the Euro- pean Community." Journal of Common Market Studies 33:47-65. Overman, Henry, and Diego Puga. 2002. "Unemployment Clusters across Europe's Regions and Countries " Economic Policy: A European Forum. 34.115-43. Puga, Diego. 2002. "European Regional Policies in Light of Recent Location Theo- nes." Journal of Economic Geography 2(4):373-406. CHAPTER 3 )Regional Policies and EU Enlargement iMichele Boldrin and Fabio Canova The European Union (EU) is moving toward the sixth enlargement of its history. The process started in 1988, when the first Trade and Cooperation Agreement was signed with Hungary; took its main turn with the 1993 European Council of Copenhagen; and is supposed to come to its first completion with the Intergovernmental Confer- ence of 2004. Candidates to enter the EU are 10 Central and Eastern European coun- tries (CEECs) that were ruled by socialist regimes until about 1990-Bulgaria, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, the Slovak Republic, and Slovenia. To this group (which we will refer to, collectively, as CEECs or the CEEC10) one should add Cyprus and Malta, for a total of 12 candidates. In 2000, the European Council of Nice established that negotiations with a first group of candidate countries about the exact timing of their accession should be completed by the end of 2002. Recent political signals from Brussels suggest that, roughly, the planned admission schedule should be followed, with Cyprus, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, the Slovak Republic, and Slove- nia gaining full admission in 2004, while Bulgaria and Romania should join about four years later. Given the very small size of Cyprus and Malta, our analysis con- centrates on the CEECI0, and the term "enlargement" refers to the entrance in the EU of the CEECIO. We are interested in the following broad policy question Tak- ing both the enlargement process, as currently defined by the EU, and the Structural Funds as a given, what is the best way for candidate countries to fuel real conver- gence at the national and regional levels? 33 34 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH To put it differently, assume that the ongoing admission process is completed by 2004. What changes, if any, in the principles and methodologies underlying current EU structural policies should be recommended to foster fast and homogeneous growth among the new members, at both the national and the regional levels? What role should national and supranational policies play in this process? The word "regional" here means Nomenclature des Unites Territoriales Statistiques (NUTS) 2 territorial units or lower, as this is the level at which EU structural policies are designed and evaluated. We pointed out elsewhere (for example, Boldrin and Canova 2001) that the NUTS 2 regions are highly heterogeneous and that pursuing economic convergence at such a territonal level makes little sense. We argue later that this is even less sensible for most CEECs. Nevertheless, taking this policy objective at face value, we ask how Structural and Cohesion Funds and their allocation should be mod- ified to reach it, and what kinds of national policies may create an environment con- ducive to growth. Our attention concentrates on the twin issues of economic run growth and convergence for the CEEC 10 after joining the EU. In the second section of this chapter, Initial Conditions in the Enlargement Coun- tries, we quantify the extent to which enlargement countries are economically back- ward with respect to the EU 15. We document the evolution of the main macroeconomic indicators since the transition from socialism began, and we find that, despite the fact that the process began to bear fruit in the mid- 1 990s, the average CEEC is still where it was 10 years ago relative to the average EU 15 country. Heterogeneity across CEECs, measured either in levels or m rates of change, is probably more substantial than the regional differences within each country. Some countries (Hungary, Slovenia, and, depending on the indicators, Poland or the Czech Republic) appear to be ahead of the rest. We document that, with the current investment rates in physical capital, it will take a considerable amount of time for CEECs to reach thc capital-output ratio existing in the EU. Whether this will be a problem depends on the dynamics of technological acquisition in these countries. Foreign direct investment (FDI) is growing but is still too small to create dramatic changes in productivity. We also examine, very briefly, the labor market conditions of CEECs. Much has been wrntten on this issue, and our con- tribution here is to point out that most of the imbalances noted by commentators are endemic also to the labor markets of current EU members. In this sense, sweeping reforms affecting incentives and eliminating dead-weight losses should be advocated for both groups of countries. We complement our descriptive analysis by reporting a simple growth accounting exercise. Consistent with studies of this type, we find that, from the viewpoint of growth accounting, productivity gains, as measured by the growth rate of total factor productivity (TFP), constitute the most important engine of long-term growth. We examine possible sources of TFP growth in the CEECs: FDI, shrinkage of the public sector or the agricultural sector, and institutional changes. All appear to be important, but it is impossible to find a single cause of TFP growth. Fmally, we examine the sectoral composition of employment and value added (VA) across CEECs and relative to the EU. In CEECs, more people are employed and more VA IREGIONAL POLICIES AND EU ENLARGEMENT 35 comes from agnculture and low-productivity sectors than in the EU. However, when we compare the CEEC IO with the poorer Mediterranean countries of the EU, many similarities emerge, in terms of both snapshot conditions in 2000 and dynamic evolu- tion over the transition Given current conditions, convergence, if any, should therefore be expected to occur toward these countries and not toward the EU average. In the next section-Initial Conditions, Former Newcomers, New Steady States- we ask how CEEC regional/national dispanties compare with those currently existing within the EU and, more crucially, with those that existed in the mid- I 980s within the EU. Economic dispanties between the EU15 and the CEECIO are not different from those the EU learned to manage dunng previous enlargement episodes. Regional inequalities within the CEECIO are also comparable with, if not lower than, those in the EUI5. There are historical differences between the enlargement process of the 1980s and that of today, but they seem to be playing out in favor of CEECs, not against them. We perform a simple econometnc exercise trying to quantify the most likely effects of accession on the growth rates of the CEECs. With some important qualifica- tions, we estimate the changes to be relatively small. We conclude that the new enlarge- ment is not substantially different from previous ones and that, as in those cases, holding internal policies constant, the economic gains from accession will be positive but not dramatically large-accession (and the arrval of EU funds) will not magically eliminate economic backwardness. National policies could dramatically alter these conclusions, and we detail at the end of this chapter a set of recommendations that, in our opinion, could make a difference. In the next section-Old, Useful Lessons-we look at other studies of economic convergence in Europe to gather some lessons about policies that may increase the growth rate of the CEEC 10 relative to that of the EU Our provisional conclusions are that (a) barring particularly dramatic circumstances or changes in institutional settings, economic divergence is not a likely outcome of trade liberalization, (b) countries belonging to common trade areas with a reasonable degree of factor mobility seem to grow at a fairly common rate in the long run, and some degree of catching up should be expected in the intermediate period, (c) policies of extensive privatization, reduc- tion in the fiscal burden, openness to FDI, and product/labor market liberalization seem to be the main engines behind the "robust" growth experiences observed so far; (d) technologically and economically backward countnes that enter into a free trade arrangement with more advanced ones undergo a fairly rapid and dramatic process of structural change that leads to the destruction of employment in the agricultural sec- tor and other backward sectors Such a transition process, already under way in the CEECIO since the early 1990s, is the key threat to long-term prosperity This is because the induced regional patterns of long-term unemployment may persist over time. Elimination of structural unemployment requires strong and clearly targeted mobility and training policies. Such policies have seldom been implemented because, in the face of growing unemployment, political pressure for the adoption of grossly inefficient transfer policies is usually victorious. The likelihood of such mistaken 36 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH policy choices is enhanced by the availability of external aid. Economic theory aside, historical evidence shows that subsidies and transfer policies make regional imbalances more, not less, persistent. We discuss several histoncal episodes within the EU15. We look in particular at Greece, Ireland, Portugal, and Spain and relate their different performances to national and EU policies. We also look at the regions of the Italian Mezzogiorno, which have been the objects of almost 50 years of national and European structural policies. Finally, we compare the kinder of the former German Democratic Republic with some of the most dynamic CEECs. While they shared almost 50 years of communist rule, the German Democratic Republic and the dynamic CEECs have been exposed to very different policies since the transition started. The former German Democratic Republic, which had an advantage as far as initial conditions were con- cerned, has been the privileged target of a large amount of external funding and support, coming both from the EU Structural Fund programs and from the German government. Hungary, Poland, and other CEECs have received practically no external aid and have been exposed to a fast transition from socialism to the free markets. The next section, Assessment of Regional Policies in the Enlargement Process, looks directly at the effects of Structural Funds. We are interested in ascertaining whether and how the allocation and use of Structural Funds is likely to affect the growth path of the EU accession economies and regions. Our analysis here proceeds in two stages. In the first, we argue why economic theory and empirical evidence sug- gest that Structural Funds (and the Common Agricultural Policy [CAP]) should be terminated, because they amount to nothing more than pork barrel spending favoring one lobby after another. As net transfers add up roughly to zero for most countries, all one is left with is the dead-weight loss of taxation, and the cost of maintaining an expensive and unproductive bureaucracy both in Brussels and at the national and regional levels. In the second step we recognize that, political games being what they are, structural policies are unlikely to be terminated anytime soon. It is therefore worth asking how those funds could be spent more efficiently. First, there is the need to redefine the convergence objective at a territorial level less mappropriate than the NUTS 2 regions. Second, external transfers seem to yield their highest payoff in extremely poor regions, when appropnate national policies are also in place to take advantage of such external funding. This suggests that structural funding should be concentrated only in regions that are below 50 percent of the EU average, and that conditionality should be made stronger than it currently is. Third, the focus of the sup- port should be shifted away from funding hundreds of small investment projects in the most disparate areas. While theory is quite ambiguous on this point, historical experience suggests that investments in transport and communication infrastructures, public utilities, and productive human capital are almost invariably among the precon- ditions of episodes of extraordinary economic growth. The chapter by Frank Barry in this volume, detailing some important aspects of the Irish miracle, further supports this claun. We recommend, therefore, that Structural Funds be concentrated in only these areas and that the objectives of European structural policies be redefined accordingly. REGIONAL POLICIES AND EU ENLARGEMENT 37 The last section contains a summary of our findings and a list of policy recommendations Initial Conditions in the Enlargement Countries The purpose of this section is to describe the current macroeconomic conditions of ('EECs, to assess the extent oftheirbackwardness with respect to the EU15, to high- light the heterogeneities existing both within and across candidate countries, and to evaluate the stance of those variables that theory or empirical practice found impor- tant in favonng long-term economic convergence. We concentrate discussion on those macroeconomic indicators that are more relevant from a growth perspective, leaving aside variables such as inflation, debt, and government deficit, which are out- side the scope of our analysis. It should go without saying that a stable and, espe- c;ially, predictable fiscal and monetary environment constitutes a key precondition for economic development. Sustained economic growth is impossible in an economy facing high and uncertain inflation and tax rates. Necessary conditions are, never- theless, just necessary low and stable inflation is not enough Low and undistor- tionary taxation is not necessanly a consequence of a stable fiscal policy. A variety of other policies may provide the wrong set of incentives In particular, historical experience shows that distortionary and wasteful industnal, labor market, and wel- iare policies may be very detrimental to long-term growth We focus on this second set of policies. Unless otherwise noted, data for the year 2000 are used to provide a snapshot characterization of initial conditions, while the evolution over the decade 1990-2000 is used to gauge the underlying tendencies We divide our discussion into several sub- s,ections. The first deals with synthetic measures of the wealth of a nation gross domestic product (GDP) per capita and labor productivity. The second addresses investment, savings, and FDI. The third discusses labor market conditions. The fourth reports the results of a growth accounting exercise and looks at the sectoral compo- sition of output and employment. In the fifth, we look at regional disparities. Per Capita Income and Labor Productivity CEECs are poor according to GDP per capita, measured in purchasing power stan- *dards (PPS). Figure 3-1 graphs this indicator relative to the EU15 average for the years 1991-2000 The GDP per capita of the mean CEEC has oscillated between about 36 and 42 percent of the EU average, the oscillations reflecting, to a large extent, changing business cycle conditions. The most recent estimates put it at about 40 percent The beginning and ending points of the time series are roughly the same, indicating that growth rates in the two economic areas have been, on average, similar While a minor degree of catching up is noticeable in some coun- tries in the second half of the 1990s, this is far from appearing to be an established and uniform trend 38 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH FIGURE 3-1 GDP PER CAPITA AND LABOR PRODUCTIVITY (share to EU average; EU average = 1) Bulgana Lithuania 0QJQ- 9435 - 02-N 55- 0795 O \ 02 \ 0-< Q2 IQ3Q R i9 tB9 107 IQ7 I97 199 197 9 IQ7 199Q 17,7,1 1997 1997 1791 1975 195Q *197 19~ *~99 Czech Republic - I_ Poland O025 0Q75 - 303 026 5990 0-3Q li9 *"1 19O 1999 S I9Q5 99Q IOQ 199Q *9 0725 5955 1951 5990 y7 1955 550 555 1Q79 1907 5953 89 Estonia Romania Hungary Slovak Republic 0.55- S0 0 2CQ Om 5 0`31 - ~ ~ ~ ~ --- 5950 5550 1999 1505 1994 99 99 19 91 191/5 li * 907 *0f 1970 *931 55 19 23 IQ 99 4 5W*955 5996 5097 1959 1599 0GDP pe cat -0-- L 0 So7wrce Eurostat 2001c, 2002c, autbors' calcula0o5s 03W- on ~~~~~~~~~~~~~~00- 0 1900 0951 Ii 1905 5995 59 4 15 6 59527 198 .90 Om 995; 195 5905 595 594 1995 -95 97 195 99 I- GDP per capita - -- Labor productivity Source Eurostat 200 Ic, 2002c, authors' calculations. REGIONAL POLICIES AND EU ENLARGEMENT 39 Focusing the discussion on averages is somewhat misleading since the past 10 years have witnessed substantial cross-country differences in the growth perfor- mances of CEECs, and, from this point of view, the heterogeneities are stronger than those among current EU members For example, while over the decade the income per capita of Romania and Bulgaria has fallen from 28 and 31 percent to 24 and 26 per- cent of the EU average, respectively, during the same period the per capita income of Slovenia and Poland relative to the EU increased by about I percentage point per year. As of 2000, Slovenia is the richest of the group, with income per capita above 70 percent of the EU average, higher than Greece and approximately at the level of Portugal; the Czech Republic, Hungary, and the Slovak Republic all have per capita iiicome in excess of 50 percent of the EU 15 average, Poland and Estonia lag behind at about 40 percent but appear to be growing faster than average; and for the rest (Bulgaria, Latvia, Lithuania, and Romania) it is around or below 30 percent There are at least two reasons why GDP per capita may not be a good indicator of the wealth of the CEECs. First, underground and unrecorded activities may be large relative to total market production. recent estimates obtained using electncity consumption (see IMFStaffPapers 2001, p. 75) putthis number at around 40 to 45 per- cent of current GDP. Consumption of electncity may be a distorted measure of eco- nomic activity because of different subsidies to electricity prices in different countries. Still, a reasonable estimate is that official GDP values are downwardly biased by about 20 percent. Second, because of the transition process, many nontradable activ- ities are likely to be mismeasured. Although it is hard to quantify, this source of mis- measurement could add up to 10 percent of current estimates. An alternative indicator is labor productivity, here measured by gross value added (GVA) per employed person, in thousands of PPS We report the time series for labor productivity in the CEECI Irelative to the EU also in figure 3-1 Roughly speaking, there is little difference between relative labor productivity and relative CiDP per capita. For example, the level of labor productivity of the average CEEC oscillates between 36 and 41 percent of the EU average and, despite some differen- tial growth in labor productivity in favor of CEECs during the first five years, the average value has not changed much over the decade In fact, factonng out the cycli- cal tendencies and excluding the first two years of the transition, when labor pro- ductivity gains were obtained through massive labor shedding, the EU/CEEC labor productivity differential in 2000 is approximately the same as it was in 1992-93. 1:lungary, Poland, Slovenia, and Latvia stand out relative to the others The first three display a clear upward trend up to 1998, which was apparently reversed in the past two years, and are responsible for most of the regions' productivity gains in the late 1990s. Latvia's relative labor productivity has doubled in the past five years, but its level is still among the lowest in the group. A word of waming about reading too much into the productivity measures is also necessary because of some aspects that characterize all transition periods. CEECs eKperienced huge job losses in the industrial (state) sector during the past decade, 40 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH which have not yet been completely compensated by the growth of new jobs in the service and advanced industrial (pnvate) sectors The expulsion of workers from inefficient sectors generates mass unemployment and labor productivity gains, some- what independently of the speed at which the rest of the economy modemizes. This is because most of the people expelled from traditional state-managed activities are unemployable in the new ones. Demography, then, plays a role in the adjustment process, with older workers entering long-term unemployment, which is terminated only by retirement. To give a concrete example, because of significantly different employment rates and demographics, Greece appears to be richer than Portugal when labor productivity is considered (more than 15 percent so, relative to the EU average), while in terms of GDP per capita the picture is reversed, with Portugal overtaking Greece by more than 10 percentage points. In any case, since both the absolute levels and the relative ranking obtained with the two measures coincide, we may want to attribute a certain degree of reliability to our findings. This leads to the conclusion that, on average, CEECs are at about 40 percent of EU15 in per capita income, with Poland and Slovenia displaying above-average growth performances while the rest grow at more or less the same rate as the EU. Saving and Investment Rates High investment rates have been, by and large, a critical ingredient of the most suc- cessful growth experiences in the past 40 years. China, Hong Kong, Korea, Singapore, and a number of other "miracle" economies have had investment rates well above the developed world average (rates as high as 35 percent are not unusual) either right before or during their growth boom. At the opposite end, countries that have either stagnated or displayed negative long-term growth feature very low levels of invest- ment (and saving) rates (for example, most Latin Amencan countries). Over the last 40 years, the empincal literature on economic growth has consistently reported strong correlations between the rate of investment (in equipment and machinery in particu- lar), the growth rates of labor and TFP and subsequent growth in per capita income (Miles and Scott 2002 is a recent reference). While the theoretical reasons for this cor- relation are far from obvious, and abundant evidence shows that capital accumulation per se is not the main cause of high labor productivity, it is important to recognize that a high investment rate provides a reliable signal of both short- and long-term poten- tials for growth. Capital comes in two forms: human and physical. The first is much harder to measure than the second, so economists are usually satisfied with some weighted measure of the number of years of formal schooling accumulated by the average member of the labor force. This is, obviously, a very imprecise measure, as not all school systems are identical from the point of view of productive human capital. This may easily be the case in the CEECs, where the school system had historically been used as an instrument of ideological control. We are insisting on these caveats REGIONAL POLICIES AND EU ENLARGEMENT 41 because, when one looks at a pure measure of the average number of years of school attended, the CEEC10, with 9.8 years of schooling on average, comes out ahead of the EU15 mean, which was only 9.5 in 1999 (OECD 2000). Controlling for content and for quality of schooling, the average CEEC may be somewhat below the EUI5 average human capital stock, but this adjustment is unlikely to make a major difference. This is an important aspect: the quality of the labor force of the CEECs is already comparable to that of the EU-it is highly educated and potentially capable of adapting to new economic circumstances. Frank Barry's chapter stresses the key role played by the high-quality Irish human capital in that country's successful growth. Hence, this positive initial condition should be prop- erly taken into consideration when evaluating policies: a highly educated work f'orce could prove a huge asset if the proper kind of high-tech investment were attracted. The picture is less rosy when one looks at machines, equipment, plants, infra- structures, and so forth. There are two problems in this respect: bad initial condi- tions (obsolete factories and infrastructures) and relatively low investment rates since the transition started. In the EU over the past 10 years, the investment rate has been 17.6 percent on average, with little variation around that level. The four cohe- sion countries all had investment rates in excess of 20 percent during the 1990s ALssuming similar depreciation rates of about 10 percent a year, a steady-state EU capital/output ratio of 1.7, and an initial capital/output ratio of 1 3 for the cohesion countnes, it will take them approximately another 10 years to match the EU capital/ output ratio. A similar calculation can be performed for the CEEC 10. Figure 3-2 plots sav- ing and investment rates for the CEECs over the past decade The average rate of gross capital formation was 24 5 percent in 2000, and above 20 percent for most of the last decade, but there are remarkable differences across countries and time peri- ods. For example, the Czech Republic and the Slovak Republic have investment rates exceeding 30 percent on average, while the Bulgana, Latvia, and Romania average is only 14 percent. Furthermore, the volatility of investment is high in all countries. More important, the countries that show or have shown the strongest potential for catching up (Hungary, Slovenia, and, to a lesser extent, Poland) are those with an investment rate that is trending more consistently upward. This fits with the received wisdom: it is not high investment per se that causes growth. Instead, when policy determines conditions that are appropriate for growth, invest- ment flows become significant It is important to stress that for the CEEC10 even an investment rate of 30 per- cent is not especially high once we take into account that depreciation and obso- lescence of old capital is significantly higher in the CEECI0 than in EU countries. As we will see, the estimated depreciation rate of capital over the decade is very high, and about half of gross investment is used to replace depreciated capacity. Clearly, these numbers overestimate the depreciation rate we should expect in the 42 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH FIGURE 3-2 SAVING AND INVESTMENT RATES (share to EU average, EU average = 1) Bulgana - Lithuania 0226 - 022 - om075 2 07- 000 SZ 141 IOP2 IM199 giC9 136 S6 470 .96 4W nK3 =I0 tKM 1RW 19'2893 164 12i 1KA 10"7 19; 6 MWK 9900 0 Czech Repubic 0.Poland 0 20 9 O m _ O4m 4 40 1990A IA 49196 192 44 49941965 944 4441 194 1999 40;. 41 190 4994 194 499 744 1999 4 1 996 49940 1999 200 2997 Estoia Romama Om- - Z \ 0240- , , , , , , 0.19- 01 1 19 7 74 19 44 90 1 79 I 00 01 4900 7991 I 726 799 1994 1993 1994 490 104 I290 20 DI 004 Hungary Slovak Republic 02409 04 - U~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~O , 3 r - 0 v r 02 - 2 a '5 - 00etmn ae7--7STmg7ae Source Eurostat 2001 a, authors' calculatioons FIEGIONAL POLICIES AND EU ENLARGEMENT 43 future, but a more optimistic view of what is to come does not eliminate the fact that net investment rates in CEECs are currently too low to guarantee a fast con- vergence to the EU capital/output and capital/labor ratios Under the reasonable assumption that the current capital/output ratio of CEECs is about half that of the EU, and taking a depreciation rate twice as high as that in the EU, we estimate that i1a will take more than 20 years for the thriftiest CEEC to reach the 1.7 level-the steady-state level consistent with the current investment and depreciation rate in the EU. Politically, 20 years may not be too long to attain convergence (Greece, F'ortugal, and Spain have been in the EU for 16 or more years and are still below the average), and depreciation rates may drop substantially in the near future. Nevertheless, this simple calculation indicates that hopes of miraculous conver- gence should not be harbored. High investment rates can be financed by local or foreign savings. The openness of capital markets and the security of the legal system play an important role in deter- mining the extent to which investment is financed by foreign savings. In all coun- tiies, domestic savings are the major source of funds for investment, and in only the C'zech Republic does a sustained flow of foreign saving allow investment rates to exceed domestic savings by about 3 to 4 percent in every year of the sample. In fact, direct portfolio investments have been relatively small in the CEECs, and their mag- nitude is completely dwarfed by FDI (see Eurostat 2002d). The role of FDI in bringing backward countries' capital stock and labor produc- tivity in line with those of the developed world is well understood, and it is confirmed by a number of positive growth experiences in Europe and elsewhere (see Barry in this volume and Martin and others 2001). However, we do not feel comfortable taking a stand on the direction of causality and, in particular, whether technological spillovers fiom FDI are high or low, relative to the productivity gains internalized by firms. The proportion of FDI in the total investment of CEECs has been increasing, in particular since 1995-96. The ratio of FDI to GDP is, on average, more than twice as large as the corresponding number for the EU (5 percent versus 2 percent). For some CEECs the net FDI inflow as a percentage of GDP has reached fairly high lev- els (II percent in Slovakia, 8 in the Czech Republic, 8 in Estonia). For others, it is still around 2 or 3 percent of GDP A 5 percent average, however, is substantially bet- ter than the one recorded in, for example, Russia, where in the 1990s the FDI rate was below 1.0 percent. It is also about 10 times better than the rate recorded by Greece, PDrtugal, and Spain in the early 1980s. Over two-thirds of the FDI in CEECs comes from the EU; Austria, Germany, and the Netherlands are providing the largest amounts, while the Czech Republic, Hungary, and Poland are the largest recipients, taking about 70 percent of the total flow to the region (see Eurostat 2000a, 2002c). Except in the Baltic states, FDI is concentrated in the private manufactunng sector (between 36 and 50 percent of the total), while the service sectors (trade, repairs, and financial intermediation) account for 23 to 40 percent of the total. 44 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH A distinct possibility is that, although an increase in the capital/labor ratio per se (driven by either domestic direct investment or FDI) is insufficient to ensure fast convergence of labor productivity, FDI helps to boost the TFP of CEECs by bring- ing in new technologies, expertise, and methods of production. Later, we provide a detailed discussion of the role of TFP in the convergence process. Here we note only that FDI rates positively correlate with TFP in many countries (figure 3-3). This seems to be particularly evident in Hungary and Poland, two of the three largest recipients of FDI, while the surprising counterexample is the Czech Republic. This, however, is consistent with the results of our growth accounting exercise: in the Czech Republic capital accumulation, and not TFP changes, accounts for most of the growth in per capita GDP. Labor Market Conditions Labor markets in CEECs displayed complex trends in the past decade. Activity rates in the 15 to 64 age group oscillated without a precise trend and are currently compa- rable to EU15 averages (slightly below 70 percent), with Hungary and Bulgaria being, from a historical perspective, the worst (at about 60 percent) and Romania the best (at about 75 percent). Employment rates in the same age group are not worse than those registered, on average, in the EUI 5 (about 64 percent). In companson with the EU15 laggards (Greece, Italy, and Spain, all hovering around a 55 percent employment rate), only Bulgana (51 percent) scores lower, while Romania, with a 69 percent employment rate, still stands out at the top. These snapshot figures have to be contrasted with differing tendencies during the decade. In figure 3-4, we plot unemployment rates together with the participation rate (defined as the number of employed and unemployed divided by the total population) in each country. Two clear trends emerge. First, the unemployment rate has dramatically increased in most countnes, except perhaps Hungary and Slovenia, with long-term unemployment rates being the reason for the strong upward trend. As of 2000, in fact, the long-term unem- ployed account for about 10 to 15 percent of the active labor force. The exception is Poland, where about one-third of the unemployed are long term. A large and fairly rapid increase in long-term (and short-term) unemployment is a well-known conse- quence of transition processes. Similar trends appeared in Portugal and Spain after their transition began. Labor shedding from inefficient sectors generates mass unem- ployment, both because new sectors are unable to grow at the rate needed to absorb those who are displaced and because workers expelled from decaying sectors have skills that are not usable in the new sectors. This phenomenon has important social implications for the transition generations and requires appropriate welfare and labor market policies to keep it within politically acceptable bounds, an issue to which we will return later. Second, even if now equal to those in the EUI 5, participation rates have declined substantially over the decade and are currently about 5 percent lower than in 1991. The largest decline was experienced by Bulgaria, whereas the best performer is llEGIONAL POLICIES AND EU ENLARGEMENT 45 F IGURE 3-3 FOREIGN DIRECT INVESTMENT (FDI) RATES AND TOTAL FACTOR FPRODUCTIVITY (TFP) (FDI and TFP divided by GDP) Bulgana = Lithuania Ols ale~0 -. 0,s - Om-0 -0 _0 10- ~ . 00 19i1 lI 1992 , I4 1990 1995 IX1 2990 IPog 20 2001 logo '0W loo lo90 ,. 1995 1995 190 9965 190 0, 201 oln ~~Czech Repubilc 0 Poland (125 ~~~Estonila OXRomania 1005 00 ~~~~~~~~~~~1 000 02 - 0 000 - , -000 -0-0 - 0110 190 191, 990z 900 990, 12,I9 27 " l2 w . 901 1999 1990 ," 1990 009 0 9 4 0991 0995 199 910 50 0,,2~ ~ ~~0 t Hunar 0055 Slva Reub AM .21- \ 0 * 1990 'Oil .li2 ,23 z9.4 Iti Im2 ,,7 29s 9Im wz 2; 19W0 19S1 I,¢ Ik lg l9;5 lg2s 4 I W41 ,zg0 2.;O WD W Estonia Slovmani 0l'15 0 - 05 =0, 075- 0 - 0a5 1990 16 91 1 .9 2 9 .9 199 0995 099 log 1995 999 9 25 . 990 9,9 00 1905 9 04 59 1 9 9 5 I1 99. 0 .95 099 05 l001 SDuIcM Eurstat2000, 202d, uthos'Ic--uTFPon Sol c uott200,20d uhos acldn 46 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH FIGURE 3-4 UNIT LABOR COSTS (real wages divided by productivity, base year = 1) Czech Republic Poland 290D- 7- 223 - 50~~~~~~~~~~~~5 r,-- 299 2991 19M 299 299 2999 299 1997 292 16 2 060 22 962 I99 293 299 299 2994 M 1995 296 997 1993 2999 20W0 2092 2itHungary sRomania 225 E 2 t l 2750 3200- 2.20 2~~~~~~~~~~~~~~~~~~~~~~~50- 210 292' m 0 ~ i 9 96 m 201I m m i m Latvia - Slovak Republic 275 2 1226 75~~~~~~~~~~~~~~~~~~~~~. 22.5 IM199 992 299 2093 22 99 2W 599 207 96 2999 2W 2922990 292 29 1 99 2999 I 1999 23 2997 ¶996 20Wg 2DW 2W, Lithuania Slovenia 2- 2.2- to - Iwo 2992 992 2993 294 295 2999 299 2999 2999 2 2922 199 2992 209 299 239 299 2g0W ' 1997 29 2990 20W 293 Source Eurostat 2002a, authors' calculations RIEGIONAL POLICIES AND EU ENLARGEMENT 47 probably the Czech Republic In general, the decline in participation rates in the CEECIO is the product of two phenomena. The first, common to the EU15, is the aging of the population, accompanied by a tendency to early retirement for older workers displaced by industrial restructuring Significantly, the country with the largest decline in the labor force participation rate, Bulgaria, is also the one in which population aging is most pronounced. A second factor, also present in some EU coun- tries but exacerbated in the CEEC 10 by the transition process, is the exit of many workers from the legal labor markets and toward underground activities Overall, the dynamics of labor market indicators in the CEEC10 and EU15 are similar. This is true for averages, degrees of variability, and overall trends The large increase in unemployment and the fall in participation rates contrast dramatically with the dynamics of unit labor costs (defined as real wages divided by productivity). We plot unit labor cost indexes for eight of the CEECI 0 in figure 3-5, where we normalize them to one in 1995 for all countries: the growth rate of average compensation greatly surpassed the growth rate of labor productivity over the whole period. This is an alarming signal for these countries and an important source of dif- ference from the EU15 (see Eurostat 2002d). According to the unit labor costs met- nc only, Slovenia and, to a lesser extent, the Slovak Republic have managed to keep the growth rate of real wages roughly in line with labor productivity gains. At the opposite extreme is Romania, where, because of the long stagnation in labor pro- ductivity, unit labor costs have more than doubled over the period. One should stress, though, that as of 2000 unit labor costs are still 50 percent lower in the average CEEC than in the EU (see Eurostat 2001 c). To the extent that the quality of the human cap- ital is about the same, this difference should make the CEECs attractive for EU FDI, especially after the enlargement is completed. Although along this dimension the CEECs are not so different from their EU15 counterparts, one may wonder why wage growth has exceeded labor productivity growth so much in the presence of high and nsing unemployment rates. Slowly, prob- ably too slowly, the noncompetitive features of national labor markets are being dis- mantled. In general, we share the view of the specialized literature, according to which, while some noncompetitive aspects still remain in the CEEC labor markets, high unem- ployment compensation and the high level of labor income (or payroll) taxation are the main culpnts in the current situation, making the switch from unemployment to employment unattractive for certain types of workers (see, for example, World Bank 2001a and 2001b). The level and the duration of unemployment compensation in tran- sition countries with high unemployment rates clearly reflect political and social pres- suires, and do not require further discussion here (detailed examination of these issues appears in World Bank 2001 a and 200 lb, and in Vaitilingam 2002). From our growth perspective, we stress that payroll taxes in CEECs are very high, even relative to EU or Organisation for Economic Co-operation and Development (OECD) standards, which are already high compared with those of the United States. For example, EU payroll taxes are 23.5 percent, and only Italy exceeds 30 percent; in Anglo-Saxon countnes the 48 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH FIGURE 3-5 UNEMPLOYMENT AND PARTICIPATION RATES (percent of labor force) Bulgana Lithuania 45 _ / - *1 , _ _ ; , --- - 24-_ 099 1991 992 099 199 099 19999 099 1099 2090 9900 099 MI3 003 09 0930 1935 09990 1 9 99 096 9 IV 0 2900 Czech Republic Poland 22- ~ - -- - - -- - -- - - --- -- -- -- ---- 2 - 04 - 29 - - 099 099 1 2 1990 19 099 1995 19 12 0097 1997 19 20 20 9 1 990 19 0X1 1995 A 997 199 199 200 70WI |-Unemployment- rate- - - - - - rt |-- 097rc Euosa 202b 201,atos-aclt 15 35- 25- 29- 20- 90- Is - , , , , , , , - 0 - Latvia Slovenia 94- ----- 45-~ ~ -------- 099 0990 099 099 099 099 099 1299 099 099 2900 0290 00 0092 0993 0934 0993 0990 0997 0992 0999 2932 2900 1- Unemployment rate - -- Participation rate Source Euroutat 2002b, 2001c, authors' catculations REGIONAL POLICIES AND EU ENLARGEMENT 49 taKes vary between 15 and 20 percent. In the CEECs, however, the taxes exceed 40 per- cent, except in Estonia (33 percent) and Slovenia (38 percent). One may want to exam- ine the rationale for keeping such high tax rates in labor markets facing tough transition problems: increasing unemployment forced governments to continuously increase spending on labor insurance policies. The recession of the mid-I 990s exacerbated this tendency, since other sources of revenue declined and governments relied heavily on payroll (and inflation) taxes to finance current expenditure. Payroll taxes are slowly decreasing in some countries-for example, Hungary-but this is far from being a gen- eralized process. Furthermore, the tax burden on labor exceeds the amount of payroll taxes as consumption, and income taxes also take away important parts of the worker's compensation. Garibaldi and others (2001) estimate that, on average in CEECs, the total tax rate on labor is around 74.7 percent, about 50 percent higher than in the EU (53 per- cent), with Poland (80 percent) and the Slovak Republic (81 percent) being the extreme. Since 1999 the unemployment level has slowly begun to decrease in most of the CEECs, Bulgaria being again the most serious exception to this trend. This also appears to follow a tendency common to the EU15, even if there are no examples of employment miracles that could be compared with, for example, the Netherlands or Ireland. But then again, there are no examples among the CEECs of courageous labor market reforms either. Therefore, the appropriate companson to be made is with those EU economies in which unemployment is still high and labor markets are still heavily regulated, such as France, Germany, Italy, and Spain. When this comparison is made, recent movements in labor markets in the CEECs very much resemble those in the EU countries. Growth Accounting, Once Again A. growth accounting exercise requires several assumptions, most of which are some- what heroic when applied to CEECs. Nevertheless, we believe that the exercise sheds useful light upon three questions- (a) whether the same factors account for growth across CEECs; (b) whether those factors are the same as those fueling economic growth in more developed countries; and (c) whether there is any evidence that FDI and technological spillovers have so far contributed to economic growth in the CEECs. We make the following set of assumptions: * The share of labor in national income is 70 percent. This roughly corre- sponds to the estimate obtained using data for the Czech Republic, Hungary, and Poland. * Capital stock increments are computed summing up investment over the period and subtracting yearly depreciation. Data for depreciation are avail- able for only the three Baltic states, in which the depreciation rate is esti- mated to be 40, 47, and 52 percent of the gross investment rate. These estimates seem to be on the high end of the distribution; hence, we use a value of 40 percent for the remaining countries. 50 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH • Since no information about part-time versus full-time labor is available, the increments of the labor input are computed using bodies. Data for Hungary indicate that the accounting discrepancy between using bodies and hours is small. We expect the same to hold for the remaining nine countries. o The increment in the domestic stock of capital is calculated as a residual, subtracting FDI increments from total increments. The same depreciation rate is applied to domestic capital and FDI. Since the technological content of FDI is higher and its depreciation is probably lower, this procedure biases FDI's contribution downward. o Since reliable data on FDI flows are avallable for only a subset of the 10 years we consider, we make the assumption that FDI was zero in all the non- available years. Again, this assumption biases downward the estimated contribution of FDI to growth. o We calculate the effect on growth of the shrinking of the agricultural sector by multiplying the average labor share by the decrement in the population employed in agriculture. Table 3-1 reports the results of our exercise. Since the years of the sample period differ across countries, the first column reports the range of years for each country. Average growth rates of GDP vary substantially, from -4.0 percent per year in Bulgaria to 7.7 percent per year in Latvia. Apart from Latvia, success sto- ries appear to be Slovenia, with an average growth rate of 5.0 percent, and Lithuania, at 4.9 percent. After Bulgaria, the worst performers are Romania (-0.4), Estonia (2.0), and the Czech Republic (2.0). The relative contributions of labor and capital follow analogous pattems across countries: the capital stock has a small influence while the contribution of labor to growth is negative. To be precise, the contribution of capital to aggregate growth is fairly constant and centered at around 1.4 percent on average. Exceptions are Bul- garia (0.6) and Latvia (0.8) on the negative side, and the Slovak (2.2) and Czech (2.0) Republics on the positive side. In the latter two countries, growth in capital accounts almost entirely for the growth in aggregate output, with vanations in the labor inputs or TFP playing a secondary role. As already mentioned, labor dismissals were intense during the 1 990s. This is reflected in the uniformly negative contribution of the labor input, which averages about 1.1 percent. The drop is extremely large for Bulgaria (-4.0 percent) and quite small for Romania (-0.001 percent), with all the other coun- tries falling in between. Qualitatively speaking, these pattems are not dissimilar from those observed in Spain and Portugal after 1975 (see Manmon 1996). Quantitatively, the difference is one of speed and magnitude, both of which appear to be higher in the CEECs, mak- REGIONAL POLICIES AND EU ENLARGEMENT 51 TABLE 3-1 GROWTH ACCOUNTING Timne AFDI/ AGR A Country period A y/y A k/k A n/n A TFP FDI n/n Bulgaria 1991-2001 -0 040 0 006 -0040 -0 006 0 001 -0 0008 Czech Republic 1993-2000 0 020 0 020 -0 002 0 002 0003 -0 0006 Estonia 1992-2001 0 020 0 015 -0.015 0 020 0004 -0 0055 Hungary 1992-2001 0 023 0 014 -0 004 0 013 0 003 -0 0008 Latvia 1991-2000 0 077 0 008 -0 013 0082 0 002 -0 0027 Lithuania 1991-2000 0 049 0 018 -0 014 0 044 0 002 -0 0026 Poland 1991-2001 0 032 0 012 -0 006 0 040 0 001 0 0002 Romania 1991-2001 -0 004 0 010 -0 000 -0 015 0001 0 0000 Slovak Republic 1993-2000 0 030 0 022 -0 010 0 017 0 003 -0 0017 Slovenia 1992-2000 0 050 0 017 -0 010 0 042 0 001 -0 0016 CEEC average n a 0 025 0 014 -0 011 0 022 0 002 -0 0016 n a = not applicable, Ay/y = output growth, Ak/k = capital growth, An/n = growth, ATFP = TFP change, AFDI = FDI gr-owth, AGR An/n =the contnbutlion of agricultural employment Note The increments in capital stock are computed by summing up investment over the penod and subtracting yearly d-preciation Data for depreciation are available for only the three Baltic states the depreciation rate is estimated to be atound40 to 50percentofthe investment rate We applya value of 45 percent to the otherseven countries The incre- mrrents on the labor input are computed using bodies and not hours, which are not available in many countries Data for Hungary indicate that the difference is small The domestic increment in the capital stock is calculated as residual The same depreciation rate is applied to domestic capital and FDI, therefore biasing downward the contnbution of FDI to growth The contnbution of agnculture to growth is calculated by multiplying &ii/n by the decrement in the population employed in agnculture Source IMF's intemational finance statistics, authors' calculations ing their transition process even more remarkable when compared wtth the one that took place in the Ibenan Peninsula. On average, the seriously downward phase of the transition process lasted about seven years in the CEECs thts is about half the time it took for Spain and Portugal to start growing faster than the EU average after their transition began. The relative size of the sectors affected by the transition is also rnuch larger tn CEECs. The Spanish and Portuguese economies were heavily pro- tected from foreign competition and had a strong state presence in all sectors but nev- ertheless were market economies. State-controlled firms never accounted for more than 20 or 25 percent of GVA in the tradable sector In contrast, private enterprises never came to control more than 20 percent of tradable economic activity in the CEECs, excluding perhaps Hungary, before the socialist system collapsed An alternative measure of the relative size of the two transitions can be gathered by looking at the drop in total employment during the first decade From 1975 to 1 985, Spain, which suffered more than Portugal, shed 2 million workers out of an ini- tial labor force of 12 million-that is, the size of the force dropped by about 17 per- c;ent. Among CEECs, the one with the smallest percentage drop in employment from 1990 to 2000 is Poland (-11 percent), followed by Latvia (-14.5 percent), the Czech 52 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Republic (-15), and the Slovak Republic (-16). All others have larger percentage drops, with Bulgaria the most affected (-33 percent loss in total employment). We move next to the contribution of TFP. Averagmg over the whole CEECIO, TFP changes have contributed to growth of about 2.2 percent per year out of a total of 2.5 percent per year (88 percent). This result is not unusual. With few exceptions, TFP accounts for most of the growth in per capita output. However, this average masks sub- stantial differences among countries. For example, in Latvia, Lithuania, Poland, and Slovenia, TFP growth would have implied GDP growth in excess of 4.0 percent, had the two production inputs remained constant, while in Bulgaria, the Czech Republic, and Romania the contribution of TFP changes to growth has been either negative or negligible. For Hungary and Poland, two of the largest recipients of FDI in the group, the TFP contribution is positive but not large. While this heterogeneous behavior may require further country-by-country investigation, what we learn from this exercise is that the CEEC 10 transition process has not diverged from the growth accounting rule: GDP increases because TFP does, whatever the cause of the latter may be. As mentioned, a natural candidate behind the (positive) change in TFP is FDI. One would expect this to be especially true for countries that had been sealed from the rest of the world for almost 50 years and that, at the start of the transition, were using obsolete and highly inefficient production techmques. In particular, we would expect the FDI contnbution to growth to be highly correlated, across countries, with the TFP contribution. It turns out that, for the sample under consideration, this is not much the case. The overall contribution of FDI to growth has been small, approxi- mately 0.2 to 0.3 percentage points per year. Furthermore, its cross-country correla- tion with TFP growth is not significantly different from zero. As pointed out already, some of the assumptions may have biased the estimated FDI contribution downward. Still, even if we double the estimate, the contribution of FDI remains quite small. It is more so when compared with the values estimated in fast-growing countries, such as Ireland or China. Maybe we should not be surprised by this negative result. After all, the CEEC1O have not been growing very fast during the past decade, not even during the last half of it. Some of them may be starting to display fast growth patterns, but it is far too soon to know for sure. CEECs have come close to completing the transition, shutting down inefficient factories and farms, displacing a large fraction of labor force mem- bers from their former occupations, and starting to adopt advanced and efficient pro- duction techniques. This is clearly reflected in our growth accounting results. FDI has started to come in, but not yet at a very high rate. Apart from Estonia, CEECs have not yet established conditions as favorable to FDI as those found, for example, in Ireland. Hence, it would have been surprising to find that FDI's contnbution to growth in the CEEC 10 was comparable to that in Ireland. One last possibility is that the shedding of labor in the inefficient sectors is the key force driving TFP changes. This conjecture seems to receive mixed support from our estimates. On the one hand, it seems to be rejected by the aggregate employment data: positive changes in TFP are uncorrelated, across countries, with the magnitude RE-GIONAL POLICIES AND EU ENLARGEMENT 53 of the decrease in total employment. On the other hand, the correlation between TFP changes and the magnitude of the drop in agricultural employment appears to be sta- tistically significant. Given the nature of agricultural production in the CEECs, treat- irng the drop in employment in that sector as a proxy for the amount of labor displacement out of inefficient sectors does not seem unreasonable. Lacking better data on labor reallocation from protected to market-oriented activities, we must sat- isfy ourselves with this weak evidence. In an attempt to find some evidence on thie sources of cross-country variability in TFP changes, we look at where the CEECIO stand as far as trade liberalization, capital flows, privatization of firms, and banking and legal system reforms in rela- tion to the EU. Table 3-2 summarizes the relevant information We report three indexes: a capital flow restriction index, an index of structural reforms, and an index of legal proxies. The capital flow index refers to data for 1997 and ranges from -0.2 to 6, with 6 indicating the most restrictive institutions and negative numbers indicat- ing the presence of positive incentives to capital (in)flows The reform index refers to 1999 and weights price liberalization and competition policies (0.3), trade and exchange rate liberalization (0.3), and privatization and banking reform (0.4). A value of 1 corresponds to a market economy. The index of legal proxies refers to 1997 and equally weights the predictability of law and policies, the political stability and the secunty of properties, government/business interface, red tape, and the efficien- TABLE 3-2 INSTITUTIONAL INDEXES Capital flow Structiral Legal Counlry restriction refornms proxies EBulgaria 1 01 0 79 n a Czech Republic 0 05 0 90 3 40 Estonia 0 00 0 93 3 33 Elungary 0 62 093 3 30 Latvia 0 50 0 86 3 77 Lithuania 1 40 0 82 3 76 Ploland 1 03 0 86 3 66 Romania 190 0 82 n a Slovak Republic 0 75 0 90 3 78 Slovema 1 35 n a n a CEEC average n a n a 3 66 EU na 100 na ri a = not applicable Mote The capital flow index vanes from -0 2 to 6(6 is most restnctive) (Ganbaldi and others 2001) The refonm index veights pnce lihberalization and competition policies (0 3), trade and exchange rate liberalization (0 3), and privatiza- tion and banking refomn (0 4) A value of I Is a market ecoiiomy (Aslund, Boone, and Johnson 2001) The index of legal proxies weights predictability of law and policies, political stability and security of properties, govemoment/busi- iiess interface, red tape, and efficiency of government infrastricture The scale goes from 0 to 6, with 6 being the worst performer (Garibaldi and others 2001) S,ource Garibaldi and others 2001 54 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH cies of government infrastructures. The index ranges from 0 to 6, with 6 being the worst performer. In general, there appear to be limited differences along institutional lines. For example, the legal proxies index varies from 3.30 (Hungary) to 3.78 (Slovak Repub- lic), while the average is 3.66. By comparison, Russia and the other Commonwealth of Independent States (CIS) countries are at about the 4.0 level. The index of struc- tural reforms also indicates that all 10 countries are close to the EU, except Bulgaria (0.79), Romania (0.82), and Lithuania (0.82). Top performers in this category are Estonia and Hungary (both 0.93). Capital flows restrictions are now relatively low in all countnes: the least restrictive is Estonia-which actually provides incentives to FDI-with the Czech Republic a close second (0.05). The most restrictive are Roma- nia (1.90), Lithuania (1.40), and Slovenia (1.35). Once again, by way of comparison, Russia scored 1.81 in this scale and the other CIS countries 1.33. Changes in the quality of institutions may affcct TFP growth. Theory tells us that more efficient and more open countries may boost their wealth by pushing their pro- duction possibility frontier forward Empirically, and apart from a few anecdotal cases, the relevance of institutional factors largely remains to be proved. In the last column of figure 3-3, we confirm this fact by plotting the time series of the reform index. The relationship between the trend in the index and the trend in TFP is weak: most of the changes in the index across countries took place in the early 1990s, while TFP moves over time and displays trends, if any, in only the past five years. The size and efficiency of the financial system may constitute an important bottleneck for future growth in the area. Financial and banking systems are still somewhat underdeveloped. for example, the combined size of CEEC stock markets is about the size of the Irish stock market, and the turnover per year in the most devel- oped CEEC (Poland) is about equal to the tumover of two regular days of trading in Frankfurt. Similarly, the size of the total banking system in all the CEECs together is equal to one large EU bank, and many banks are too small to enjoy any positive scale effect. These features of the banking system are not, however, different from those existing at the time of accession in the most backward regions of the EU.' Com- pared with 10 years ago, capital markets are different in terms of efficiency and func- tioning. Nevertheless, the development of an appropnate financial system is the area where CEECs may benefit most from accession to the EU In sum, the liberalization indexes indicate that most of the CEECs are already similar to current EU members. Since the data are slightly outdated, we expect the divergences to have been further reduced in the last few years. Although institutional reforms may still be sought in the future, the need for reform does not appear to con- stitute a fundamental stumbling block in harmonizing EU and CEEC economies. Economuc theory and previous transition processes suggest that the sectoral com- position of VA and employment play a role in determining the impact of the trade- openng shock and that they are useful indicators of a country's stage of economic REGIONAL POLICIES AND EU ENLARGEMENT 55 development Differences between the EU and CEECs in the sectoral composition of VA may provide information useful to predict the shape of things to come. Table 3-3 presents this information. Three important features of the table deserve comment First, the agricultural sector in the CEECs is much more important than in the EU. On average, it contributes 2 6 percent of total VA and employs 4 8 percent of the active EU 15 labor force, while in the CEECs agriculture represents 10.6 per- cent of the total VA and employs 29.3 percent of the active labor force. Such per- centages also suggest that the agricultural sector of the CEEC IO is about 50 percent less productive (relative to average productivity) than the (already inefficient) agri- TABLE 3-3 GVA AND EMPLOYMENT BY BRANCH, PERCENTAGE SHARE Services, Services, Country Agrciclture M and M Utilities Construction private public G VA Bulgaria 145 19 1 50 36 429 13 1 CzechRepublic 39 278 40 74 41 1 127 Estonia 6 1 19 0 3 3 5 8 46 6 14 3 Hungary 4 1 25 5 3 8 4 6 42 6 16 5 Latvia 4 1 146 39 68 508 150 Lithuania 7 5 22 1 4 1 6 1 39 7 17 3 Poland 3 9 23 7 3 4 8 8 42 7 13 4 Romania 15 8 274 2 9 5 5 36 9 9 1 Slovak Republic 4 0 25 0 4.1 5 2 45 9 n a Slovenia 3 2 28 2 3 2 6 0 38 5 17 0 CEEC average 10 6 26 5 3 5 6 7 37 5 15 1 EU 26 238 50 54 410 211 Employment Bulgana 27 6 21 1 2 0 4 1 27 4 15 4 Czech Republic 5 0 28 6 1 6 9 3 319 17 6 Estonia 6 9 23 9 2 6 7 0 35 5 18 8 Hungary 75 280 20 62 288 277 Latvia 147 179 1 7 63 346 206 Lithuania 196 164 24 6 1 270 217 Poland 25 7 20 7 1 7 6 0 28 0 15 4 Romania 414 212 2 0 4 1 19 0 10 8 Slovak Republic 8 3 30 1 1 9 9 3 25 8 24 6 Slovenia 10 2 28 7 1 3 7.5 314 17 1 CEEC average 29 3 22 9 2 0 6 0 20 0 16 9 EU 48 203 30 54 38 1 270 M and M = manufaettinng and mining, n a = not applicable WVote Shares are computed using data for 2000, except for employment in Hungary and the Slovak Republic, where data for 1998 are used Source Eurostat 2001c,2002c 56 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH cultural sector of the EU 15. This is important in light of EU agncultural policies. In spite of its relatively minor weight in total VA, the agricultural sector is a political giant in CEECs because of its low productivity: many unproductive people are tied to the agricultural sector as the only source of income in the near future. Second, the service sector is much more developed in the EU than in CEECs. Private and public services produce about 62 percent of total VA and employ about 65 percent of the active population in the EUI5, while in the CEECI0 the share of VA produced by the service sector is about 53 percent and only 37 percent of the active population is employed there. Again, these numbers also suggest that the average service sector is fairly efficient in the CEEC 10. Third, the sectoral distribution of VA looks very sim- ilar to that of the most backward countries in Europe, but the distribution of employ- ment is still skewed toward agriculture and the industrial sector. Within the CEECs, several important facts need discussing. First, the service sector is already the largest contributor to GVA in all countries Percentages vary from a low 48 percent for Romania to a surprisingly high 66 percent for Latvia. Agri- culture is still an important source of GVA in the poor countries of the region- Bulgaria and Romania score around 15 percent-while in other countries the share of GVA produced by agriculture is around 4 to 8 percent. The share of manufactur- ing and mining is roughly constant across countnes and on the order of 22 to 25 per- cent; construction is important primanly in the Czech Republic and Poland, while the utility sector is everywhere a small contributor to GVA. Second, the growth rates of sectoral GVA have differed substantially across coun- tries. For example, over the period 1995-2000, growth of GVA in agnculture has been significantly positive in the Czech Republic (around 4.0 percent a year) and Lithuania (around 2.4 percent a year) and negative in Bulgaria (-2.0 percent a year) and Roma- nia (-3.0 percent a year). In manufacturing and mining, Lithuania experienced the largest growth rate in VA (an astonishing 15 percent a year), while Romania and the Czech Republic had the smallest growth rates. The utility sector has not shown much GVA growth, except in Bulgana, and the construction sector displayed significant VA growth in Estonia, Latvia, Poland, and Slovenia (between 7 and 10 percent per year). Finally, VA in services has grown at a significantly positive rate in all countries except Romania and the Czech Republic. Such large variations in the growth rate of VA, and especially labor productivity, across sectors point to different institutional and politi- cal constraints across countries and sectors. The data are too limited to establish an explicit correlation, but economic theory suggests that the largest productivity (and VA) gains should take place in those sectors in which competition is allowed. It would be worthwhile to look carefully into the industrial organization aspects of such large differentials to uncover the lmkage between labor market and competition policies on the one hand and sectoral growth and labor productivity performances on the other. Given the regional concentration of some sectors, particularly those in small countries, looking at which policies help or hamper the development of a specific sector in one country may be extremely useful to designers of growth-oriented regional policies. REGIONAL POLICIES AND EU ENLARGEMENT 57 Third, the sectoral pattern of employment also varies across countries. Agricul- tuare is the main reason for cross-country differences. There are two polar extremes- the Czech Republic (5 percent) and Romania (41 percent)-while in Bulgaria, Poland, and Lithuania employment in agriculture is still around 20 percent. Employ- rnent in manufacturing and mining varies between 16 and 29 percent, while con- struction employs between 4 and 9 percent of the active population. Finally, the service sector employs 50 percent or more of the population in all but Bulgaria, I'oland, Romania, Lithuania, and Slovenia. Fourth, employment growth in the past five years displays the typical pattern of transition countnes: employment growth in agriculture was negative in all countries but Romania; employment in manufacturing and mining declined on average by 15 to 20 percent; and job creation has been most intense in the construction and ser- vices sectors. Employment in construction mcreased in two of the three Baltic states and in Slovenia (3 to 4 percent a year), and the growth in the service sector was sig- nificant in every country except Romania and Bulgaria, with Poland's being the high- est (3 percent a year). It is important to look at the relationship between labor productivity and wage by sector both in relationship to the EU and within each country to identify the sectors for which employment growth is more probable. Table 3-4 presents this information. tOn average, labor productivity and wages in the CEECs are around 41 to 42 percent Of the EU level. However, sectoral heterogeneities are important. For example, wages in agriculture are more than twice the productivity level, while the opposite is true in all other sectors, with construction being the sector where unit labor costs are the low- est. In substance, the agricultural sector is a highly subsidized sector. Productivity is TABLE 3-4 SECTORAL PRODUCTIVITY AND WAGES BY BRANCH, RELATIVE TO EU Services, Total Country Agriculture M and M Construction private economy Bulgana 37/36 20/21 29/27 48/29 25/24 Czech Republic 88/81 53/46 72/74 59/82 58/60 Estonia 46/47 26/26 41/32 49/45 37/33 Hungary 77/75 49/41 54/43 83/75 58/50 Latvia 12/36 29/26 39/30 36/33 27/27 Lithuania 26/32 30/24 45/39 46/33 30/28 Poland 13/64 38/38 68/47 52/46 38/44 Romania 24/n a 31/n a 53/n a 68/n a 32/n a. Slovak Republic 54/56 42/32 51/40 90/69 53/40 Slovenia 94/121 58/56 72/61 81/74 71/70 CEEC average 28/63 41/34 58/41 68/53 41/42 M and M = manufacturmg and mining, n a = not available Note Productivity and wages are computed using data for 1998 Source Eurostat 200 1c, 2002c, authors' calculations 58 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH highest in the private service sector-primarily because international competition is most effective in this sector-and lowest, as expected, in agriculture. If one excludes agnculture, the VA produced by each employed worker exceeds gross labor income by about 20 percent. Within each sector, enormous differences exist between countries: for example, in Slovenia, productivity and wages in agriculture are at the EU level, while in Latvia productivity is 12 percent of EU level and wages 36 percent. Similar disparities exist in the construction sector (compare, for example, Bulgaria with the Czech Republic). The private service sector also shows differences, but on average it is the one whose features are most similar to those of the EU. For example, in Hungary, the Slovak Republic, and Slovenia, labor productivity and remuneration in the private service sector are slightly above 75 percent of the EU average. In sum, on the one hand the sectoral composition of employment and VA in the CEEC 10 is substantially different from that of the EU 15 and resembles that of the most backward EU countries. On the other hand, its dynamics are those that always char- acterize the development process of backward countries. Within the group, similar- ities and differences coexist. In all countries, the service sector contributes most to GVA creation, followed by manufacturing and agriculture. Growth patterns reflect national idiosyncrasies, but, to a large extent, growth in VA is driven by growth in the service sector. Around 50 percent of the population in every country is already employed in the service sector, and this share is increasing over time. Manufactur- ing and mining, and especially agriculture, are shrinking as a result of the scrappmg of old enterprises The speed at which this sectoral adjustment takes place is differ- ent across countries because of both comparative advantages and policy choices. Most important, employment in agriculture is still very high in 5 (Latvia, Lithuania, Poland, Romania, and Bulgaria) of the 10 states. The evolution of the agricultural sec- tor presents some worrying features from the viewpoint of enlargement and economic growth. The generalized presence of a large wedge between labor productivity and labor income is important in the light of current EU15 policies. This is both cause and effect of the slow expulsion rate from agriculture and the parallel slow produc- tivity gains Unless local and EU support programs perpetuate this anomaly, in the years to come a substantial amount of the labor force should be expelled from farm- ing. The construction sector has been an important (temporary) source ofjob creation in some countries, and its low unit labor costs indicate that it may also be able to absorb redundant agricultural employees in the near future. This sectoral shift is, however, unlikely to boost average labor productivity, given the low productivity of this sector. The Czech Republic, Hungary, and Slovenia appear to be the countries where increments in service employment are more likely to occur in the future. Since these are also the countries in which the private service sector is more productive, we should expect sectoral reallocation of labor to be more conducive to labor produc- tivity convergence in these countries than in the rest of the CEECs. RElGIONAL POLICIES AND EU ENLARGEMENT 59 Regional Inequalities in the CEECs Regional inequalities are not very large in the CEEC10; to some extent, they are smaller than or at most comparable to those already present in the EU 15. There are two reasons for this- most CEECs are small, in size and population, and this fact lim- its heterogeneity among reasonably sized internal territorial units, the high level of inequality between countries (one to three in per capita GDP) dwarfs the within- country differences This is a crucial fact to keep in mind when thinking about eco- nomic growth and convergence in the CEECs: regional and national economic convergence are, to a first approximation, the same problem. The basis for our statement is simple: eight of the CEECIO can be treated in the same way as moderately large NUTS 2 units of the EU. For example, Lithua- nia, which is the median among these eight, is only 65,300 square kilometers, and the population is less than 4 million people, smaller than Catalonia (Spain) and roughly equal to Veneto (Italy). The largest of these eight is the Czech Republic, which is equal in population (10 million) to Bayem (Germany) or the Regione Lom- bardia (Italy) and much smaller than Nordrhein-Westfalen (Germany). Obviously, one can try to find large differences within these countnes by looking at smaller ter- ritorial units There is no doubt that, with enough effort, one may succeed. Never- theless, one needs to try hard to find differences larger than those characterizing Italian and Spanish provinces, for which both unemployment and per capita GDP ratios between highest and lowest reach levels of 6 or even 7. In Hungary and the Czech Republic, for example, the ratio between the maximum and minimum unem- ployment rate at the regional level is about 4. Igor Strmsnik's chapter confirms this fact for the case of Slovenia, another CEEC comparable to an average NUTS 2 region. Slovenia is divided into three areas; the ratio in per capita income between the first and the second is 1.38, while that between the second and the third is 1. 10. No matter how one twists it, these are small differences, especially if one takes into account that the largest share of the population lives in the richest area around Ljubljana. The remaining two countnes, Poland and Romania, are fairly large. In Poland, 38.5 million people live on a piece of land that is approximately the size of Italy. Romania is less densely populated, with only 22 million people, but the country is roughly the size of the United Kingdom. For these countries, comparison with NUTS 2 units is inappropriate and care should be exercised in interpreting the relative num- bers. Nevertheless, while regional disparities exist in these two countnes, they are small relative to those in several of the EU countries. For example, the unemploy- ment rates in Poland average 16 percent (using 2001 data), ranging from a maximum cf 25 percent in Warmi#msko-Mazurskie to a minimum of 12 percent in the metro- politan Warsaw area of Mazowieckie. Internal labor migration is low, but it is already substantially higher than that within EU countries with similar or greater employ- ment and income per capita differences. 60 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Hence, the extent to which regional disparities are a problem for the CEECs is smaller than that for some of the EUI 5. Furthermore, regional disparities are relevant only in countries (such as Poland and Romania) that are large enough for regional comparisons to make sense. For the rest, concentrating on regional income inequali- ties would be tantamount to asking whether provincial or intercity inequalities in, say, Lombardia or Andalucia are important for aggregate economic growth and require intervention via some specific structural policies. For example, the city-region of Prague, in the Czech Republic, has a population of about 1.5 million people (about 15 percent of the country's total) and enjoys an income per capita of 122 percent of EU average and an unemployment rate of 3.4 percent. The other eight Czech regions host the remaining 8.9 million people, have 49 to 56 percent of the EU average m terms of income, and have unemployment rates between 5 and 15 percent. These are significant vanations, although not much different from those existing between any capital city in the EU and its countryside. Should this inequality be a cause of major concern for growth-onented policymakers? If so, then every EU 15 country, with the exception perhaps of Luxembourg and the Netherlands, is a problem. Summary We would like to highlight a few summary points of this general overview. First, despite extremely different initial conditions, we do not observe dramatically differ- ent growth rates in GDP per capita or labor productivity between CEECs and the EU. This is particularly true for the past five years, after the end of the first transition period and in the face of an already very high degree of openness to trade with the EU (as documented in the next section). Barring dramatic changes in fiscal and labor market policies, it is hard to forecast a very rapid convergence toward the EU aver- age levels of per capita income and labor productivity. Second, the CEECs are not homogeneous. Vastly different current levels and potentials can be observed. Treat- ing the 10 countries as if they were one single unit is likely to produce tensions as well as contradictory policy prescriptions. Third, investment rates in the CEECs are not especially high, which may prevent a rapid convergence of the capital/output ratio of these countries to the EU average. FDI does not appear to have played a big role in this respect, at least so far. Nevertheless, its upward trends during the past five years may lead to pleasant surprises should enlargement proceed along its course. Fourth, labor market incentives are grossly distorted, probably as a consequence of the harsh transition period. Still, the labor markets in CEECs are not much worse than those of many EU countries. Fifth, TFP growth seems to be the main driving force behind output growth, with the exception of the Czech Republic and the Slovak Republic. Sources of TFP growth are always mysterious. The data show a weak rela- tion to FDI flows and to the rate of employment reduction in agriculture, but not to recent changes in institutions, a factor often considered to be Important. The amount of time since such changes took place and the dubious quantitative nature of some indexes may be the reason for this irrelevance Sixth, the sectoral composition of VA FIEGIONAL POLICIES AND EU ENLARGEMENT 61 and employment in the CEEC1O resembles that of the backward countries in the EU and is still far from the average. This is not surpnsing. More attention should be paid to other distinctive features of the CEEC I0. The important role of agriculture in 4 of the 10 countnes, its relative inefficiency, and the inability of more advanced sectors to employ the excess labor expelled by agriculture appear to be key features, at least Irom a social point of view Substantial heterogeneity, both within and between CEECs, in sectoral productivity suggests that a fairly large process of intersectoral and international adjustment, following comparative advantage lines, still needs to lake place. In this and many other respects, CEECs are not different from the poor regions in southern Greece, Italy, and Spain, and a number of common insights apply to both. Seventh, regional income disparities exist, but their relevance should be played down relative to the differences between countries, which appear to be much more substantial. Furthermore, and with the sole exceptions of Poland and Romania, one should look at the CEECs as territorial and economic units altogether equivalent to the current NUTS 2 units of the EU 15. Finally, nominal and Maastricht indicators are already in line with those of the EU, but, except in extreme and remote (it is hoped) cases, we do not expect such factors to play a major role in determining the path of future income inequality between the EU and CEECs. Initial Conditions, Former Newcomers, New Steady States The EU has experienced several enlargements since its creation. During 1981-86, three countries, all poorer than the EU average, were admitted' Greece in 1980-81, and Portugal and Spain in 1985-86. We use this experience to examine two issues. (a) how the CEECs compare, in relation to EU averages, with the three earlier entrants; and (b) whether the experience of Greece, Portugal, and Spain can teach us something about the future ofthe CEECIO in the EU. The punch line is, althoughthe historical circumstances are different, macroeconomic conditions in CEECs resem- ble very much those of Greece, Portugal, and Spain at the time of accession. On the ground of similar initial conditions and accession to the same free trade area, and under the assumption of similar national and supranational policies, we find that the differential effect ofjoining the EU will be small (in terms of both levels and growth rates), so that current inequalities are likely to persist for a few decades. Basic Statistics Our claim is that the economic systems of the CEECI0, and in particular those of the first accession group, are comparable to those of Greece, Portugal, and Spain at the time of their accession to the EU. One major difference is the number of people involved. fewer than 60 million in the 1980s, and more than 100 million this time. A second difference is the number of countries involved, 10 versus 3, and the CEECs' much smaller average size, which generates procedural and administrative complex- ities probably not encountered before Similanties, on the other hand, abound Like 62 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH the earlier three, the CEEC IO have been part of "cultural Europe" since at least the eighteenth century and are currently emerging from a long period of political repres- sion, autarchy, and state control of the economy. Also, in both cases (with Greece as an exception), about a decade had elapsed since the previous regime collapsed, and a fair number of changes have already been implemented in the accession countries. Structural and socioeconomic features are also very similar, as is widely docu- mented (see, for example, Martin and others 2002, World Bank 2001 c, 2002, for pub- lic expenditure and its composition, education, demographic evolution, road and transportation systems, public pension expenditure, research and development invest- ment, size of the information technology sector). We will therefore concentrate on a restricted number of aggregate statistics, which can provide a parsimonious measure of the "distance" between CEECs and Greece, Portugal, and Spain at the time of accession. Ideally, we would like to come up with a single number. Because we do not have a rationale for choosing particular weights, we have decided to present the various components of our index separately in table 3-5 The factors we use to mea- sure similarities and differences are the openness of the economy (measured by exports plus imports over GDP), the share of employment in agriculture, GDP per capita relative to the EU average, and labor productivity relative to the EU average. For CEECs we take 2000 as our benchmark, for Spain and Portugal we average the conditions existing in 1985 and 1986, and for Greece we average the conditions exist- ing in 1980 and 1981. One could consider many other indicators, but we believe these four provide a reasonably good synthesis of the kind of information we seek. Fur- thermore, they are highly consistent with the information in the quoted studies, as well as that reported earlier, in the Initial Conditions in the Enlargement Countries section. A few aspects of the table need to be commented upon First, the share of employment in agriculture (our preferred measure of backwardness) is compara- ble between the two groups of countries. Portugal and Spain had shares of 22 and 15 percent, respectively, while the average of CEECs (excluding Romania) is about 20 percent.2 Similar results would have been obtained had we used other sectoral indexes of the two groups of economies-for example, the share of manufactunng in total VA of Greece, Portugal, and Spain was 15, 29, and 26 percent, respectively, and in CEECs it averages 26.5 currently. Second, the range of GDP per capita values is also very similar Spain was at a level similar to Slovenia at the time of accession, while Portugal and Greece were approximately at the current level of Hungary and the Czech Republic. Latvia, Lithuama, Romania, Bulgaria, Estonia, and Poland fall below the range of relative per capita GDP determined by the last three entrants. Again, the picture would have not changed if, instead of levels, we had been using growth rates of per capita GDP in our comparison. The growth rate of GDP was a meager 0. I percent in Greece at the time of admission, while Portugal and Spam displayed healthier values at 4.1 and 3.2 percent, respectively. The range for the annualized average growth rate of GDP in the CEEC 10 over the period 1996-2000 was -1.5 (Romania) to 5.1 (Estonia), with REGIONAL POLICIES AND EU ENLARGEMENT 63 TABLE 3-5 INDEXES OF SIMILARITY Countrv Openness Employment in agricultuire GDP per capita LP (Greece 1980-81 0 42 n a 0 55 0 52 Portugal 1985-86 0 56 0 22 049 044 "pain 1985-86 029 015 062 095 Bulgana 1 28 0 27 0 25 0 28 Czech Republic 1 44 0 05 0 59 0 47 IEstonia 0 87 0 07 0 41 0 34 liungary 1 23 0 07 0 53 047 Latvia 1 02 0 14 030 030 Lithuania 1 06 0 19 0 30 0 26 I'oland 0 63 0 25 0 41 0 34 IRomania 0 75 041 0 27 0 15 Slovak Republic 1 60 0 08 0 49 0 47 Slovenia 1 20 0 10 0 73 0 50 ii a = not applicable, LP = labor productivity aVote Openness is measured by exports plus imports over GDP For Greece, Portugal, and Spain we averaged over the iwo years For CEECs the values refer to 2000 The employment share and labor productivity for Greece, Portugal, and Spain are computed using the OECD Structural Statistics for Industry and Services database and the OECD Structural Analysis Database for Industrial Analysis Values for the CEECs come from previous tables GDP per capita is com- pited using the Eurostat Regio data set 7 countries out of 10 growing at 3.4 percent or more per year. Furthermore, the absolute values of GDP per capita at PPS also match For example, in 2000 all CEECs, except Bulgaria and Romania, were ncher than Greece, Portugal, and Spain -at the time of their admission; Bulgaria and Romania matched Greece and Portugal and were no more than 10 percent below Spain Third, except for Spain, relative labor productivity is also comparable. To explain the Spanish outlier, it is useful to remember that at the time of accession, the unemployment rate had reached 23 percent, with falling participation rates, and that Spain's labor productivity index decreased, relative to the EU average, after acces- sion. As a further element to evaluate relative labor productivity potentials, it may be worth pointing out that schooling rates are uniformly higher in the CEECs than they were in Greece, Portugal, and Spain in the early 1980s. The two sets of countries differ very much in terms of openness, but the differ- ence is all in favor of the CEEC group. CEECs are much more open than previous poor newcomers. The average index of openness is two to three times as large in the CEECs as it was in Greece, Portugal, and Spain when they joined For small economies (the Czech Republic and the Slovak Republic), the numbers are larger. In fact, even when we compare the current degree of trade openness of Greece, Portu- gal, and Spain with that of the CEECs, the latter come out ahead on average. More 64 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH important, between 40 and 70 percent of CEEC trade takes place with the EU already; all these percentages have been growing steadily over the past five years. Capital flows are also larger than those for Greece, Portugal, and Spain at the time of acces- sion, even adjusting for the fact that gross flows are much larger today. As we have seen, on average FDI to CEECs has been above 4.0 percent of GDP during the last few years. FDI as a percentage of GDP in Greece, Portugal, and Spain in 1986 was at 1 2, 0.7, and 1.3, respectively (Martin, Velazquez, and Funck 2001). The post-1990 labor market dynamics in the CEEC 10 strongly resemble those of Spain and Portugal after their transitions began. Rapid expulsion of workers from agriculture and from traditional industrial sectors (total employment m Spain dropped from 12.6 million to 10.6 million between 1975 and 1985), equally fast increase in unemployment (from 6 to over 20 percent in Spain, to around 16 percent in Portu- gal), reduction in the labor force participation rates (especially among young and old individuals), and creation of a substantial and lasting stock of long-term unemployed people at around 40 to 50 percent of total unemployment are important common fea- tures of the first 10 years of both transition processes. Similarly, productivity per employed person increased while income per capita decreased initially, to rebound only after 1986. In fact, relative to the EU average, it took until 1991-92 for Spanish per capita GDP to go back to where it was before 1975 It has taken less time for the best-performing CEECs to achieve the same result. These facts, together with those illustrated in the second section of this chapter, strongly suggest that initial CEEC conditions are similar to those of previous entrants and that the costs of and gains from joining the EU will probably be comparable to those experienced by the previous three newcomers. Different Historical Circumstances History matters. Having stressed the similarities between the socioeconomic con- ditions of the CEECs and those of earlier entrants into the EU, we should elaborate further on the differences briefly mentioned above. As our goal is to evaluate the extent to which the performances of earlier entrants can predict those of the new can- didates, we should limit ourselves to those factors that may influence, in one direc- tion or another, the process of economic integration of the CEEC1O into the EU. Overall, the CEEC 10 will have some advantages that the former newcomers did not. Paradoxically, while these advantages may ease adaptation to EU market conditions, they may also reduce the size of the gains the CEECs should expect from accession. The first important difference is that the CEECs will enter a larger and richer market than the early entrants did. This should facilitate the accession process, at least in the presence of a reasonable degree of flexibility on the part of CEEC firms. The second crucial difference is that the current level of economic integration within the EU is much higher than it was in the 1980s. A startling measure of such difference is the amount of European FDI already flowing to the CEEC 10, as a per- centage of their GDP, in companson with the amount of EU-originated FDI enter- REGIONAL POLICIES AND EU ENLARGEMENT 65 ing, say, Spain before 1986. For the latter it was, at most, half a percentage point a year, while for the best CEECs it exceeds 5 percentage points. Total capital or trade flows follow a similar pattern. This feature should benefit CEECs and should facili- tate their integration into the EU. On the other hand, unless accession brings dramatic policy changes, this fact should also make the postaccessionjump in FDI smaller than the one observed after previous enlargements. A similar argument applies to trade: trade openness of CEECs is remarkably higher than that of earlier entrants Therefore, comparatively speaking, CEECs have less to gain from the general tariff and baffler reduction that comes with accession. However, many tariff and trade restrictions are still in place. Given the CEECs' sec- toral heterogeneity, one could expect certain sectoral gains to be comparable to those of earlier entrants even if the aggregate effect may be smaller Finally, exchange rate risk is now gone, at least within the EU 15 area. If CEECs manage to satisfy the Maastricht criteria, they wouldjoin the EU and (possibly) adopt the euro. Given that the share of EU trade within total CEEC trade oscillates between 45 (Bulgaria and Lithuania) and 68 percent (Estonia, Hungary), the absence of exchange rate risk gives the CEEClO a huge advantage relative to earlier entrants Relative Steady States: Old and New Smce initial conditions in the CEECs are approximately the same as those of the for- mer newcomers, and since, given current policies, we find it likely thatjoining the EU will have the same effect on these countnes that it had on Greece, Portugal, and Spain, we can estimate where CEECs will be in their postaccession steady state relative to the EU average. Like all reduced-form econometric exercises, this has obvious limitations First and foremost is the assumption of policy and behavioral invariance postadmis- sion What this means is that we assume that no CEEC will be able to replicate, for example, the policies that Ireland has adopted in the past 16 years. Rather, we assume that CEECs will behave, more or less, like Greece, Portugal, and Spain did during the last 16 years; that the EU structural policies toward poorer regions will remain more or less unaltered; and that the overall degree of trade liberalization within the EU and between the EU and the rest of the world will not change. But there are no guarantees. Spain and Portugal have progressively modernized and liberalized their own internal product and labor markets; slightly reduced labor taxation; improved capital mobil- ity; and implemented a fairly large, albeit not overwhelming, privatization process. Furthermore, inflation rates have declined, budget deficits have been brought under control, public debt has been restrained, and the euro has been adopted. Making a cetens panbus assumption implies that the CEEClO will accomplish something sim- ilar, at least in relative terms, in the next 10 to 15 years While feasible, this achieve- ment will require keeping up the pace of market-oriented reforms. With these caveats in mind, we now illustrate our econometric procedure. We plan to draw from the information we have available to infer something about the (asymptotic) distribution of growth rates for CEECs postaccession. We construct 66 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH three scenarios. The first, the no-change scenario, assumes that the current histori- cal conditions (including current cross-country heterogeneities) will be perpetuated indefinitely. This scenario implicitly assumes that the gains from free trade and inte- gration have already been absorbed over the past 10 years and that entering the EU will not change the underlying trend behavior of the CEEC 10 Rather than taking this assumption literally, we consider it a benchmark against which to evaluate the long- term gains that these countries may enjoy once they are in the EU. For this scenario, we estimate a simple autoregressive (1) model for relative (to EU average) income per capita/labor productivity for each country and, given parameter estimates, we pro- ject past behavior far into the future. Since the data set is short, estimates of the steady states are biased. Therefore, these numbers should be taken as indicating more qual- itative than quantitative tendencies.3 In the second scenario, the level-effect scenario, we assume that after joining the EU, CEECs will settle into a steady-state position that is similar (in terms of the parameters regulating the asymptotic distnbution of relative growth rates) to that of Greece, Portugal, and Spain since they entered the EU. Theoretically, one can justify this scenario by using either endogenous or exogenous growth models in which coun- tries that are similar in terms of economic fundamentals face the same distribution of long-term growth rates. Notice that, because of the normalization by the unknown EU growth rate, this reduced-form statistical model makes absolutely no assumption as to the source of growth. It is consistent with old, new, and even postmodem growth theoretical models as long as they satisfy a probabilistic version of the tertium non datur assumption: countries that are identical at the beginning will draw their final outcomes from the same probability distribution. EU accession would therefore mean, for the CEEC 10, a transition to the same distribution of growth rates faced, since 1986, by the three earlier entrants. To implement this idea, we estimate steady states for NUTS 2 regions of Greece, Portugal, and Spain using data after their EU accession. We then use this information as a prior for the AR model we estimate for the CEEC 10. The prior takes a simple form: we assume that, postaccession, the distribution of balanced growth rates of the CEEC 10 (scaled by the EU average) will have the same mean and variance as the dis- tribution of poor EU temtorial units after they joined the EU. Clearly, we do not pre- vent miracles (or busts) from happening; we simply require similanties in the support of the two distnbutions. This scenario does imply, though, that CEEC miracles will not be larger or more frequent than those experienced by poor EU regions, and simi- larly for busts. We examined whether taking Greece, Portugal, and Spain as a whole or (to have exactly comparable GDP per capita) the poorest 22 regions in these coun- tnes4 makes a difference for our calculations We also examined whether the choice of the short sample made a difference-by concentrating totally on data for Greece, for which almost 20 years of data exist. Neither of these two variations made a sizable difference. The mean steady states we obtamn oscillate, depending on the assumptions made, between 60 and 75 percent of the EU average, and the dispersion of the regional REGIONAL POLICIES AND EU ENLARGEMENT 67 steady state ranges from 45 to about 90 percent. We capture these tendencies by assuming a normal distribution centered at 65 percent of the EU average with a stan- dard error equal to 15 percent. In the third scenano, the growth-effect scenario, we assume that joining the EU will proportionally boost the growth rate of CEECs by the same factor it boosted the growth rates of Greece, Portugal, and Spain after their accession. This scenario assumes a linear production function in which the asymptotic growth rate is country specific.5 The scenario assumes that when two countries are exposed to the same change in their fundamentals, a common proportionality factor affects country- specific growth rates. As a trivial example, think of an A,Kmodel in whichA, #A2 are the production function parameters for the two countries. Joinig the trade area implies that a common proportionality factor, say m > 0, multiplies the A,, so that ,BA,K is the aggregate production function for country i after accession Obviously, this is just an example, a more complicated model in which the parameter A, is endogenized by means of external effects, public capital, agglomeration effects, rate of technological innovation, and so on would also be consistent with our statistical procedure For the case at hand, joining the EU will make the country-specific growth rates in the CEEClO jump, so that for a given initial condition, output will grow at a permanently higher (or lower if D < 1) rate. To implement the idea behind this scenano, we esti- mate average and dispersion of growth rates using regional data for Greece, Portugal, and Spain before and after accession We use these two distributions to estimate the common proportionality factors f3 (for the mean) and f3 (for the standard deviation) As before, we use this information as a prior for the AR model we estimate using CEECIO data. Since we have very few data points before accession, estimates of the jump are very imprecise. The mean effect is, however, very small (1.009), probably because most of the growth gains in Greece, Portugal, and Spain predated EU acces- sion. We take into account the uncertainty present in the data by letting the dispersion of the potential jump be relatively large (the maximum range we allow is [1.0, 1.03]) Table 3-6 provides a summary of the results. For each scenario, the first column refers to relative GDP per capita and the second to relative labor productivity. We would like to emphasize two results. First, in the absence of any structural change (no-change scenario), we should not observe significant variations either in the rank- ing or in the level of GDP per capita of the CEECIO relative to the one observed in 2000. In other words, CEECs are already close to their steady state. The only two countries that, at the steady state, seem able to reduce significantly their distance from the EU are Poland (currently at 41 percent) and Slovenia (currently at 73 percent) Second, adding the information about the behavior of the poor EU countries after accession does not radically alter the estimates of the relative steady states. The dis- tribution of relative steady states in the level- and growth-effect scenarios is only marginally different from the one estimated for the no-change scenario. This follows from the fact that the growth patterns experienced by CEECs dunng the 1 990s were not so different from the growth patterns experienced by Mediterranean countries 68 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH TABLE 3-6 ESTIMATED STEADY STATES No-change Level-effect Growth-effect scenario scenario scenario Country GDP LP GDP LP GDP LP Bulgaria 023 026 034 0 35 0.37 0.37 Czech Republic 0 61 0 50 0 50 0.56 0 51 0 53 Estonia 0 33 0 33 0 39 0 38 0 40 0.39 Hungary 0 48 0 46 0.45 0.49 0 45 0 45 Latvia 0 26 0 25 0 35 0.36 0 37 0 36 Lithuania 0 28 0 26 0 37 0.35 0 36 0.37 Poland 0 48 0 34 0 46 0.39 0 47 0.37 Romania 0 27 0.23 0 42 0.31 0.38 0 33 Slovak Republic 0 49 0.48 0 45 0.49 0 46 0 48 Slovenia 0 80 0.61 0.76 0 68 0 74 0 60 CEEC average 0 38 0.38 0 41 0 45 0.42 0.41 Note Estimates obtained in the columns "Level-effect scenano" and "Growth-effect scenario" are computed usrng a Bayesian procedure that weights infonnation contained in domestic data and umformation in the time senes of poor EU regions after they joined the EU In the first case, it is assumed that the distnbution of steady states to which the CEECs will belong is the same as the one of poor EU units In the second, it is assumed thatjoining the EU has a level effect on the growth pattem that is the same for CEECs as it was for poor Spanish and Portuguese regions Source Authors' calculations after theyjoined the EU Our estimated gain for the average CEEC is about 3 to 4 per- cent (GDP per capita would go from 38, to 41 or 42 percent of the EU average), but this is not statistically different from zero. There are some sizable gains for the poor- est of the CEEC 10; for example, the steady-state levels for Bulgaria and Romania will be about 50 percent higher if they join the EU, and they are predicted to do slightly better in the growth scenario. Conversely, the nchest among the CEEC 10 are predicted to do worse after accession than they would do otherwise; see, for exam- ple, Slovenia. The reason behind this reversal of fortunes should be kept in mind when discussing policy implications. Greece, Portugal, and Spain have performed well but far from spectacularly after accession and only marginally better than they had in earlier decades. The strongest among the CEEC 10 have done well in the last half of the 1 990s; In fact, they have done better than the earlier entrants did, on aver- age, after accession. The opposite is true for the weakest five among the CEEC10. Hence, applying to the CEECs the distribution of growth rates experienced by Greece, Portugal, and Spain induces a "reversion to the mean" effect in which the currently fast growers grow more slowly and the laggards' growth accelerates. Closeness to the steady state, lack of measurable effects from joining the EU, and, absent changes in national policies, relatively similar "engines of growth" lead us to predict that the EU and CEECs will grow at fairly similar rates in the next couple of decades. Simi- lar steady-state growth implies constant ranking in the distribution of GDP per capita REGIONAL POLICIES AND EU ENLARGEMENT 69 and persistence of current inequalities. Integration and extension of the current EU regional policies to CEECs will not sweep away income inequalities. National poli- cies, if anything, must do the job. Why compare CEECs with Mediterranean countries and not Ireland? We have shown that there are similanties in the historical circumstances of the two experi- ences that make the exercises meaningful. One could, however, argue that Ireland in 1976 was as backward and as closed as Mediterranean countries but has experienced extraordinary growth over the past 10 years. Clearly, we would have had a much rosier picture had we assumed that the growth experience of CEECs after accession would resemble that of Ireland in the 1990s. However, Irish growth did not come in a vacuum: as Barry tells us, national policies make a difference. If CEECs are will- ing to adopt the policies that Ireland implemented in the last decade, our current esti- mates are on the pessimistic side. Accession, however, does not guarantee good national policies bad govenmments exist both within and outside the EU. Since his- tory and social norms matter, we find it unlikely that the CEECs will enact the kinds of sweeping changes in labor, financial, and capital market conditions that would jus- tify the alternative scenario This sobering fact should not take our attention away from the fact that there are differences in the dynamics of per capita GDP within CEECs and that these differ- ences may be relevant in projecting current conditions far ahead. For example, the regions of Prague, in the Czech Republic, and Bratislava, in the Slovak Republic, have enjoyed sustained growth dunng the past five years, which pushed their regional GDP per capita respectively to 122 percent and 99 percent of the EU average. Nev- ertheless, if we exclude some dynamic region, typically centered around the capitals (Prague, Bratislava, Ljubljana, and Budapest), growing at a reasonable pace relative to the average of the EU, the majority of the CEEC inhabitants are expected to live in the future with an income per capita that is about 55 or 65 percent of the EU average. Finally, since the dynamics of labor productivity in CEECs are comparable to the GDP per capita, steady-state calculations are unaffected if we substitute one for the other. Projecting current labor productivity into the future implies that the Czech Republic, Poland, and Slovenia look more dynamic in terms of GDP per capita than in terms of labor productivity, but the uncertainty in the estimates does not permit us to make statistical statements. Bringing in information from the previous newcomers in the EU does not change the pattern we have described. If anything, the higher rel- ative productivity level recorded for Spain at the time of accession implies a distnb- ution of steady states that is more spread out than the one of GDP per capita in the second scenario Old, Useful Lessons Probably the most important lesson from previous studies of economic growth in trade-integrated areas is that the predictions of "new growth" or "new trade" theory 70 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH models are comfortably rejected by the data These kinds of models almost unani- mously predict that trade openness combined with increasing returns and a variety of external effects will produce agglomeration phenomena, poverty traps, economic divergence, and increased inequality. Hence, when two differently endowed coun- tries start trading with each other, the richer or more advanced one "wins" while the other "loses." Victory takes the form of a higher rate of income growth (because of faster capital accumulation or faster rate of innovation), a concentration of produc- tive factors, or both. Agglomeration theories, which seem to be particularly popular among policymakers and technocrats involved with regional and structural policies, predict that capital and labor move toward where their complementary factor of pro- duction is more abundant, leadmg to a concentration of economic activity in a few privileged areas and leaving the rest far behind. Hence the need for active public inter- vention to prevent factors (especially labor) from moving around too much and for subsidizing economic activity in poorer areas where it would not otherwise take place. We are not aware of any histoncal experience of trade integration supporting this kind of prediction. In fact, all recorded episodes of increased trade openness, at the national or international level, have generated the opposite outcome: poorer areas have either strictly gained (in both absolute and relative terms) on the leader or kept distances roughly constant. What we have in mind here are the increasing trade and factor mobility among the 50 U S. states since the end of the Civil War; the increas- ing trade and economic integration among the initial 6 members of the European Economic Community (EC) since the 1950s and then, since the end of the 1980s, among the current 15; the recent successful integration of Canada, the United States, and Mexico in the North American Free Trade Agreement; and the almost 50 years of progressive and still increasing trade integration of Japan and Southeast Asian countries among themselves and with Europe and the United States. One is hard- pressed to find a single "loser" in any of these episodes It is quite important to stress that, in all but one case (the post-1980s EU), trade integration and increased factor mobility took place without any kind of regional, structural, or transfer policy meant to compensate the poor countries or areas for the losses of trade integration with the richer ones. While the extent to which trade openness and integration have generated convergence varies greatly from one situation to the next, divergence has never been observed at any reasonable level of spatial disaggregation. The latter is not a minor point. The European Community integrated trade first among six and then among a higher number of countries for about 20 years, practicing very few "compensatory" or "structural" policies (aside from the infamous CAP, the consequences of which are well known), and income differences sharply decreased both across and within the EC countries. Dunng the past 20 years the EU has increased dramatically the amount of funds invested in structural and regional policies, without any visible impact on the rate of economic convergence within countries. Empirical estimates suggest, in fact, that regional convergence has come close to a halt just at the time FIEGIONAL POLICIES AND EU ENLARGEMENT 71 structural and cohesion policies have been introduced (see Boldrin and Canova 2001). East Asian countries have practiced almost no compensation or structural poli- cies. Still, their convergence to the average income of their trading partners (that is, EIurope, the United States, Canada, and Japan) seems certain The list could be con- tinued indefinitely, but it would become redundant. The message is clear: opening tip trade among regions that are economically heterogeneous does not lead to diver- gence, not spontaneously at least. Unfortunately, when looking at convergence data we have a strong tendency to see what our personal prejudices would like them to show. It is for this reason, we believe, that in spite of all the evidence to the contrary, the "agglomeration" hypo- thesis seems to be more popular than ever. Research produced at or around the Euro- pean Commission, especially the Directorate for Regional Policies, is an excellent example of this phenomenon. NUTS 2 and even NUTS 3 regions are often used to rneasure inequality At a fine enough level one can certainly find fairly large inequal- ities in per capita income or other measures of economic well-being. Less obvious is the conclusion that public policy aimed at eliminating economic differences at such a fine territorial level should be carried out centrally by the EU (violating the most elementary interpretation of subsidiarity) and that income transfers and subsidies are the appropriate instruments to make the poor regions grow faster than average. The second lesson we have learned is that trade integration facilitates economic growth but is far from guaranteeing it, especially if appropriate internal economic policies are not adopted Large amounts of regional aid may temporarily increase the income of recipient regions and in this sense postpone the need for serious structural reforms, but there is no evidence that they generate higher growth rates in the long run Empirical evidence has consistently shown that when reasonably large temtorial units are chosen for the analysis, opening up trade and allowing internal markets to vvork lead to a certain degree of convergence (see, for example, Ben David 1994) flow fast this "spontaneous" or "automatic" convergence takes place is still a topic of debate In fact, on average and across very many political systems and fiscal and mon- etary policies, the most likely result seems to be that convergence takes place in growth rates but not in levels countries that start ahead tend to stay ahead, even if dis- tances are somewhat reduced Somehow, a combination of initial conditions, factor endowments, and, most important, national policies seems to determine country- specific steady states (or balanced growth paths) to which individual countnes con- verge. Such steady states are different across countnes and are affected by trade policies, but trade integration alone improves the relative performances of the poorer countries by only a handful of percentage points relative to the ncher ones. In certain circumstances, convergence in levels does seem to take place, but such circumstances are rare. For example, Western European countries and Japan came very close to the per capita income levels of the United States in a period of roughly 30 years between the end of World War II and the oil crisis of the 1 970s. However, convergence was 72 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH not, and still is not, complete; in fact, since about the mid- 1 970s, the three parties (the EU, Japan, and the United States) have kept their relative positions approximately unchanged. For the "miracles" of Southeast Asia a large amount of convergence toward the U.S. level has taken place, but one has the impression that these countries converge to some relative steady-state position that is strongly determined by national characteristics. Portugal, Spain, and, to a much lesser extent, Greece have somewhat reduced their distances from the EU average income level since 1986, while Ireland has managed to overtake the EU average income level in a period of less than 20 years. One may legitimately wonder whether structural and regional policies are behind the miracles or, at least, the growth convergence we observed. We doubt that this is the case for several reasons. First, as mentioned above, no regional policies were ever implemented among Southeast Asian countries, and divergence was not observed. Second, one may think of the Marshall Plan-the historical analogy to the current European transfer policies-as key to the European convergence to the United States. However, both the financial size and the duration of the Marshall Plan are orders of magnitude smaller than those of the European structural policies or, at the national level, of the German transfers to the East German lander and the Italian transfers to the southern regions. If what it takes for convergence is a Marshall Plan, then Sicily and Calabria have received approximately 20 of them since the 1950s. Furthermore, in Boldrin and Canova (2001), we have shown econometrically that, at least in the EU15, the conjecture that regional transfer policies are behind the partial conver- gence episodes is not supported by the data. Regional policies, at least in the form implemented by the EU since the mid-1980s, made little difference on long-run growth at the regional level. It should be self-evident that this does not mean that the transfers involved with the structural policies made no difference for the countries and regions on the receiv- ing end. They certainly did and still do: receiving a nice yearly check for an amount between 2 and 5 percent of national income is valuable. The net of transfers, health and social insurance payments, and public expenditure in the EU is never above 25 percent of GDP. European transfers increase the funds for public expenditure available by 10 and 20 percentage points and by much more when we look only at public invest- ments. When used appropriately, these funds can help to ease social tensions, espe- cially in transition situations. When used inappropriately, they enrich unscrupulous politicians and those backing them. No wonder that transfers are most welcome by receiving countries, and members of the EU, old and new, are no exception to this rule. But to claim that transfers have made a difference for growth is an entirely dif- ferent matter. The valuable contribution by Angel de la Fuente to this volume should be inter- preted in the light of these observations. He shows that the Structural Funds flowing to Spain were mostly used for productive investments and that such investments have indeed been productive. That is, they contribute to gross national product, propor- tionally to their estimated share in the aggregate production function. This is certainly REGIONAL POLICIES AND EU ENLARGEMENT 73 correct: transfer of funds from the EU to Spain has helped the accumulation of pro- ductive infrastructures in that country. It would be really dramatic if it were other- wise. But the crucial claim is not that Structural Funds are wasted. The crucial claim is that they do not alter the long-term growth rate of the recipient regions in any sig- nificant way. In our statistical work we looked in many different ways for such an effect but could not find it Pedro Arevalo (2002) has carried out a painstaking and meticulous investigation of Spanish regional development since the late 1950s, using a high-quality data set of both provincial and regional human, public, and pnvate cap- ital stocks, and sectoral VA. He shows that TFP growth accounts for the lion's share cf economic growth and convergence across Spanish regions, with little left for pub- lIc and pnvate capital and a somewhat larger share for human capital. More impor- tant, he shows once again that, even at this very detailed and disaggregated level, there is no sign of a positive impact of Structural Funds on provincial and regional 1TFP growth rates. While speaking against current EU regional policy is a political taboo, other people have also started looking at the question. A recent paper by E.derveen, de Groot, and Nahuis (2002) is a prime example It is relevant here also because, in our view, de la Fuente quotes it erroneously as a piece of statistical evi- clence contradicting our earlier analysis. It does not do so; in fact, it reinforces it The authors use a statistical methodology that is quite different from the one we used in Eloldrin and Canova (2001) but reach the same conclusions: Structural Funds by themselves are ineffective. Their estimates show a statistically significant negative effect of Structural Funds on regional growth rates and convergence. They find a small positive effect of Structural Funds only for countries with the "right policies " P'lacing this "nght policies" condition does not work in Ederveen and others either: low inflation, low budget deficit, and a cohesive social policy do not make Structural Funds effective. Only high institutional quality and low corruption do. More inter- esting is Ederveen and others' country-by-country breakdown of this conditional effect (table 4-2): in all cohesion countries but Ireland the impact of Structural Funds is strongly negative even when one accounts for the variables measuring corruption and institutional quality. In fact, the impact is also negative in Italy and, in two cases out of three, France. The authors apply their methodology to the EU accession coun- tries and show that, also in these countries and even after conditioning for institu- tional variables, the likely impact of Structural Funds on convergence is negative Only openness, not surprising in the light of our earlier discussion, may help make the impact somewhat positive for some of the accession countries. Hence, contrary to de la Fuente's assertion, Ederveen and others (2002) reach conclusions quite sup- portive of our analysis.6 In summary, neither de la Fuente nor any other serious study we know of shows that the growth rate of TFP is significantly affected by structural policies Overwhelming evidence shows that long-term growth is the product of increasing TFP, notjust of capital accumulation. TFP growth is, indeed, a fairly mys- terious object. Nevertheless, no data show that regional TFP growth has been affected by Structural Funds. 74 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH A second point should be made. Structural Funds are transfers supported by dis- torting taxation, which implies a dead-weight loss. The dead-weight loss could be justified on public policy grounds if the social rate of return from the investments financed via Structural Funds were large enough to compensate both for this dead- weight loss and for the opportunity cost of the funds. The latter, after all, could have been used by private agents in other productive activities. Deciding what a reason- able social rate of return on public investment would be is hard, and we are not going to try to quantify it here. The Congressional Budget Office of the U S. Congress, for example, recommends a 10 percent real annual return as an absolute minimum for any public investment project. Has such a minimum rate of return been obtained by the investments financed by the EU Structural Funds? Even without capitalization, the accumulated investment financed by Structural and Cohesion Funds in Spain amounts to at least 40 percent of Spanish GDP. This is a benevolent choice, first because we are not capitalizing and second because the number would have been 70 percent for Portugal and more than 100 percent for Greece. Have Structural Funds increased the Spanish GDP by at least 4 percentage points each year? If sound empir- ical analyses can show numbers anywhere comparable to this figure, we will be will- ing to step back and revise our position on the effectiveness of Structural Funds. Until that is the case, we will keep our current position. Sometimes some poorer countries grow faster than richer ones. The debate is still open as to the magical mix of policy and circumstances that makes miracles happen. We do not have an answer to such a question, but we can come up with an incom- plete list of factors that help, that hurt, and that are more or less irrelevant. To illus- trate the point, look again at the four cohesion countries of the EU-Greece, Ireland, Portugal, and Spain-and at the Italian southern regions of Basilicata, Calabria, Cam- pania, Molise, Puglia, Sardegna, and Sicilia (that is, the Mezzogiorno). We can break the group into two, based on initial conditions. Back in 1960, Ireland and Spain were both at about 60 percent of the EU average in terms of per capita GDP, while Greece, the Mezzogiomo, and Portugal stood at 50 percent During the 1960s and until 1974 Spain liberalized trade unilaterally and adopted internal policies that, relative to earlier policies, were market oriented. By 1974 its income per capita stood at almost 80 percent of the European average, which is where it still stands now, after 25 years of oscillations in both growth rates and internal policies Irish relative per capita income stagnated for more than 20 years, without apparent effect from either the 1973 EU accession or the receipt of EU Structural Funds: by the mid- 1980s its income per capita was around 65 per- cent of the EU average. Since then the government has fully embraced free trade, low taxes, low public spending,7 and well-known competition-oriented policies. Its income per capita now exceeds 110 percent of the EU average. Look next at Greece, the Mezzogiomo, and Portugal. Greece stands now at about 65 percent of the EU average. This is exactly where it was 21 years ago, when it first joined the EU and REGIONAL POLICIES AND EU ENLARGEMENT 75 began receiving transfers: all the catching up that Greece managed to do since 1960 took place before it started to receive Structural Funds. It is also a fact that, since accession, Greece has been the EU country least likely to indulge in market-friendly internal policies, reduction of public expenditure and taxes, and privatization and liberalization of its markets. Portugal seems to have done almost the opposite: it fol- lowed its neighbor Spain in unilaterally liberalizing trade in 1960, shifted to a regime of high public spending and taxation coupled with heavy state intervention in labor and product markets right after the 1974 revolution; and resumed liberal- ization, privatization, and labor market reform in the late 1980s. Its per capita GDP, relative to the EU 15's, followed a similar sequence from 45 to 60 percent between 1960 and 1974, unchanged between 1974 and 1988, and from 60 to 78 percent between 1988 and 2000 The Mezzogiorno's itinerary is slightly more complex, as it has been the object of Italian and EU transfers at the same time In any case, dur- ing the 1960s the flow of external funds to the Mezzogiomo was relatively low: the l950s and 1960s were the decades of the labor migration to northern Italy and Northern Europe. By the time of the oil crisis in 1974, the Mezzogiorno per capita GDP was about 63 percent of the EU average. Since then, the Mezzogiomo has become one of the privileged targets of EU Structural Funds and the Italian gov- ernment has stepped up its subsidy and transfer policies. Migration flows ended and official unemployment started rising By the year 2000, the Mezzogiomo per capita income was around 68 percent of the EU average Historical experience and economic analysis have also taught us something Important about the short-term effects of trade integration policies. Technologically and economically backward countries that enter into a free trade arrangement with more advanced ones undergo a fairly rapid and dramatic process of structural change that leads to the destruction of employment in the agricultural sector and other back- ward sectors. The adjustment is socially costly, and the job destruction it generates is the key threat to long-terrn prospenty because the unemployment wave induced by the transition tends to be persistent over time. Elimination of structural unemploy- ment requires strong and clearly targeted mobility and training policies. Such poli- cies have seldom been implemented because, in the face of growing unemployment, political pressure to adopt grossly inefficient transfer policies usually succeeds. The likelihood of such erroneous policy choices is enhanced by the availability of exter- nal aid funds. Economic theory aside, historical evidence shows that subsidies and transfer policies make regional imbalances more persistent. To summarize the points, (a) while structural transfers carried out at the Euro- pean level leave growth rates roughly unaffected, national policies do make a huge difference, (b) national policies that reduce distorting taxes and unproductive pub- lic spending, liberalize labor markets, foster job search and retraining, attract FDI, and minimize income support transfers seem to lead to a sustained period of above- average growth. 76 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH To lend support to our claims, we look at the recent East German experience as another emblematic case of badly conceived and highly perverse transfer policies. The Big Brother Effect Unlike CEECs, the German Democratic Republic enjoyed the protection of its big Western brother during its transition From an economic point of view, this has meant that (a) a number of domestic income support policies, financed with taxation from the West, have shielded the residents of the former German Democratic Republic from the most unpleasant costs of the transition; (b) firns operating or installing themselves in the former German Democratic Republic received abundant public subsidies; and (c) the East Gernan Iinder have been classified as Objective I regions, receiving Structural Funds from the EU. The total amount of transfers involved is enor- mous. For example, during 1991-97 the German government alone transferred to the East German kInder a total of E571 billion corresponding to about 50 percent of their GDP over the period. During that same period, total lending and transfers from the World Bank to CEECs and Central Asian transition countries equaled approximately 3 percent of this quantity (see World Bank 2001c). Beside these windfalls, firms of the East German lknder had immediate access to the EU1 5 markets and did not face the restnctions, tariffs, and other admirustrative obstacles that apply to products and ser- vices coming from outside the EU. Given that the initial conditions in the German Democratic Republic were substantially better than those of other CEECs when tran- sition started and given the plenitude of public support, how did the East German kIn- der economy perform relative to the less fortunate CEEC1 0? It is well known that the growth perfornance of the former German Democratic Republic has been quite mediocre. After a spurt of rapid growth, right after unifica- tion, growth for the East German kinder slowed down, and during the past six years they have grown more slowly than the rest of the EU. From 1995 to 2000, growth in GDP per capita averaged less than 2.0 percent, much lower than growth in every CEEC but Bulgaria and Romania. Labor productivity, which started at about one- third of the West's in 1991, grew to 65 percent in 1995 but has stagnated ever since. Over the 1995-2000 period, growth in the former German Democratic Republic's labor productivity was only 1.5 percent a year, the lowest in Europe, with manufac- tunng and the business service sectors performing particularly poorly. The observed slowdown has lasted long enough to rule out any "cyclical" inter- pretation of this phenomenon: in spite of the enormous transfers, the economy of the former German Democratic Republic is not converging to either the EU or the German averages. While it may be a bit too early to classify the East Idnder as the Mezzogiorno of Germany, it is certainly another example of persistent backwardness, lower produc- tivity, low labor force participation rates, higher unemployment, and high subsidies. The analogies with the Mezzogiorno go further. In particular, the "backwardness" or "poverty" of the East German lander is, like that of southern Italy, quite relative. After all, GDP per capita in the former Gernan Democratic Republic has reached 76 percent REGIONAL POLICIES AND EU ENLARGEMENT 77 of the EU average, approximately the level of Portugal and several points above the level of Greece However, to the extent that half of this amount is financed by the Fed- eral Republic of Germany, it is clear that in comparing the GDR's lander with CEECs, a much lower level, say 45 or 50 percent of the EU average, should be taken as a true value. This places the East German lander just below the Czech Republic, Hungary, and Slovenia in terms of both labor productivity and per capita GDP. In other words, once the large income transfers from the West are factored out, the East German lander are, in terms of domestic productive capacity and produced income, a few percentage points below where they were 12 years ago. Since the best performing CEECs appear tc, have done better and are projected to do significantly better in the near future, one should seriously ask why the extraordinary amount of resources poured mto the econ- omy of the former German Democratic Republic have provided income support but failed to deliver sustained growth incentives. There may be several reasons why this has happened. A number of hypotheses have been analyzed in the existing literature (see, for example, Canova and Ravn 2000; Sinn 2000; Franz and Steiner 2000; and Ragnitz 2001 for some of the most insightful examples). A highly counterfactual but nevertheless popular point of view, paralleling earlier literature on the failures of state aids to the Mezzogiorno or the poor growth performances of Objective I regions, has reached the Panglossian conclusion that while the amount of aid transferred was large, it was evidently not enough and/or it was not well spent. This refrain, heard also at the Barcelona conference, reminds one of the old story according to which socialism had not been implemented correctly in the U.S.S.R, but it could and would produce wonders in, say, Albania. In our opinion, the most coherent explanation has to do with the perverse effects that income support policies generate on the incentives to work, produce, and invest for citizens of eastern Germany. Since 1990 total employment in eastern Germany has fallen by 30 percent, from 9 million to 6 million, roughly the fall observed in Bul- garia, the largest fall in the CEECs. As in most transition countries, the manufactur- ing sector has expenenced the largest decline in employment, and job creation has occurred only in the construction sector. The expansion of the latter has been partic- ularly abnormal in eastern Germany because funds from western Germany for recon- struction have led to an oversized construction sector with poor or negative efficiency gains in the sector itself, which had low TFP and a low technological content at the beginning. Obsolete and unproductive firms have been scrapped, and despite an ini- tial boom (1990-91) in the creation of new firms, the entrepreneurial process stopped right after the first enthusiasm for reunification waned. The registration of new enter- prnises in the east, which was almost twice as large as that in the west in 1991, fell to only one-tenth of the west's in 2000. Interestingly, over the period 1995-2000 Ger- man FDI flow to CEECs was, larger than to eastern.Germany (Eurostat 2002d). This may not be altogether surprising In 1990 gross nominal wages in eastern Germany were 46.7 percent of western Germany's, while in 1999 they had reached 73.9 per- cent of western Germany's. In companson, average labor productivity was only 31 per- 78 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH cent of western Germany's level in 1990 (implymg unit labor costs 50 percent higher in East Germany) and was 59.4 percent in 1999 (implying unit labor costs that are still 24 percent higher m eastern than in western Germany). As we have seen, unit labor costs in the CEEC 10 as a whole are substantially lower than in the EU, where in turn they are lower than the unit labor costs of western Germany. Like employment, labor force participation has dramatically declined as a con- sequence of generous early retirement programs and the large number of women exit- ing the labor market. Overall, labor force participation has shrunk by 7 to 8 percent in the past 10 years, 50 percent more than in the average CEEC. Income support to the unemployed has become substantial: one-fifth of eastern Germany's population received income support in the form of benefits or state assistance in 1999. The level of support is high in absolute (about 1,000 per capita in the first case and about 750 in the second case) and relative terms. So-called active labor market retraming programs provide extremely generous support for participants, in some cases higher than the monthly wage earned by a low-skilled worker. Not surprisingly, they have a remarkably poor record at increasing participants' employment levels. Overall, more than 60 percent of total transfers to eastern Germany are for social protection. In this situation, we should expect to see stagnating growth perfornance, little down- ward pressure on wages, rent-seeking activities by nonparticipants and the unem- ployed, and a strong incentive to maintam the status quo-a picture that duplicates to a large extent the experience of southern Italy and of a number of provmces of south- ern Spain. Subsidized income support has become a modus vivendi and has created incentives for its perpetuation. Ragnitz (2001) has also suggested that the abundance of assistance programs has produced a very inefficient allocation of new capital. Funds have been used to maintain an inefficient and unproductive manufacturing sector, and new enterprises have been established mostly in sectors with low levels of competition, low levels of research and development, and low technological content. In comparison, new investments in CEECs have been typically directed to the most productive sectors and, at least in the most successful countries (Hungary and Slovenia, and Poland to a lesser extent), direct income support has been low and indirect income support via a high minimum wage negligible enough to prevent the creation of counterproductive incentives and gross malfunctioning of labor markets. The case of Poland is, from this point of view, almost paradigmatic since the economy is large and heterogeneous and its per capita income levels were, and still are, lower than those in eastern Germany. Poland initially followed a well-established EU policy of pro- moting interregional labor immobility, subsidizing or giving workers mcentives not to leave areas where unemployment rates were higher, ignoring their poor housing situa- tion, and mcreasing minimum wages to the point of making them binding, especially in the less developed areas. Facing nsing unemployment, Polish authorities have started reversing these tendencies in recent years. To the businesspeople of western Germany, Poland may seem to offer much better growth prospects than eastern Germany. RIEGIONAL POLICIES AND EU ENLARGEMENT 79 Assessment of Regional Policies in the Enlargement Process Wie begin this section with a very short summary of what regional policies are, how much funding EU regional policies currently provide, and how those funds are dis- tributed The EU budget was, in 2000, equal to about 1.05 percent of aggregate GDP, of which 46 percent was taken by the CAP and 36 5 percent by the Structural and Cohesion Funds. Greece, Ireland, Portugal, and Spain are the big net receivers, while a.] the other countries are net contributors. Relative to national GDP, the Netherlands is the largest contributor. The Irish position as a large net receiver of funds (about 4 percent of GDP) still reflects the recent past, when Ilnsh per capita GDP was below 90 percent of the EU average From the point of view of gross flows, Germany and Italy (with 1999 30 billion each, over the 2000-06 cycle) and France and the United Kingdom (with 15 and 16 billion, respectively) should also be added to the list of great beneficiaries of regional funds. For the sake of comparison, over the same bud- get cycle Spain is set to receive 56 billion, Greece 25 billion, Portugal 23 billion, and lheland 4 billion (all figures in 1999 euro). The Structural Funds (European Agricultural Guidance and Guarantee Fund []EAGGF], European Regional Development Fund [ERDF], European Social Fund [ESF], and Financial Instrument for Fisheries Guidance [FIFG]) are supposed to finance projects pursuing at least one among six (three as of 2001) policy objectives. Each of them corresponds to a different subset of regions of the EU, even if the Com- mission makes a distinction between "regional objectives," which concentrate about 85 percent of the budget, and "nonregional objectives." Objective regions are desig- nated at the NUTS 2 level. The Structural Funds cofinance multiannual programs in the member states. In contrast with the expenditure of the CAP, where payments from the fund are largely determined by a formula, there is a substantial discretionary element in the spending of the Structural Funds. Taking full advantage of the Structural Funds is a labor- intensive exercise for member states, both because of the technocratic standards proj- ects are expected to match and because of the extensive coordination required to satisfy the partnership principle. The Cohesion Fund, introduced by the December 1992 European Council (Edinburgh), is the second pillar ofthe current EU regional policy. The Cohesion Fund aims specifically at improving European transport networks and overall environmen- tal conditions. Interventions have not a regional but a national basis, and eligibility requires a national GDP per capita below 90 percent of the EU average. This require- nient has de facto limited the Cohesion Fund in the four poorer countries (Greece, Ire- land, Portugal, and Spain). Ireland, which has been effectively well above that thireshold since the late 1990s, is still receiving funds during the 2000-06 budget cycle A European Commission communication of March 1998 titled "Reform of the Structural Funds" aims at reducing the number of regions covered by the funds. The 80 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH proposals, which were not incorporated into the deliberation of the European Council of Berlin and which (as far as we know) have not yet been implemented, envision a reduction in the number of objectives from six to three and a stricter enforcement of the 75 percent threshold for the NUTS 2 regions that are covered under Objective I. Should the current criteria for eligibility for Structural Funds be maintained after accession of the CEEC10, current recipients would see their transfers evaporate almost entirely For example, in the financing cycle that is supposed to start in 2007, only two Spanish regions (Andalucia and Extremadura) are likely to qualify for Objective I funds. In Italy, only Calabria may remain. At the same time, Spain's and Portugal's income per capita would most likely move above the 90 percent threshold for admission to Cohesion Fund transfers, so these transfers would also disappear Other likely losers of funds are the lander of the former German Democratic Repub- lic. At the same time, nearly every administrative division of the new countries (with an exception made for the cities of Prague, Bratislava, Ljubljana, and possibly Budapest) would qualify for Structural Funds support. Preaccession aid to the CEECs, a form of Structural Funds aid, has been operat- ing since the 1994-99 budget cycle. The Commission's suggested policy for the postaccession system, so far, implies an effective cap at 4 percent of GDP for the receipts of any country and an overall target of spending on Structural Funds equal to 0.46 percent of total EU GDP. Such limits will be effective until the end of the cur- rent budget cycle (2006), after which they will be open to modification. A crucial pol- icy issue that is actively debated concems the new limits (if any) for such spending. As far as the Cohesion Fund is concemed, the issue is blurrier. Since the Cohe- sion Fund was created to facilitate the respect of the so-called stability pact by the poorest countries of the EU 15 in the wake of the euro adoption, it is not obvious that it should be maintained now that (a) the euro has been successfully adopted, and (b) only Greece is still below the 75 percent threshold and it may not remain there if CEECIO countries are admitted in 2004 and 2008. Taking the planned admission as a given, two options seem possible: keep the Cohesion Fund, diverting its resources almost entirely to the newcomers, or abolish it. Our main suggestion is that European regional policies be terminated after the current budget cycle ends in 2006. This is, we believe, the best choice of policy because current regional policies are ineffective, based on incorrect or at least unsub- stantiated economic theory, badly designed, poorly carried out, and a source of wrong incentives and, in some cases, corruption. The lack of direct enforcement mecha- nisms that evaluate the efficiency and effectiveness of fund allocation creates an environment where corruption is bred. The recent episode of misallocation of ESF and CAP funds (highly publicized in the Spanish press) indicates that these are not merely "theoretical" problems. The presence of Structural Funds may have some indirect economic benefits for CEECs: solidify the democratic rule in the East, help implement institutional changes, create intemational partnerships, and reduce social tensions generated dur- REGIONAL POLICIES AND EU ENLARGEMENT 81 ing the transition. The funds' existence, however, generates two sets of senous prob- lems for the growth prospects of CEECs. On the one hand, they create a clear political obstacle to the enlargement process: all current beneficiaries of the regional transfers are trying to avoid losing them, while the current contributors are trying to avoid paying for a much larger bill Under current policies both events are very likely, especially when one adds the Structural Funds and the CAP together Barring the courageous choice of shutting down re- gional policies altogether, the two opposite camps are likely to find a murky com- promise, with exemptions and suspension provisions, compensatory transfers, and so oni. If the accession of Finland and Sweden into the EU has led to the creation of Objective 6 to compensate the two countnes for what they contributed to Brussels, the accession of the CEECs is likely to create a whole set of new objectives, special clauses, exemptions, and "temporary" compensatory transfers. On the other hand, making the vast resources of the EU regional policy avail- able to the CEECs may provide them with a substantial incentive to insist on the wrong policies. Practically, they may postpone the elimination of state subsidies to obsolete and inefficient enterprises. They may also create new or reinforce existing "income maintenance programs," thereby reducing labor mobility and providing incorrect incentives to entrepreneurial capital. The availability of such funds and their tendency to focus on large projects of public utility may also provide a further reason to delay adjustments in the utilities sector, with negative implications for overall efficiency. As we argue in the next section, the sequence of policy choices is not only crucial to foster growth but also a key precondition to a much-needed reduction in labor income taxation. The accumulated experience of eastern Ger- many, Greece, the Mezzogiomo, and southern Spain suggests that the risk of falling into the "big brother trap" when large external subsidies are available is very high. Overall, we believe that the costs greatly surpass the benefits and that the radical suggestion of eliminating them after 2006 could make a large percentage of the future EU better off Since our preferred option will not find many supporters in Brussels, it is worth loDking at which reforms would keep Structural Funds alive but limit their damag- ing effects and enhance their positive impact on economic growth in the CEEC 10. The theoretical principles underlying the EU regional and structural policies are, prima facie, commendable and hard to dispute. The Commission calls for (a) con- centration of funding where it is most needed, on the basis of explicit and certified (b) planning of such intervention in (c) cooperation with local and national authori- ties whose funding the EU transfers are supposed to (d) complement (with cofund- ing going from 50 to 80 percent of the value of the project). As is often the case, the reality is quite different. We have already emphasized the lack of both common and economic sense behind the choice of NUTS 2 and NUTS 3 regions as the territonal levels at which economic convergence should be measured. We will not harp further on this point, but we list it as the natural first step in a long-overdue reform of Euro- 82 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH pean regional policies: elect territorial umts that are both homogeneous and large enough to make convergence in per capita GDP a reasonable target and coordination at the European level justifiable. Common sense suggests choosing areas with a pop- ulation of about 10 million people. In the light of the CEEC accession, this would imply that convergence should be measured at a country level, with an exception made for Poland and Romania. The choice of appropriate territorial units for measuring convergence leads to the issue of the level at which resources are funneled. Currently, various subnational administrative levels are involved, sometimes particularly small ones. Theoretically, the choice of subnational unit is meant to stimulate decentralization. However, the restrictions imposed make the approach resemble a degenerate form of fiscal feder- alism. First, the administrative entities mvolved are very unequal, and since the Com- mission imposes homogeneous technical requirements on planning, financing, and implementing the projects, this places a huge burden on small regions or adminis- trative units. All but a handful of very large local administrations use the services of consulting companies located in Brussels to handle Structural Funds projects. Alter- natively, they let their central governments elaborate, present, bargain for, and man- age those projects on their behalf-not much decentralization or federalism. Italy, where Structural Funds for the Mezzogiorno are de facto handled, coordinated, and almost dished out by a dedicated Direzione Generale at the Italian Treasury, is the most egregious example. Second, while EU funding is not supposed to replace local spending, it obviously does because of the aggregate budget constraint at the level at which resources are funneled. Furthermore, as central governments are active partners in the funding process and are allocating national resources to the same regional entities to which European funds go, it is at the level of central government budgeting that substitu- tion takes place. With the exception of Germany, the administrative units involved have little or no autonomous fiscal power: their resources flow from central govern- ments that, obviously, count Structural Funds provisions as part of total financing. Finally, the desired territorial concentration of funding is, to say the least, long gone: all EU15 countnes receive some regional subsidy Counting in a map of Europe, the number of NUTS 2 regions receiving some transfer under some Objective 1 reaches more than 90 percent of the total. In fact, as we learn from Chapter 11, at most 70 per- cent of the total amount of funding goes to areas with an income per capita lower than the EU average. One way or another, almost all regions of Europe need to converge to the average European income. These arguments reinforce the need for radical reform of both criteria and meth- ods for funding. Concentration of funding where most needed should be reestablished as a relevant cntenon by setting a much lower threshold for the definition of a dis- advantaged region. Cooperation and complementarity should also be ensured by funding only regions that are true federal units, with autonomous taxing/financing power, and that are able to handle their budgeting process in cooperation with the REGIONAL POLICIES AND EU ENLARGEMENT 83 European Commission and independent of their central governments. Where such entities are not present, one should choose between financing central governments directly (as should be the case with most CEECs) or not providing funds at all. If realpolitik implies that some side payments going from the nchest to the poorest members of the EU have to be maintained as a polite form of "bribing" and political "consensus building," then such side payments should be handled at the country level with the exception of those subnational units that have achieved some form of true federal autonomy. This is similar to the views expressed by de la Fuente in this vol- ume and, for different but complementary reasons, by Guido Tabellini, who argues, "More likely, the main goal of structural and cohesion funds was redistributive: not tc, increase economic efficiency, but to redistribute the benefits of integration among countnes, providing side payments so as to facilitate compromise in bargaining sit- uations The question then is whether the same goal could have been achieved in less distorting ways Participants at the bargaining table are countries, not regions. Side payments are thus needed among countries, not among regions or groups of individ- uals" (Tabellini 2002, p. 19). We are convinced that the recognition that Structural and Cohesion Funds are just transfer payments used to facilitate political bargaining and coalition building is not forthcoming. Hence, the fiction of the "convergence goal" and of "growth and ef ficiency enhancing" objectives is likely to be maintained, scientific evidence to the contrary notwithstanding. In this case, two reforms should still be advocated. One is a drastic lowering of the maximum income for admission to funding. A level equal to 50 percent of the EU average would, in our view, be a good choice, allowing funds tc, be concentrated where they are most needed. Such a cutoff would not only exclude all current EU members from funding but, among the CEECIO, also exclude the Czech Republic, Slovenia, and possibly Hungary. Of the first entrants, only Poland (rninus the metropolitan area of Warsaw) would clearly be a potential beneficiary of Structural Funds It seems most likely that (by 2008) Bulgaria, Latvia, Lithuania, and Romania would still be below 50 percent of EU average income per capita and there- fore would qualify for this target. Second, we recommend a drastic reduction of the number of objectives to be pursued (as proposed by the Commission in 1998 and being implemented currently). In our view, Objective 1, properly rephrased to focus onl structural deficiencies (especially large public goods, transportation and commu- nication infrastructures, and environmental protection), is the only one that should be retained on a permanent basis. In the light of the CEECIO accession, it appears that Objective 2 (recovery from industrial restructuring) and Objective 5 (agriculture structural transformation) should also be maintained during the first budgeting cycle following admission (2007-13) because of the relevance of both industrial and agri- cultural restructuring in these countries. From a practical point of view, the tough part consists of effectively financing large-scale projects that favor recovery from industrial crisis and agricultural restruc- tunng. A couple of criteria are worth suggesting. First, we would, contrary to much 84 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH common wisdom, deemphasize the support for small and medium enterprises. Poli- cies of support to small and medium firms should be the task of national governments via properly designing fiscal and labor market legislation and granting small firms easier access to financial markets. It is our view that such national policies are far more effective (when properly implemented) and relevant than some general subsi- dies coming from Brussels that are linked to complicated business plans, the elabo- ration of which is often too demanding for truly small companies. Anecdotal evidence and common sense coincide in suggesting that supporting small and medium enter- prises via Structural Funds equates to supporting those firms that look small, have good political connections, and have a comparative advantage in rent-seeking (rather than VA) activities. Second, we recommend insisting quite strictly on the "public good" nature of the projects to be financed. We recommend concentrating funds to provide productive infrastructures: transport, communication, power and water dis- tribution, and educational infrastructures. This we advocate for three simple reasons. First, both theory and empirical evidence suggest that if there is anything like a poverty trap, this is determined by the shortage of the kind of productive public goods we have just listed. Second, besides a favorable fiscal and labor market environment, these are the kinds of public goods that generate the "absolute advantage" (in the sense of Jones 2000) that is crucial to attracting mobile factors of production, FDI in particular. Third, no matter how corrupt the allocational system for the production of such large public goods may be, it allows more control from the EU and engenders less damaging collusion between the private sector and the political system than the subsidization of a large number of small private enterprises. There are also a number of actions we would discourage. One is lifting current spending limits on Structural Funds (as suggested in Vaitilingam 2002). The subtle plan behind such advice seems to be that of getting the old beneficiaries to agree to the admission of the new members without having to give up the sacred cow of the "structural policies." Such commentators have often insisted on the existence of "enormous income gaps" between current and future members. This is incorrect: the current differences between the EU 15 and CEEC 10 are similar to, or even smaller than, those that the EU managed to overcome quite successfully during previous enlargements. Second, various commentators have insisted on the opportunity and necessity for enlargement countries to forge ahead with deep labor market reforms, such as relaxing minimum wage requirements and lowering labor income taxes. These rec- ommendations are most welcome. Nevertheless, it is hard to see why one would ad- vocate labor market reforms in these countries when labor markets in many EU countries are less flexible and when labor market policies, at both the national and supranational levels, in the EU do not meet the standards. The labor market in, say, Hungary is more open and liberalized, although maybe less efficient, than that of sev- eral of the members of the EUI 5. Similarly, the structure and the dynamic of the labor market structure in Slovenia are not dissimilar from those of its largest EU neighbor, REGIONAL POLICIES AND EU ENLARGEMENT 85 Italy, and growth rates of labor productivity and employment in Slovenia actually look significantly better in recent years. Third, mixing good intentions with bad economics, other commentators have recommended the creation of minimum guaranteed income schemes similar to those in place in most current EU member countries as a condition for accession, and even EU coordination of the level of these minimum guaranteed income schemes The long-term plan seems to be that of building up a pan-European social safety net as one of the pillar institutions of the EU. We find such proposals dangerous, especially for the future growth perspectives of the CEEC IO and other poorer areas. The ratio- nale is simple the big brother effect kills labor mobility and entrepreneurial efforts and price flexibility, all of which are key ingredients of successful growth. Further- more, minimum guaranteed income schemes need financing, and this financing can come either from additional transfers from richer countries (increasing their fiscal burden and reinforcing the subsidization culture on the receiving side) or from labor income taxation at home. Surely, CEECs do not need an increase m their labor income taxes Nevertheless, given that we do not expect income support to poor countries to end with accession, we would like to suggest two general principles to minimize the induced distortions. First, income support policies should target job seekers (as is done right now in the Netherlands). Second, Structural Funds directed to the creation of employment opportunities-as opposed to those directed to the creation of high productivity levels via provision of public goods-should aim at enlarging labor force participation, which plays a crucial role (as the Irish experience shows) in fos- tering income convergence. This leads us to another theme, which is also stressed in the next section labor income taxation. Theory tells us that, especially when unemployment is high and sec- toral reallocation of labor an important priority, the optimal tax rate on labor should be low When total taxation on labor is too high, two types of distortions are created: potential job seekers are discouraged, and the productive side of the population is heavily penalized to subsidize the income of those who are unproductive. These dis- tortions may perpetuate a vicious cycle, which is particularly vicious in countnes (like the CEEC10) facing major restructuring, destruction of a large number of firms, eiitrance of new firms from abroad, and a rapidly changing distnbution of skills in the labor force. If a labor market reform is needed in the CEEC 10, it consists of reduc- ing unemployment subsidies and income maintenance programs (together with enter- prise subsidies) in order to free labor income of the gigantic fiscal burden it cames. Labor market reforms should probably be coupled with policies that favor the eificient reallocation of labor across not only sectors but also regions. "Labor migra- tion" seems to be a well-guarded taboo among both current EU members and CEECs right-wing politicians use the potential size of migration from the East to increase EU citizens' fears of losingjobs and cuirrent benefits. On the other side, CEECs worry that labor mobility may exacerbate the drain of high-skilled workers, thereby rendering national disparities even larger. Overall, we believe that these fears have been over- 86 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH stated and that regional misallocation of labor and persistent poverty is greatly ampli- fied by the equity-without-efficiency approach upon which many EU policies have been forged. This regional misallocation is going to be even larger if the existing poli- cies are mechanically extended to CEECs effective removal of Objectives 3 and 4 of the Structural Funds should be sought By creating incentives against labor mobility across regions, these objectives perpetuate income inequalities across regions. Furthermore, like Vaitilingam (2001), we believe that the fears of mass migration from the CEEC10 are overstated. Most of those who wanted to and were able to migrate from the CEEC10 to the EU15 countries have already done so since 1990. This may sound surprising, given the current fears of migration from the CEECs to the EU and the relatively small number of workers from those countries who have made it inside the EU so far. Available estimates place the stock of CEEC immigrants residing in the EU at around 800,000, 60 percent of whom are concentrated in Aus- tria and Germany. After an initial surge, the flows from the CEECs to the EU have subsided, with minor movements from one country or the other, all clearly associated with cyclical fluctuations in the home country. There is no evidence that current restrictions on labor migration from the CEECs to the EU are holding back a gigantic mass of potential migrants. Just to cite an example, while before unification a net 300,000 people from the German Democratic Republic migrated to West Germany in 1990, in the period 1992-2000 the net flow dropped to about 50,000 a year Because of the natural linguistic and cultural links we think of this as an extreme case not to be used as a reference point to forecast what will happen in the case of accession but only as a very high upper bound. Furthermore, although nonnegligible in size, the percent- age of people potentially migrating toward the EU represents only 0.2 to 0.5 percent of the total EU labor force, an amount that a large labor market like the EU should be able to absorb with little difficulty. This number is substantially smaller, for example, than the annual flow of immigrants who find employment in the United States. His- torical experience shows that whcn the migrant masses want to come in, they do. Medium-term economic perspectives in the home country relative to those in the host country count a lot more than absolute levels in determining the intensity of migra- tion. To the extent that growth rates In the CEEC IO equal or exceed those of the EU, the migration flow will be kept within very reasonable limits. Finally, to go back to the comparison with early enlargements, the flow of immigrants from Greece, Portu- gal, and Spain after their admission to the EU was quantitatively irrelevant- we see no reason to fear that the flow from Poland or Slovenia after 2004 will be much larger. ConcIusiolrns Overall, experience from earlier EU enlargements and current economic conditions within the CEECI0 suggest that placing very high expectations on enlargement per se is not useful. It will help politicians, but it will not lead to economic convergence. Furthermore, regional transfers taking place under the structural and cohesion poli- RE GIONAL POLICIES AND EU ENLARGEMENT 87 cies are unlikely to become the growth engines of the CEECIO They may increase income in the receiving countries by an amount equal to the one transferred, but there is no evidence that they will have an impact on long-term growth rates. To achieve long-term growth at rates higher than average, an appropnate mix of European and national policies is needed. This includes further fostering of trade integration within the EU, restructuring of public spending, creation of supply-side incentives by proper reforms of fiscal and social insurance policies, and free movement of capital and labor, together with a competitive level of labor income taxation. Based on historical experience, two assertions appear to be particularly relevant. First, public programs for long-tenn income support, corporate subsidies, and other fomls of income trans- fer have a negative effect on economic growth. We believe they should be terminated as soon as possible. Second, labor and capital mobility are good for growth and eco- nomic convergence. In particular, the adoption or continuation of various transfer aiid/or regulation policies aim'ed at eliminating labor migration from CEECs is wrong and damaging. The fear of migration has been magnified by skillful politicians. nitigration after past enlargements has been small. There is no reason to expect it to be large in this case. A more detailed list of findings and policy implications follows First, although theoretically possible under fairly special circumstances, there is no reason to believe that trade integration per se would lead economies to diverge On the contrary, all past experiences of trade integration, especially those that have taken place in Europe since the 1950s, have led to sizable improvements in the fac- tor endowments of the poorest partner and in the efficiency with which such factors were allocated in production These improvements have reduced income inequalities and enhanced production possibilities across participating countries. Hence, as a principle, further trade integration should be pursued among European countries and between the EU 15 and the CEEC 10 in particular. Second, pure trade integration generates a leveling effect more than a growth effect on participating countries. In particular, there is no hard evidence supporting the idea that trade integration alone may increase the long-term growth rates of par- ticipating countries. This implies that, when absolute convergence is the objective, other national policies than liberalization of international trade play an important role. The experience of various European countries shows that reduction in fiscal pressure, accompanied by parallel reduction in public spending, is among such poli- cies. Capital and labor mobility, together with a competitive level of labor income taxation, also play a role in fostering real convergence Third, while a fairly stable macroeconomic environment is certainly necessary for growth, there is no evidence that, by themselves, a low inflation rate, low public deficit, and low public debt will foster economic convergence. In fact, the experience of many regions within the EU proves that, when facing the same monetary, fiscal, and exchange rate policy, poor regions need not grow faster than rich ones. Fourth, both evidence and economic theory suggest that, given a stable macro- economic environment, the presence or lack of supply-side incentives plays a crucial 88 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH role in determining long-term regional performance. The half-century experience with southern Italy and the more than two decades of experience with southern Spain show that the availability of large and permanent income support transfer programs has a negative impact on economic efficiency and long-term growth. The relatively more recent experience of the East German kinder leads to the same conclusion: pub- lic programs for long-term income support, corporate subsidies, and other forms of income transfer have a negative effect on economic growth. They hamper instead of foster economic convergence. Fifth, the experiences of Ireland, Portugal, and Veneto in the EUI5 and of Poland and Estonia in the CEECIO show that sustained above-average economic growth is the consequence of an attractive environment for FDI and new small firm creation, risk-taking entrepreneurial behavior, and exploitation of local comparative advantages via enhanced labor and capital mobility. Low marginal taxes, efficient transportation and communication infrastructures, good financial facilities, and a rel- atively flexible supply of high-level human capital appear to be the key ingredients of a growth-friendly environment. Sixth, labor and capital mobility are good for growth and economic conver- gence. Free capital movement across national borders seems to have become an obvi- ous and accepted policy stance in the EU. Things are different with respect to labor movement. A number of researchers are recommending the adoption or continuation of various transfer and/or regulation policies aimed at reducing or even eliminating labor migration from European regions. We consider this prescription wrong and damaging. While the social and political costs of mass migration are certainly large and cannot be underestimated, a certain amount of free labor migration is necessary to a well-functioning labor market. Furthermore, labor migration from one EU coun- try or region to another has often been temporary and has been associated with episodes of rapid economic growth and convergence. Evidence shows that move- ments of labor within Europe and the United States have been among the major forces behind economic convergence. Labor migration is one of the most important channels through which precious productive skills are acquired in advanced regions and brought into poorer regions to be applied. Finally, a vanety of arguments and considerations suggest that the size of the postadmission migration flows from the CEECs to the EU countnes will be much smaller than envisaged by catastrophe-predicting politicians and will have very little impact on the labor markets of the EU 15. In essence, we find no reason to adopt direct or indirect policies to restram free movement of labor from the CEECs to the EU. The best policy would allow free movement of labor without creating incentives or disincentives in either direction. Seventh, experience from earlier EU enlargements and current economic condi- tions within the CEEC 10 suggest that placing very high expectations on the economic consequences of the enlargement would be incorrect. Hopes that EU regional and structural policies will be the key to rapid economic growth and convergence are not REGIONAL POLICIES AND EU ENLARGEMENT 89 likely to be fulfilled. Regional transfers taking place under the structural and cohe- sion policies are just that, transfers. To achieve long-term growth at rates higher than average, an appropnate mix of the policies described in the second, third, fifth, and sixth points above seems to be needed. Eighth, in the light of their very secondary effect on long-term growth and of the particularly acute political tensions their availability and allocation create among cur- rent members, one should reconsider the very existence of regional Structural Funds within the enlarged EU. This also applies, with stronger force, to the funding of the CAP. Theory and evidence show that Structural Funds are pure income transfers with few positive long-term effects. The availability of such transfers generates two very negative effects. First, it leads to rent-seeking behavior on the part of poorer regions that want such funds. It also creates rent-seekmg coalitions of the "half-poor" against the "even poorer" or the "very nch," giving rise to spurious coalitions whose only objective is to increase the amount of transfers accruing to one particular region or country. Both activities cloud the political discourse and, as in the case of the current enlargement, create artificial and purely redistnbutive obstacles to otherwise valuable and efficiency-enhancing political decisions. Second, it leads to inefficient allocation cf resources within regions that are the beneficiaries of such transfers. All the micro- economic and anecdotal evidence available shows that a large share of Structural Fund resources are wasted in the lobbymg and advocacy of projects. Many projects, either public or pnvate, that would not have been financed under normal competitive condi- tions are financed by Structural Funds simply because the latter are tied to a certain area. This leads to a suboptimal allocation of regional labor, capital, and entrepreneur- ial resources and to a self-perpetuating system of expectations in which below-average income levels are almost "sought" by the regional administrations as a conduit for addi- tional structural funding. In the long run, both of these effects lead to the misallocation of resources, corruption, underground activities, and lack of sustained growth that char- acterize the Mezzogiomo. This is bad for growth and most definitely does not help eco- iiomic convergence. Structural Funds should be phased out over the next EU budget cycle (2006-12) The Cohesion Fund, whose objective has been achieved with the suc- cessful establishment of the euro, should be terminated at the end of the current spend- ing cycle (2006). Short of this, the recommendations listed in the second part of the Assessment of Regional Policies in the Enlargement Process should be followed. Appendix A-Data Sources The majority of the data used in this study come from International Monetary Fund's International Finance Statistics. Data for investments, savings, FDI, depreciation, government deficits, employment, unemployment, population, the consumer price index, nominal wages, and nominal GDP are all taken from that database. Discrep- ancies and incoherencies are checked agamst the OECD Main Economic Indicator data bank and integrated when needed. 90 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Data for GDP in PPS relative to the EU come from Funck and Pizzati (2002, p. 31). Regional data on GDP for CEECs is obtained from Eurostat (2000b, 200 Id, 2002e). Data for labor productivity are reconstructed by the authors using data on a labor productivity index provided in Martin, Velazquez, and Funck (2001) and labor productivity data contained in a number of Eurostat publications (2001c, 2002c). Important information about FDI in candidate countries is obtained from Euro- stat (2000a, 2002d). Data for the EU are also obtained from a number of Eurostat publications. In par- ticular, data on the investment rate come from Eurostat (2001 a), those on employ- ment from Eurostat (2002a), those on unemployment from Eurostat (2002b), and those on labor force participation from Eurostat (200 le). GVA and employment by branch are constructed by the authors using the infor- mation contained in Eurostat (2001c, 2002c). Notes 1. Spain would make a positive exception in this case Since the beginning of the transition, and thanks mostly to the presence of four or five large private banks, the Spanish bankmg sys- tem has been much more advanced and its banks much larger than those of Spain's Mediter- ranean counterparts. 2. Romania is a clear outlier in this sector, see the chapter by Giurescu in Funck and Pizzati (2002) detailing the circumstances surrounding this phenomenon. 3. What we call "steady state" here and in what follows is, in reality, a growth path. It can be treated as a steady state, that is, as a fixed point of a stationary dynamic system, because we are normalizing everything by the (unknown) average long-term growth rate of the EU. It is relative to this unknown growth rate, whatever it may turn out to be, that the position we com- pute is a steady state. 4 These are all 14 Greek regions, 4 of the 5 Portuguese regions (Lisbon is excluded), and all 4 southern Spanish regions (Extremadura, Andalucia, Murcia, and Cananas) 5 We recall here that the asymptotlc growth rate depends on the production function param- eters, the discount factor, and the intertemporal elasticity of substitution in consumption. Hence, we are assuming that at least one of these parameters may vary across countries. 6 We refrain here from getting into the methodological, but unsubstantiated, other point advanced by de la Fuente, according to which, if one conditions by regional fixed effects, Structural Funds may turn out to have a positive impact. On the one hand, fixed-effect esti- REGIONAL POLICIES AND EU ENLARGEMENT 91 mates are inappropriate with heterogeneous dynamics. On the other hand, at a more technical level, assume this did happen- regional fixed effects are catchall variables that may well hide the true determinants of convergence, and no conclusion could be reached without an explicit model and measurement of such regional fixed effects. 7. Ireland's public expenditure and taxes are slightly below 30 percent of GDP, while the EU avrerage is about 42 percent Biibliography The word processed descrtbes informally reproduced works that may not be com- monly available through libranes. Arevalo, P. 2002. "Crecimiento y Convergencia en Espafia y sus Regiones: 1960-2000." Ph D. diss Universidad Carlos III de Madrid, Spain. Aslund, Anders, Peter Boone, and Simon Johnson. 2001. "Escaping the Under- Reform Trap." IMF Staff Papers 48:88-108. Ben David. 1994 Quarterly Journal of Economics. Boldrin, Michele, and Fabio Canova. 2001. "Inequality and Convergence Recon- sidenng European Regional Policies." Economic Policy 32 (April):205-45 Canova, Fabio, and Morten Ravn. 2000 "The Macroeconomic Effects of German Unification: Real Adjustment and the Welfare State" Review of Economic Dynamics 3(3):423-60 Ederveen, Sief, Henri L. F de Groot, and Richard Nahuis. 2002. "Fertile Soil for Structural Funds? A Panel Data Analysis of the Conditional Effectiveness of European Cohesion Policy " Netherlands Bureau for Economic Policy Analysis, Amsterdam. Processed. EU Commission. 1998. Reform of Structural Fund. Brussels European Council. 1999. "Presidency Conclusions, European Council, 24 and 25 March 1999 " Available at http://www inforegio.cec.eu int. Eurostat. 2000a Statistics in Focus, Economic and Finance, Theme 2, "European Union FDI with Candidate Countries. An Overview " 26/2000. - . 2000b. "Statistics in Focus, General Statistics, Theme 1, Regional GDP in Candidate Countnes." 2/2000. 92 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH . 2001a. "Free Data, General Statistics, Theme 1, Business Investment." 11/5/2001. . 200 lb. "News Release, General Statistics, Key Data on the Candidate Coun- tries." 129/2001. . 2001 c. "Statistics in Focus, Economic and Finance, Theme 2, Value Added, Employment, Remuneration and Labor Productivity in the Candidate Countries" 13/2001. . 2001d. "Statistics in Focus, General Statistics, Theme 1, Regional GDP in Candidate Countries." 4/2001. . 2001e. "Statistics in Focus, General Statistics, Theme 1, Regional Labor Force in the EU: Recent Patterns and Future Perspectives " 2/2001. . 2001 f. "Statistics in Focus, General Statistics, Theme 1, Regional Popula- tion Changes in Candidate Countries." 6/2001. . 2002a. "Free Data, General Statistics, Theme 1, Employment Rate-Total." 12/7/2002. . 2002b. "Free Data, General Statistics, Theme 1, Unemployment Rate- Total." 12/7/2002. . 2002c. "Statistics in Focus, Economic and Finance, Theme 2, Candidate Countries' National Accounts by Industry." 17/2002. .2002d. "Statistics in Focus, Economic and Finance, Theme 2, the Evolution of FDI in Candidate Countries: Data 1995-2000." 3/2002. - . 2002e. "Statistics in Focus, General Statistics, Theme 1, Regional GDP in Candidate Countries." 2/2002. Franz, Wolfgang, and Viktor Steiner. 2000. "Wages in the East German Transition Process: Facts and Explanations." German EconomLc Review 1:241-70. Funck, Bernard, and Lodovic Pizzati, eds. 2002. Labor, Employment, and Social Policies in the EUEnlargement Process. Washington, D.C.: World Bank. Garibaldi, Pietro, Nada Mora, Ratna Sahay, and Jeromin Zettelmeyer. 2001. "What Moves Capital in Transition Economies." IMF StaffPapers 48:109-45. IMF. 2001. "Special Issue on Transition Economies: How Much Progress?" IMF Staff Papers. Washington, D.C. REGIONAL POLICIES AND EU ENLARGEMENT 93 Jones, Ronald W 2000 Globalization and the Theory of Input Trade. Cambridge, Mass. MIT Press. Mlarmon, Ramon, ed. 1996 "La Economia Espafiola: una visi6n diferente " Uni- versidad Pompeu Fabra, Barcelona, and Ministerio de Economia y Hacienda. Barcelona. Antoni Bosch. NMartin, Carmela, Francisco J. Velazquez, and Bernard Funck. 2001. "European Inte- gration and Income Convergence: Lessons for Central and Eastern European Countries." World Bank Technical Paper 514. Washington, D.C Processed. NMartin, Carmela, Jose A. Herce, Simon Sosvilla-Rivero, and Francisco J. Velazquez. 2002 "La Amphaci6n de la Uni6n Europea Efectos sobre la economia Espafiola " Colleci6n Estudios Economicos de la Caixa, No 27. NMiles, David, and Andrew Scott 2002 Understanding the International Economy. London School of Business, U.K. OECD. 2000. Main Economic Indicators Paris. Ragnitz, Johachim 2001. "Lagging Productivity in East German Economy Obsta- cles to Fast Convergence." Institut fir Wirtschaftsforschung, Halle, Germany Processed Sinn, Hans A. 2000. "Germany's Economic Unification: An Assessment after Ten Years." NBER Working Paper 7586 Boston: National Bureau of Economic Research. Tabellini, Guido. 2002. Pnnciples of Policymaking in the European Union: An Eco- nomic Perspective IGIER-Universita Bocconi, June. Milan, Italy. Processed Vaitilingam, Romesh, ed 2002. "Who's Afraid of the Big Enlargement?" CEPR Pol- icy Paper 7 London: Centre for Economic Policy Research W orld Bank. 2001 a Employment and Labor Market in the Czech Republic Wash- ington, D C. - . 200 lb. Poland's Labor Market. Washington, D.C. - . 2001c Transition The First Ten Years. Washington, D C - . 2002. Expenditure Policies toward EU Accession. Washington, D C. CHAPTER 4 I)iscussion of "Regional Policies and EU Enlargement" by Boldrin and Canova, and "Public Policies and Economic G-eography" by Martin Carole Gamier Each of the two chapters offers ample scope for debate. They will, however, be discussed jointly. This is no easy task smce they start from very different theoretical views, namely the neoclassical and the new economic geography models There may nevertheless be some advantages in a joint discussion. It may illustrate the dilemmas faced when con- sidering long-term growth and convergence issues with a view to policy implications. These dilemmas are worth summarizing briefly. Different views of the world are captured by different models based on different sets of assumptions. None of these views can be ruled out with any degree of certainty, not only because there are limi- tations in the data and in the methodologies used by the models presently available but also because researchers tend to focus on a single theory without attempting to imtegrate or at least compare alternative theories. These different views are supported by limited empirical evidence, which furthermore has often no clear link with policy guidelines. When, nevertheless, strong policy statements are made-as is the case here-they may entail divergent policy recommendations. For ease of presentation, I shall follow the structure of the Boldrin and Canova chapter. The initial conditions and growth prospects of the Central and Eastern Euro- pean countries (CEECs) will first be bnefly discussed. Then consideration will be given to real convergence issues and policy recommendations, starting with what the authors consider as lessons from the past before turning to their policy suggestions for the future. It is in this policy-oriented part that the work of Martin, which does not touch upon enlargement issues, will be discussed. 95 96 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH l[nitial Conditions and Growth Perspectiives irn the CEECs There would be no point in discussing the purely descriptive analysis of the evolu- tion over the past decade of some main aggregate indicators. More open to debate would be an analysis of the underlying explanatory factors and of their likely impact on future performance. But this goes well beyond the scope of this chapter and may explain why, contrary to expectations, there are no obvious links between the pre- sentation of the CEECs' situation, as a whole and individually; their growth per- spectives; and the policy recommendations that are formulated later. Of interest to a discussion are the future growth prospects of the CEECs, taking into account the effects of enlargement. This forward-looking question is answered through a backward-looking approach. Three scenarios are proposed. First, the "no- change" scenario projects the behavior observed in the period 1991-2000. This benchmark is highly debatable since it includes the recession that has followed the transition to a market economy. Starting from 1994, when most countries had man- aged to recover positive growth rates, would significantly alter the results and call into question the authors' opinion that the CEECs' position compared with that of the European Union (EU) 15 is no different than it was a decade ago. From 1994 to 2000, the CEECs have achieved a positive growth differential vis-a-vis the EU15. This does not imply that the 1994-2000 period would be an ideal benchmark. It would indeed be difficult to distinguish between the countries with high but transi- tional growth and those initiating a more sustainable growth trajectory. What needs to be pointed out is that the very specific context of transition that is downplayed in the chapter does not allow for a "trend" methodology. Caution is also required with regard to the other two scenarios. They are based on the claim that candidate countries are similar to the three southern countries, Greece, Portugal, and Spain. The second scenario assumes convergence of the CEECs in a probabilistic sense to the same steady-state level and the third a similar boost in growth rates after accession. The second scenario is driven by the initial pre- diction of absolute convergence of the neoclassical theory, a prediction considered implausible since the mid-1990s, when the possibility of having only conditional convergence, because of differences in fundamentals between economies, gathered support. The third scenario has no articulated theoretical point of view. In both cases, applying to each of the CEECs-regardless of their initial conditions, specificities, and potential-the same constrained "level or growth effect" leads to a biased con- vergence toward the mean. The best performers do worse after accession, the worst performers better. Such an exercise adds little to the understanding of the growth potential and patterns of countries that are rightly acknowledged as being very dif- ferent. Nor does it give much insight into their potential evolution as a whole com- pared with that of the EU 15. In both scenarios, a major problem is that the assumed similarity of the three southern countries is far from warranted. This is clearly demonstrated by the so- DISCUSSION OF "REGIONAL POLICIES" AND "PUBLIC POLICIES" STUDIES 97 called index of similarities itself, used to justify the approach, which is restricted to four indicators. Just looking at the first indicator-openness-as an example, the dif- ierence is striking. Even at the time of accession in 1973, Ireland's degree of open- ness was a lot closer to that of the CEECs, while other indicators are not more dissimilar than those of southern countries-or than their average. Assuming con- vergence to the Irish steady-state level would have drastically changed the results. This does not mean that all CEECs would be expected to mimic the Irish perfor- mance. It simply indicates that the similarity assumption used in the approach is not r obust enough for conclusions to be inferred. Beyond the lack of evidence from the index, the underlying assumption of similar transition patterns is rather dubious. The jump in unemployment in Spain from 1975 to 1985 (from some 3 percent to 21.6 per- cent) that is viewed as the effect of transition on the labor market is mainly the con- sequence of a very long and severe economic crisis, probably amplified by structural rigidities, resulting from the oil shock and world recession Other major structural differences can be identified, for instance, in the product markets. The 1976 nation- alization in Portugal basically froze the public sector until 1989, while the so-called state corporatism in Greece implied direct and indirect government interference in the running of pnvate industry. In addition, though the authors claim that history mat- ters and mention some differences in historical circumstances with earlier entrants, a crucial difference is ignored. There had been convergence between some CEECs and Western Europe until divergence set in after World War II. A country like Czechoslovakia had significantly higher gross domestic product (GDP) per capita growth rates than Western Europe in 1870-1913 and 1913-50 (more than double in the latter period). Even in 1950, its GDP per capita was much higher than those of Greece, Portugal, and Spain and close to that of Germany. Regaining positions may imply different patterns of catching up than those observed in countries that have started from initially low GDP per capita. To give a more forward-looking dimension to this econometric exercise, an addi- tional scenario that identifies the conditions and potential for catching up and exploits the full benefits of EU membership should be considered. Impulses to a growth model such as those stemming from further scope for trade and possibly foreign direct investment, the reallocation of resources to more productive sectors, the deep- ening of structural reforms, or a clearly focused investment strategy would be help- ful in outlining a realistic growth path, given initial conditions and policy developments The standard Solow growth accounting model with some adaptations would be a useful starting tool that could yield interesting insights. Policy Lessons from the Past As mentioned before, policy issues and recommendations are not considered in the light of the conditions and prospects of the CEECs but are based on what the authors call "old, useful lessons " These lessons are basically three and concern the respective 98 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH impacts on convergence of trade integration/openness, the EU cohesion policy, and national policies. Trade Integration/Openness The impact of trade on long-term growth and convergence deserves discussion since the evidence remains somewhat mixed. While the benefits of trade at the aggregate level are generally accepted, their distribution is more controversial. Nevertheless, strong and contrasting views are expressed by Boldrin and Canova, on the one hand, and Martin, on the other. Boldrin and Canova claim that the pre- dictions of new growth/trade theory on agglomeration and divergence are rejected by the data and that openness is beneficial or at least neutral to poorer areas. Martin seems to endorse Krugman's views in stating that specialization and geographical concentration, though lower than in the United States, have increased in the EU 15 over the past 20 years. Since labor mobility is low in the EU, increased disparities and even divergence would not be precluded. These are strong statements given the evidence provided by the authors and that currently available. Boldrin and Canova do not provide any data or empirical evidence. They list some examples of trade openness at national and international levels. Observing con- vergence during the same period, they conclude that causality runs from trade to con- vergence. The sensitivity of the issue would call for a more in-depth analysis together with some empirical grounding. Account should be taken of the other factors that influence convergence, of the possibility of reverse causation (that is, trade may be endogenous to the process of growth), and of scarce analysis on the impact of trade on poorer areas. For instance, it cannot be asserted that increasing trade has led to convergence between the U.S. states, since the data on interstate trade are too scarce to isolate the contribution of the latter from the contributions of capital and labor mobility, central government intervention, or other factors. According to Kim (1997), who investi- gated regional specialization and convergence in the United States from 1840 to 1987, the decrease in GDP per capita disparities between the northern and southern states is linked to the shift from industry to services. Mention is also made of trade integration and convergence among the initial six members of the European Economic Community. This relationship has mainly been investigated by Ben David (1996). He observes that in the six countnes and more generally in countries that trade extensively with one another, convergence gener- ally occurs simultaneously with the removal of trade barriers and its speed is higher than in other countries. He also acknowledges that the income gap in six of the Euro- pean Free Trade Association members did not begin falling as obstacles to trade were removed in 1961 but later, and that there had been no regime of interstate trade bar- riers that had to be abolished when income convergence between the U.S. states resumed. He does not investigate as such the impact of trade on poorer areas. He thus DISCUSSION OF "REGIONAL POLICIES" AND "PUBLIC POLICIES" STUDIES 99 concludes that trade policy is not the most important policy from a long-run-growth perspective and that investment in education and infrastructure, preservation of prop- erty rights, and other ingredients are needed to enable a country to enjoy the fruits of openness. One should also add that contrary to the authors' statement, convergence in the European Commission (EC) took place in the context of national transfers. Implicit interregional transfers were made through budget revenues and expendi- tures. In addition, the German scheme of explicit interregional fiscal equalization dates back to 1949/1950-that is, to the constitutional objective to broadly equalize living conditions. Specific regional policies were also implemented in the EC, for example in the Mezzogiorno since the early 1950s and in France (amenagement du territoire) since 1960. There were no EU cohesion policy transfers, but their small financial size compared with national redistribution, state aids, and regional policies should be kept in mind. Thus, the only conclusions one can reach are those that have been evidenced by previous research. First, aggregate growth-such as the exceptional one experienced by the EC during the 1950s and 1960s-is beneficial to convergence, though the mechanisms implied have not been fully explained. Second, trade policy can con- tribute to growth but may not be the most important factor for long-run growth and convergence. Frank Barry, John Bradley, and Aoife Hannan (2001) have pointed out that "openness has been a feature of the Irish economy ever since the 1970s if not earlier and Irish performance over most of the 1960s-1 970s and 1980s exemplified iailure rather than success." Indeed, success came only in the 1990s, when Ireland also benefited from significant EU cohesion transfers. Consistent with the new trade theory that assumes increasing returns, Martin finds some evidence of increasing agglomeration and geographical concentration, at least in the EU, in the empirical work initiated by Briihlart and Torstensson to test the new economic geography theory. It should be specified that agglomeration does not necessarily lead to divergence in the long run. The most recent models of the new economic geography suggest a possible evolution of the spatial patterns of activity that can be stylized via a U-shaped relationship between trade costs and agglomera- tion. With the lowering of trade costs, firms will tend to exploit economies of scale by concentrating production close to larger markets and to other firms, and concen- tration will develop endogenously. When trade costs tend to disappear, higher wage costs in agglomeration lead industries to disperse to regions where they can benefit from labor cost advantages. As the empirical evidence to test the validity of the new economic geography theory was scarce, rather aggregated, and strongly focused on manufacturing, two studies were camed out for the EC They arrive at similar conclusions and give a nuanced view of spatial evolutions at the country level in the EU15. Based on pro- duction data for 14 EU member states and 36 manufacturing industries between 1970 and 1997, the study by Midelfart-Knarvik and others (2000) shows decreasing spe- cialization during the 1970s and a slow and uneven increase from the early 1980s. 100 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Industries have experienced some changes in their spatial concentration. A number of industries initially dispersed have become more concentrated in southem Europe. These are mainly labor-intensive industries such as textiles, clothing, and leather. Among industries initially concentrated, about half remained so (for example, air- craft, motor vehicles). A number of medium- and high-technology industries with high skill intensities (for example, office machinery, televisions and communication, professional instruments) have spread out from the central EU, to the benefit of Ire- land, Finland, and southern countries. The most concentrated services-financial, insurance, real estate, and business services-have also become less concentrated over time. EU Cohesion Policy Boldrin and Canova claim that the EU cohesion policy has had no impact on long- term growth and convergence. This is a very strong statement since EU cofmancing assistance is granted for specific investment purposes and inserted into a complex system of monitoring, evaluation, and control, ensuring that it is spent according to targets. However, one cannot find arguments or empirical evidence to support their claims. The authors refer to some econometric work presented in a previous paper (2001). In this paper, two econometric exercises are reported. In the first, which looks for indirect evidence regarding the impact of cohesion support, 13 convergence-type regressions are run using data for either 101 or 185 NUTS 2 regions for the period 1980-96. The results are mixed. However, they indicate some convergence in labor productivity. Such types of regressions have numerous methodological weaknesses, which have been highlighted in the literature. Reviewing some of the numerous exist- ing econometric studies, Ederveen and others (2002), who argue that government failures reduce the positive effect of cohesion support, stress a main criticism, the lack of data. Researchers draw different conclusions from different data, time peri- ods, and methodologies. In addition, in this specific case, the period retained is too short to adequately encapsulate both the long-term dimension of convergence and the impact of the Structural Funds. It is only at the end of the 1980s that the latter were given an efficiency-oriented design and less negligible financial means. Furthermore, because of their targets (investments in infrastructure, education, and so forth), their impact on the supply side of the economy takes time to materialize. But even more important, no link can be established between the cohesion policy and the conver- gence results of the simple regression methodology used. The impact of the former can be identified only by comparison to a counterfactual situation. This implies the use of models that are at least able to capture the factors that appear constantly in the academic literature as being important in explaining the process of growth as well as the channels via which it can be affected by Structural Funds and also account for their interactions. Obviously, no structural model of growth can be presented by a single equation with two variables. For this reason, the second econometric exercise intended to directly measure the impact of cohesion policy on regional convergence DISCUSSION OF "REGIONAL POLICIES" AND "PUBLIC POLICIES" STUDIES 101 does not throw more light onto the subject. The authors run a simple univariate riegression attempting to explicitly link indexes of regional productivity to a portion (limited to the European Regional Development Fund) of the Structural Funds ieceived Since the regression is not picking up other explanatory variables and since a negative correlation is to be expected between Structural Funds amounts and low productivity growth, the former are likely to act in such a regression as an indicator ifor low productivity growth. Funds are flowing to predominantly rural/agrarian r egions, which may show low productivity growth due to low technology spillovers compared with regions with larger industrial sectors The econometric exercise can give no insight into the impact of the EU cohesion policy on regional growth and convergence. More surpnsing is the fact that the issue of its impact on national growth and convergence is fully ignored. Data are more readily available at the national level, and more sophisticated methodologies can be used. In addition, the authors repeatedly express the view that the NUTS 2 territorial anits, which reflect the member states' subnational institutional divisions (for exam- ple, landers, Communidades Autonomas, regions), are not suitable targets for con- vergence, which would be more adequately assessed at a higher territorial level. Silence on this point may be explained by general agreement on the positive contri- bution of the Structural Funds to national convergence. Even if one excludes the impressive performance of Ireland, the GDP per capita of the three cohesion coun- tries (Greece, Portugal, Spain) compared with the EU15 average, which stood at 68 percent in 1988, has increased to 79 percent. The contnbution of the Structural Funds assessed with the help of two models with different specifications, including an explicit supply side-Hermin and Quest-is positive and significant, as reported in the Commission's (2001) publication In addition, and contrary to Boldrin and Canova's assertion, the EU transfers may have had a positive effect on the profile of convergence in a long-term perspective. Unlike what the neoclassical theory would predict-that is, faster growth while GDP levels are still low, and then a slowdown in convergence-the patterns of convergence in Spain and Portugal since 1960 (with GDP per capita increasing respectively from 59 2 percent and 40.6 percent of the EU average to some 82 percent and 74 3 percent in 2001)' seem rather linear and are even reversed for Ireland. National Policies One can only agree with the assertion that national policies are crucial for long-term growth and convergence. Indeed, the effectiveness of the Structural Funds is likely to be limited if they are inserted into a context that is not conducive to growth. However, there are two points worth discussing in Boldnn and Canova. First, proactive regional policies, focused on improving knowledge and factors of produc- tion, should not be confused with implicit and unconditional interregional transfers. Out of the average DM 180 billion per year gross transfers to eastern Germany in the second half of the 1990s, only about DM 30 billion was investment. Similarly, gross capital 102 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH formation has represented a limited part of the transfers to the Mezzogiomo, which took the form mainly of welfare payments and a high level of public employment. In 1988, support provided by the public sector m the form of public employees' salaries and transfers to households amounted to 49 percent of southern GDP and gross fixed cap- ital formation to 6 percent. Because of budget consolidation, central and local govern- ment current and capital expenditures have decreased since the early 1 990s Second, among the range of national policies that can influence convergence, macroeconomic stability should be more emphasized, as it seems to play a strong role. This can be illustrated by many examples. At the EC level, the 1960s and early 1970s were periods of exceptionally high growth and low inflation. Convergence occurred. After the two oil shocks in the 1970s, macroeconomic conditions deterio- rated in terms of inflation and budgetary positions, especially in the cohesion coun- tries, and so did convergence. At the individual country level, Ireland's GDP per capita compared with the EU average was not much different at the end of the 1980s than in 1960. High convergence set in when an economic policy combining stable public finances and a new approach to industrial relations leading to wage modera- tion was implemented. Conversely, after a phase of unbalanced policy mix and income divergence in the 1980s, Greece displayed a positive growth differential vis- a-vis the EU from 1996, when a stability-oriented policy was implemented to fulfill the criteria for participation in the European Monetary Union Macroeconomic sta- bility may not be a sufficient condition for growth and convergence, but it would seem to be a prerequisite. Recommendatdons for the EU Cohesion IPolicy The description of the EU cohesion policy that is provided does not allow for a clear understanding of the facts, a necessary step before formulating and discussing reform proposals. For instance, the revision of the Structural Funds in 1999 and their main features are overlooked; for example, objectives are only three since 2000; a perfor- mance reserve has been introduced; some of the Objective 1 regions have reached a GDP per capita higher than the 75 percent threshold and EU assistance is subse- quently being phased out; and the geographical concentration of support has been increased. Eligibility for the Cohesion Fund has been based since its creation on gross nationalproduct and not gross domestic product. This is why Ireland is a beneficiary. In many cases, the assertions are not in line with reality. The EU assistance cannot be assimilated into income maintenance programs. EU budgetary transfers are dis- tributive, as they are made to the less wealthy. But spending is allocative, as it is tied to specific investments. Financial and physical outputs are monitored and controlled. The principle of additionality, which entails verification, is neglected. The claim that convergence at the NUTS 3 level is an objective of the policy is unfounded. It has never been. Subsequently, no attempt to assess convergence at this territorial level has ever been made in the Commission's reports and evaluations. DISCUSSION OF "REGIONAL POLICIES" AND "PUBLIC POLICIES" STUDIES 103 Nevertheless, a few reforms are suggested. These basically touch on two main issues. the criterion and territorial level for assistance, and the forms of assistance Who Should Benefit? The territorial level-national or regional-to be targeted in terms of both eligibil- ity and long-term growth strategy is a recurrent issue, raised in the Commission's second cohesion report (2001), which launched a debate on the future of the cohe- sion policy post-2006. It is of special relevance to the CEECs that display stronger income disparities at the national than at the regional level The suggestions made by Boldrin and Canova are not fully clear and contain contradictions. While they start advocating convergence at the country level, they end up with two proposals targeted at the regional level. funding only regions that are federal units-a proposal that targets the constitutions of member states but has little relevance from an economic development perspective-and lowering the threshold in the definition of disadvantaged regions. Only the latter, reduced to 50 percent of the presumably current EU average to ensure concentration of the finan- cial means on the very poorest, is fully clear Since there is no empirical evidence of higher returns in the extremely poor regions, such predominance given to equity is surprising in a chapter that criticizes at length the equity-oriented policies pursued in the Mezzogiorno and eastern Germany. Efficiency considerations could point to an opposite approach, as outlined by Martin. He shows that national convergence in the EU15 has been accompanied by generally increasing regional disparities within countries and thus finds evidence of the importance of efficiency gains due to agglomeration. There is indeed some evidence that catching-up countries enjoying high national growth often see a widening of interregional disparities, as national growth tends to be dnven by growth poles effects that emerge in capital cities and other major agglomerations. This has been the case in Ireland and Spain, while Greece has until the mid-I 990s expenenced low growth and a fall in regional dis- persion.2 Thus, in the early stages of the catching-up process there may be a poten- tial trade-off between national and regional convergence. Efficiency would ultimately lead to focusing public investment on the regions with the highest poten- tial rather than on the most backward ones. At the least, an appropriate balance has to be found between equity and efficiency, and this may be best ensured in catching- up countries by an approach and strategy for economic development at the national level. Any extreme policy recommendation, such as depriving the relatively less poor regions of poor countries of EU support, can be misleading, as it may not improve economic development and convergence very much. WVhat Should Be Financed? While Martin expresses doubts about the effectiveness of state aids to attract and develop enterprises in less favored regions, Boldrin and Canova take a more "insti- tutional" view of the issue. They distinguish between large-scale projects related to 104 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH industrial and agricultural restructurng, which could be financed by the EU, and sup- port to small and medium enterprises, which should be the task of national govern- ments. It would indeed be difficult to find any empincal evidence that state aids have been instrumental in fostering regional development, as exemplified by the Mezzo- giomo and eastern Germany experiences. Such aids have important dead-weight effects. However, regional state aids represent only some 14 percent of total state aids in the EU 15, which means that the regional component has to be tackled in a wider context. This is why a coordinated stance committmg member states to a downward trend is being implemented at the EU level. Total state aids have thus decreased by 12 percent from 1995/1997 to 1997/1999, and aids to manufacturing from 35.8 to 27.6 billion Boldrin and Canova recommend concentrating funds on "public goods" under- stood as different types of infrastructure, including transport infrastructure. How- ever, as highlighted by de la Fuente, thc productivity effects of infrastructure investment have been the subject of an economic debate that is still ongoing. Mar- tin insists on reducing transaction costs on exchanges of "ideas" and calls for cau- tion in supporting transport infrastructure, as it may be detrimental to less favored regions (for example, Mezzogiomo). He acknowledges, however, that public infra- structure may have very different effects depending on the region concerned. Indeed, the available empirical evidence would hardly support Martin's views For instance, several analyses find that infrastructure has had a positive and highly sig- nificant effect on regional growth and convergence in Italy. There is some evidence that the construction of highways in some California counties and U.S. states has even been detrimental to the others. The only conclusion is that the context is impor- tant. No one could imagine that growth and catching up would take place without infrastructure. But infrastructure per se is not a panacea and may not necessarily lead to growth. Finally, the types of intervention financed by the EU cohesion policy appear to be rather consensual. Caution may be advocated m some cases, and the weight to be given to the different components of the investment mix may differ (for instance, both Martin and Boldnn and Canova seem to put less emphasis on human capital), but except for the cofinancing of state aids to the corporate sector, no drastic change is advocated concerning the basic ingredients of the policy. To conclude, the main lesson I would like to draw is one of modesty It would certainly be convenient if an economic process as complex as growth could be cap- tured by one or a few variables. Unfortunately, this is not possible A growth policy implies a mix of ingredients ranging from national policies for macroeconomic sta- bility and a smooth functioning of markets to improvements in factors of production and access to knowledge. The specific mix to be implemented is a function of each particular situation. Each expenence of growth is unique. Overlooking such diver- sity or giving predominance to one factor while ignoring the others is likely to lead to misleading policy suggestions. DISCUSSION OF "REGIONAL POLICIES" AND "PUBLIC POLICIES" STUDIES 105 Piotes The views expressed are those of the author and do not necessarily reflect those of the Euro- pean Commission. 1. These changes occurred despite relatively reduced convergence in Spain in the 1980s com- pared to the previous decade, where GDP per capita had even peaked at some 79.5 percent of the EU average in 1975 and 1976. 2. For a more complete discussion of the issue based on available data, see The EUEconomy 2000 Review (European Commission 2000) 13ibliography T'he word processed describes informally produced works that may not be commonly available through libraries. Blarry, Frank, John Bradley, and Aoife Hannan. 2001. "The Single Market, the Struc- tural Funds and Ireland's Recent Economic Growth." Journal of Common Mar- ket Studies 39(3):537-52. IBen David, 1996. Technological Convergence and International Trade. Foerder Institute for Economic Research Working Paper 4196. Tel Aviv University. IBoldrn, Michele, and Fabio Canova. 2001. "Inequality and Convergence: Recon- sidering European Regional Policies." Economic Policy 32:205-53. Ederveen and others. 2002. "Does European Cohesion Policy Reduce Regional Dis- parities? An Empirical Analysis." Netherlands Bureau for Economic Policy Analysis, The Hague. Processed. Ederveen, Sief, Joeri Gorter, and Richard Nahuis. "The Wealth of Regions The Impact of Structural Funds on Convergence in the EU." Netherlands Bureau for Economic Policy Analysis, The Hague. Processed. European Commission. 2000. The EUEconomy 2000Review. Luxembourg Official Publications Office. - . 2001. Second Report on Social and Economic Cohesion. Luxembourg Official Publications Office. 106 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Kim, Sukkoo. 1997. Economic Integration and Convergence: U.S. Regions, 1840-1987. NBER Working Paper 6335. Boston, Mass.: National Bureau of Economic Research Midelfart-Knarvik, K. H., H Overman, S. Reading, and A. Venables. 2000. The Location of European Industry. Economic Papers 142. Luxembourg: European Commission. (CHAPTER 5 Issues and Constraints of Regional Convergence Alfred Steinherr I really like de la Fuente's work, but I cannot believe the results in his chapter despite his very reasonable and extremely good points. I am not convinced by his argument that regional policies produce a dramatic increase in investment and, therefore, employment has very high returns, up to 40 percent. We saw already in previous chapters that in a way economics stacks the cards against that result simply because, as we have seen, there are missed externalities and scale economies for investments outside of the established centers. In poorer areas one cannot expect to achieve social rates of return of 40 percent Second, statistics is also a bit of a problem because the way regions are defined is problematic and gives a very unclear picture Let us think about what regional policy based on transfers could achieve. What are the effects a study may wish to capture? First there could be a financial constraint for the entire economy so that national savings are not sufficient and there could be limited access to foreign borrowing. I do not think that any member of the European Union (EU), given that we now have a very large and relatively efficient EU cap- ital market, faces such an overall macroeconomic constraint. Let us therefore put that argument aside. Second there could be a microeconomic constraint. There may be projects that do not find financing. While I can think of many projects that are so unattractive that they will not find any financing, 1 have difficulty identifying projects that are economically sound but cannot find the resources in Europe for financing. 107 108 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH I think we need to comb through projects in poorer regions and identify those that would not have been executed without transfer payments from the European regional aid program. Of course, regional funds add to the money a country has at its disposal. They may have induced effects, which Antoni Castells' and Marta Espasa's chapter takes into account. For instance, if a good project would not have needed the subsidy, the rate of profit is increased and the recipient institution may mcrease its overall investment- but not necessarily in the poor region concemed. So there could be an induced effect in the form of more investment, but this may result in overinvestment. More is not always a good thing. I think there is an optimum that is determined by the standard mar- ginal conditions. That approach is therefore, in a way, misleading. What is not shown is which projects would have been crowded out had these funds not been available. There is no support for contentions that good projects in Europe sometimes cannot be carried out because financing is lacking. Now I want to discuss the statistical problems that arise in the application of European regional policy In order to make regional policy operational, the Commission has made an attempt to classify regions. Its general geostatistical plan is known as the Nomencla- ture des Unites Territoriales Statistiques (NUTS) classification, and there are differ- ent levels of refinement. The NUTS level that is mostly used for European regional policy-NUTS 2-refers to the provincial or county level, such as the French Regions or the German Regierungsberzirke. Map 5-1-which I borrowed from the sixth periodic report on the regions- shows the variation in the standard of living among these regions. The darker the region, the wealthier it is. Clearly, there are five large problematic zones: 1. Virtually all the regions in Greece 2. The Mezzogiorno in Italy 3. Most of the regions in Portugal and southern Spain 4. The most northern regions of Europe 5. Regions in East Germany. The geographical distribution of unemployment very much reinforces this pattern. The Commission has prioritized the affected regions and labeled them Objec- tive 1 areas. From a technical point of view, Objective I concerns in general those NUTS 2 regions with a per capita income lower than 75 percent of the EU average (there are, admittedly, a few exceptions to this rule). In map 5-2 they are the dark ISSIJES AND CONSTRAINTS OF REGIONAL CONVERGENCE 109 MAIP 5-1 GDP PER HEAD BY REGION (PPS), 1996 GDP PER HEAD BY REGION (PPS), 1996 - Index, EU 15 = 100 E7° <75 LI 75-90 90-110 ouadu|a MarnnSqure| Reunlon :ostatO t 110 EUROPEAN INTEGRATION, REGIONAL POLiCY, AND GROWTH MAP 5-2 OBJECTIVE 1 AREAS OBJECTIVE 1 AREAS W Objenlve 1 Akeas . W Transironal Support Under Objectve 1I /A V ' s . KILOMEMR$ IBRD 32281 v Canary Is (SP) French 1Azones. (Port) t ( Fr (FG a (Fr) f U X@' A4'~( W: W¢ :: , ' 0.0 O -0.01 _ Falling behind -0.02 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 Real GDP per worker, 1982 Soturce Author's calculations Yet to believe that productivity gaps are immutable is a mistake. Important "within movements" take place in the distributions, which I shall illustrate again with a graph (figure 5-3), showing a plot of the average annual growth performance against the initial income per capita in 1982. The horizontal and vertical lines in the middle represent the EU averages. It is easy to distinguish four zones. Convergence toward the EU average took place in the regions situated in the northwestern and southeastern parts of the chart. Regions that have been lagging further behind are in the lower right-hand portion. Some of these "within movements" may offset one another, which may explamn the quite constant overall distribution. But, in spite of this partial convergence, the pattern in the tails of the distribu- tion has been remarkably stable For instance, the 25 regions with the lowest unem- ployment rates have hardly changed over the past decade, with rates steady at around 4 percent. In the worst affected regions, by contrast, rates have climbed from as much as 20 percent to nearly 24 percent and-in some cases-are continuing to increase. So why are the goals not being met? I do not think it would be fair to say that the EU does not try hard enough to address the problem, but two factors may prevent policies from achieving the desired result: ISSUES AND CONSTRAINTS OF REGIONAL CONVERGENCE 115 1. Market failures 2. Rigidities in the labor market. Let us discuss market failures first. Imagine an economic paradise-a system in which economic agents behave rationally and production factors are freely mobile. To be more precise, capital and labor go to this paradise because returns there are the highest In such a perfect world, there are three factors that impair the working of the mar- ket in such a way that persistent economic imbalances may still prevail. I will clar- ify them briefly First, the market outcome is less than optimal when there are substantial tech- nological externalities. By this I mean that firms learn from one another how to do things better through observation, casual conversations, and so forth. As a result, fimis tend to locate in zones where there is already economic activity and a well- developed market, because in that way they benefit from investments by other agents without having to pay for the gains Of course, companies will try to account for this externality in their location decisions, reasonmg along the following lines. "If I move to that region and invest, I might give a competitive advantage to others, for which I cannot ask for remuneration Perhaps it is better to stay put." Therefore, the market results in an outcome that further disfavors the lagging regions. This brings me to the second market failure-pecuniary externalities, another aca- demic term that requires explanation. As a result of the productivity differences among regions, both skilled workers and capital will tend to flow into the well-developed areas However, firms and workers do not take into account the impact of their relo- cation on the well-being of those who stay put or those who live in the destination region. For instance, migration will put a downward pressure on the wage level in the destination region, while demand, and hence prices, will be boosted. The market may, in this case, lead to excessive imbalances. The last market failure is a coordination problem. In some cases a region does not take off because a minimum threshold of economic activity has not been reached No one knows how a business would perform in such an area, because too little price information is available Lack of adequate information then prevents the develop- me nt of a network of services and intermediate goods suppliers, so there is a vicious circle of persistent underdevelopment With these flaws operating, it may not be very stratghtforward to develop an effective regional policy, not least because it may not be very clear what the domi- nant market failure is. The least one can say is that an efficient regional program should be sufficiently tailored to the region-specific problem. By consequence, not every region needs the same type of infrastructtre to get its growth engine going, and I think it is worth underlining this point. Moreover, the infrastructure-and regional policy in general-should help regions to develop their comparative advantages. 116 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Thus, a regional policymaker must understand the externalities at play, and the pol- icy should be a function of the local comparative advantage In fact, too loose a regional approach-one that is not sufficiently fine-tuned- might lead to perverse effects. I will illustrate this with an example. Suppose that a new highway is built between a relatively lagging region and an industrial region. Lower transportation costs will be an incentive for firms and skilled workers to move to the industrial region. In that way, their businesses can benefit from technological externalities, while the competitive price advantage associated with local production in the prehighway period is still maintainable because of the lower transport costs. Thus, the lagging region may become a net importer of goods. Of course, this does not necessarily mean that the living standard will go down. But, in the longer run, it may imply that unemployment rates will rise, especially if the relatively immobile unskilled workers become a dominant group in the local labor force and if-for what- ever reason-they are unable to retrain themselves. Yet higher unemployment is the opposite of what the policymakers had in mind. Discussing a low degree of labor mobility and skill differences leads me natu- rally to the second problem I mentioned earlier-rigidities in the labor market. As many people know, Europeans still have a tendency to stick to their geographical ori- gins, despite substantial regional differences in the standard of living. Studies indi- cate that the degree of interregional labor mobility, for instance, is roughly only one-third of that in the United States (see table 5-2). Important reasons behind this phenomenon are an inflexible market for housing-that is, substantial costs for sell- ing and buying due to high registration taxes and the like-and a lack of information on jobs elsewhere. Moreover, in a broader European context, language barriers con- tribute in an important way to the labor immobility TABLE 5-2 EUROPEAN VERSUS U.S MOBILITY United United Years States Germany Italy Kingdom Percent of the regional population 1970-79 1 20 0 27 0 37 0 47 1980-89 0 84 0 34 0.33 0 26 1990-95 087 031 0 40 0.20 Average net international migration Average 100 32 38 32 Source Obstfeld and Pine, www nber org ISSIJES AND CONSTRAINTS OF REGIONAL CONVERGENCE 117 Now, limited labor mobility would not be too much of a concern if the unem- ployed were able to price themselves back into the labor market. However, wages in many European countries are often influenced by factors decided at the national level, such as wage bargaining between unions and employers, wage indexation, unem- ployment packages, or social security taxes. Such a system does not necessarily yield wages that are in line with the degree of labor productivity in every region. The result is that workers in poor regions-mainly relatively unskilled workers-are priced out of the market In my view, this factor has played a tremendous role in the persistent unemployment rates in eastern Germany and the Mezzogiomo This story would not be complete without the buzzword "new economy." Will subsidizing "new technologies" lead to fewer regional imbalances? Even though such technologies are tremendously useful, their implementation may have undesired effects from a cohesion point of view given the current European setting Let me explain this point The "new economy" may eliminate workplaces, especially among those less skilled, but it may at the same time create well-paid jobs for people with high human capital and infonnation technology skills Add to this the European tendency toward downwardly ngid wages and labor immobility. The new economy would then hold the potential to increase both unemployment and wage inequalities. In this respect, the rec- ommendations of the Portugal summit may thus work against greater social cohesion. But there is good news, too. I understand that new technologies also have the potential to reduce the need for regional migration. People will be able to work from horne while being paid the salaries that prevail at headquarters. Given the three mar- ket failures mentioned above, this force might reduce the impact of pecuniary exter- nal [ties New technologies may thus reduce the level of imbalances. It remains an open question whether new technologies can offset the annoying problem of failure of coordination, too. Some people argue that it does not matter where your main server is located if you are running an e-commerce business. I do not agree If a region has no local network of services for equipment repair, legal and accounting services, food stores, and the like, and no flexible labor market for work- ers with the required skills, I would not be inclined to start my business in that region, not even in this "new economy" century. CHAPTER 6 European Integration, Regional Policy, and the Nonintervention "Hands-off" Approach: Some Comments on the Boldrin and Canova Study Jiri Blaiek The Boldrin and Canova study, "Regional Policies and EU Enlargement," covers a wide range of issues, provides a lot of empirical analysis, offers a wealth of opin- ions, and has several policy implications. My comments will be structured into three sections: 1. General comments on the fundamental assumptions of the authors' approach 2. Discussion of a few specific statements or policy recommendations 3. An attempt to provide some experience on regional development and regional policy in the Czech Republic in the European context. General Comments on Fundamental Assumptions The authors clearly adopt a neoclassical nonintervention approach and suggest con- tracting the public sector and removing the European Union (EU) regional policy because it is a costly, bureaucratic, and ineffective intervention. I am not going to say that the neoclassical nonintervention model of manage- ment of economy is wrong and the competing model of social market economy'- now quite widespread in the EU countries-is right The principal difference between 119 120 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH the two is in their solution of a basic dilemma-the dilemma between the pnnciple of solidanty and pnnciple of meritocracy. Neoclassical/neoconservative approaches (represented, for example, by F. Hayek or M. Fnedman) stress individual responsibility and the pnnciple of meritocracy, while the leftist or social democratic orientation (for example, J. M. Keynes, W. Molle, M. Castells, and D. Harvey) accent the role of the state and the principle of solidarity. My personal belief is that there is a need to search for a proper balance between these principles and that any one-sided approach can do a lot of harm. (Consider, for example, the economic and moral crisis caused by the nonintervention approach of the conservative Czech government in the first half of the 1990s or the reasons lead- mg to the resignation of M. Thatcher.) An important point is that in the economy and society there is frequently a trade-off between short-term and long-term efficiency and prosperity Therefore, what might seem very efficient (or inefficient) in the short term might be highly inefficient (or efficient) over the long term. In addition, a one- sided stress on efficiency arguments might lead to monopolization of the economy and-due to loss of competitive pressure-to its gradual degeneration (see Hampl and others 1999). The rationale for cohesion policies is based on both economic and social motives. The list of these arguments is quite extensive and ranges from improving efficiency by removing bamers to development and better utilizing resources to mod- erating inflation pressures by decentralizing economic activities from congested urban areas As to social motives, Molle (1997) stresses that a huge inequality is morally unacceptable for large population groups. However, in the postcommunist countries, there is at least one additional motive for cohesion policy. An insensitive approach to social and regional problems might lead to frustration in a sizable part of the population with limited opportunities and thus discredit not only the market economy but even the system of parliamentary democracy itself. This situation would have profound long-term consequences for the future development of these countries (as illustrated by the contrasts in public opinion on the acceptance of the former communmst regime in Russia versus the coun- tries of Central and Eastem Europe and the consequent differences in their political stability and their development trajectones). Moreover, in the case of regional pol- icy implemented on the European level there are also other arguments in its favor, some of which are clearly of an economic nature (see also Molle 1997). So, why is there a regional policy on the level of the EU? The main arguments can be summa- rized as follows: I. The oldest argument for EU regional policy was directly connected with the formation of a larger market with more intensive competition. The first aim was to facilitate and coordinate the restructuring of old industries under con- ditions of stronger competition in a larger market. SOAE COMMENTS ON THE BOLDRIN AND CANOVA STUDY 121 2. A second goal was to eliminate the distortion of free competition by an equal approach toward the provision of subsidies to private firms in all member countries. 3. The weakest regions are located in the weakest states with the smallest potential to support these regions sufficiently 4. Finally, and most important, creation of the EMIU (European Monetary Union) requires that member states have a similar level of socioeconomic development and similar macroeconomic charactenstics, as adaptive mech- anisms in some countries or regions are very weak. This concerns especially the limited potential of adaptation via international migration within the EU and the negligible volume of the EU fiscal policy in relation to volumes allo- cated to national fiscal policies. The discrepancy between strong and uni- fied monetary policy and weak EU fiscal policy is one of the principal threats to proper functioning of the EMU (In addition, national fiscal poli- cies are nearly uncoordinated-for example, there are different tax sys- tems ) From this point of view, the EU regional/cohesion policy stipulates at least on a very limited scale the stabilization role of fiscal policy. In this context, however, one has to stress again that the overall volume allocated to cohesion policy (0.47 percent of EU gross domestic product [GDP]) is very small, especially when compared with the share allocated to redistrib- utive policies within national fiscal policies or to the volume of international trade, which in some countries exceeds the GDP. Boldrin and Canova's main conclusion is that the EU regional policy did not make any significant contribution toward growth and convergence The problem with con- clusions derived from these types of empirical analyses is that they do not compare the present situation with a situation in which regional policy had not been imple- mented at all. Only such a comparison-however difficult-would allow for draw- ing such conclusions. In addition, there is a parallel with social policy. For example, one might say that social inequality has increased dramatically over the last decade in all transitional economies, so the social policy was ineffective-and therefore suggest abolishing it completely. Moreover, regional policy can be and should be considered an intervention that helps prevent social problems (and hence saves expenditures on both active and pas- sive labor market policy). In other words, regional policy, by stressing the enhance- ment of firms' competitive position operating in assisted regions (via elimination of infrastructure deficits, support for a productive environment, and human resource development) aims at eradication or at least moderation of the roots of the problems, nct just at mitigation of their consequences 122 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Practical examples of this relation between regional and social policy are the structurally affected regions in transitional economies that were suddenly exposed to intensive competition in the European and global markets. Without national support policies, including regional policy and also preaccession support programs, espe- cially Poland Hungary Aid for Restructuring the Economy (PHARE), a considerably higher number of people made redundant by radical restructuring of heavy indus- tries-together with their families-would fall into poverty, with related implications for social marginalization. The experience from cities in both Western Europe and the United States shows what a tremendous challenge it is to fight urban problems when a vicious circle of social deprivation has been in operation for decades or even for generations Intervening in the early stages of formation of this negative cumu- lative process is much more effective and efficient. Finally, in the case of implementation of cohesion policy there are also impor- tant benefits of a "nonfinancial" nature, such as inducing institutional changes, demonstrating new support mechanisms, and creating partnerships among relevant subjects of local and regional development. Discussion of Specific Statements and PoRlicy Recomnmendations The authors argue that a combination of initial conditions, factors endowments, and most important, national policies seems to determine country-specific steady states or growth paths to which individual countries converge. Therefore, they do admit the importance of the state's role. The EU is much more than a free trade area and is (or until recently was) moving in the direction of a federal state, despite the fact that this word had to be erased from the Maastncht treaty. Consequently, if one accepts that the EU is "half' state, then there is a role for its policies. I agree with the authors that there is a danger of creating a "dependency men- tality" in countries or regions receiving support from the Cohesion Fund (CF) and the Structural Funds (SFs). But I would challenge their support for deemphasizing small and medium enterprises (SMEs) and attracting foreign direct investment (FDI). Attracting FDI usually means creating another type of dependency-dependency on someone else's know-how and decisions. Support of SMEs might mobilize coun- tries' own initiatives. In the sphere of policy implications, the authors put forward a number of less radical proposals along with abolishing EU regional policy completely. I agree with several of them. I think everyone would agree with the need to simplify and debureaucratize EU cohesion policy, but in practice it is quite difficult to achieve consensus on such changes. For example, the increasing stress on evaluation of efficiency and effec- tiveness of EU cohesion policy as well as on promotion of public awareness of EU- funded projects is in fact increasing administrative burden. However, I can personally imagine, for example, that there could be only the Cohesion Fund and SOME COMMENTS ON THE BOLDRIN AND CANOVA STUDY 123 the number of SFs could be reduced to one, instead of four of them, each with dif- ferent regulations. Or at least, if there are just two objectives of the EU cohesion policy beginning with the next budget cycle in 2007, the number of SFs could be cut to two. Similarly, one can agree with larger territorial concentration of regional policy to the neediest regions (Objective 1), even though it is quite likely that some of the future members (Slovenia, Czech Republic) would soon completely or partially lose eligibility for Objective 1. Several criticisms raised by the authors might be mitigated by an increase of vol- ume of the Cohesion Fund and a corresponding cut of SFs, as the procedures for the CF are more straightforward; place more stress on large-scale projects with measur- able effects, and place responsibility on central governments, which usually offer a belter guarantee of efficient use of resources than do weak and in some countries newly established regional bodies. (This was already partially done for new mem- bers for the period 2004-06.) 1 also agree with the authors that there is a danger of over-expectation that struc- tural policies will solve the problems and lead to growth and convergence. At the veiy best, they can contribute to such goals by eliminating some deficits, demon- strating new approaches, and inducing some institutional changes, but each state must find its own way from a low-road to a high-road competitive strategy. Recent Trends of Regional Development in the Czech Republic ainid Outlook for the Future Orn the basis of an extensive analytical effort (for example, Bachtler, Downes, and Gorzelak 2000, Blazek 1999, Hampl 1999; Illner 2001), it can be concluded that the basis of regional development during the transition is vertical geographical position (that is, a qualitative hierarchy of regional centers). The role of vertical geographic position of cities and regions can be expected to continue as a dominant factor of regional development after the accession. While a qualitative hierarchy of regional centers will most likely continue to be the basis of regional development in the future, a second significant factor can be identified: more profound differentiation on microregional and local levels (Blazek 1999). This will be a consequence of the formation of new spatial forms of regional development such as development axes, clusters, or even "nonspatial" networks. The creation of these new forms of spatial organization of production will depend especially on the local initiative of both the private and public sectors but also will be influenced by differences in the external conditions of particular regions and localities Therefore, the vanation in local initiative-in combination with different starting conditions-will operate as a multiplier, stimulating more profound differentiation on the microregional level Embryonic versions of these new spatial forms can be 124 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH identified in the Czech Republic. The spatial agglomeration of some of the suppliers for Skoda Auto in and near Mlada Boleslav is an example of such a cluster. An exam- ple of a nonspatial network might be the association of eight Czech historical cities called "Czech Inspiration," which cooperate in the sphere of culture and attract tourism. The significance of these new forms will increase in the future, especially within the framework of the integration process, by the widespread use of modem communication technologies and by people's growing geographic mobility, fa- cilitated by the construction of new infrastructure. All these changes will enhance the opportunities for cooperation. The ability of local subjects to form or join these new organizational structures will significantly influence their future devel- opment. While development axes or clusters will be based locally, they will be more and more connected to (or even integrated into) international structures (Blazek 2002). From the policy perspective, it should be stressed that a vital asset in the forma- tion of these new spatial forms will be the quality of human resources, including the ability to cooperate and the formation of an atmosphere of commitment, profession- alism, and optimism (see Hirschman 1958; Krugman 1991). These assets are in pnn- ciple nonmobile and are almost exclusively dependent on bottom-up approaches (see also Malmberg 1997). Thus new space opens up for local and regional development strategies, which until now have been one-sidedly oriented on the construction of technical infrastructure in the Czech Republic. Policy Implications The main driving force in the sphere of Czech regional policy is preparation for the EU policy of economic and social cohesion (Blazek and Boeckhout 2000). The fol- lowing paragraphs provide a critical assessment of the current state of the art in the key spheres of preparation for EU cohesion policy. In the sphere of regional policy, the Czech Republic has recently made several important steps. In the institutional sphere, the coordination and implementation structures were strengthened, for example, by establishing 14 self-goveming regions, forming management and monitoring structures on the level of cohesion regions (NUTS 2). The first generation of programming documents has been drafted and approved or acknowledged by the government, a new legislative framework for regional policy has been developed (Regional Development Act No. 248/2000), and the volume of financial resources has increased due to state-support regional devel- opment programs for the two most affected regions (Northwest Bohemia and Ostravsko NUTS 2 regions). Moreover, these new regional programs are more inte- grated than most other support programs. However, despite these positive changes, Czech regional policy is still highly fragmented mto an array of small programs and still departs significantly from the EU cohesion policy. Table 6-1 compares the main differences between the Czech and the EU regional policies. SOME COMMENTS ON THE BOLDRIN AND CANOVA STUDY 125 TABLE 6-1 THE MAIN DIFFERENCES BETWEEN THE CZECH REGIONAL POLICY AND THE EU COHESION POLICY Sphere Czech regional policy EU cohesion policy Remarks on Czech policy Programming Czech Republic until Already the third gen- Excessive emphasis on recently without program- eration of program- analytical part, weak ming documents, now "over- ming documents strategic part, no consid- programming" (two sets of eration of altematives programming documents- one for Czech regional pol- icy, the other for EU cohesion policy), standard programs, low invention, top-down motivation for drafting pro- grammmg documents Implementa- Prevailing sectoral approach Different systems tlion structure Integrity of Narrow concept of regional Integrated multi- Progress recently, espe- approach policy and insufficient coor- sectoral approach cially "regionalization" dination with other policies of sectoral policies and implementation of more integrated state support programs for affected regions Incentives of Limited spectrum of Wide spectrum of Regional Development regional policy incentives incentives Act is consistent with the EU pnnciples Size of projects Prevailing small projects Prevailing large projects Selection of Problems with transparency Clear separation of projects management, moni- toring, and control functions Evaluation of Weak tradition, performed Systematic attention Chance posed by prepa- efficiency and infrequently and ad hoc and pressure for fur- ration of the monitoring effiectiveness ther enhancement system for Structural Funds Partnership Weak tradition, especially in Different practice the case of projects on supramunicipal level Involvement Low participation of pnvate Strong role, often of private sector in preparation for and significant initiative sector limited awareness about cohesion policy (continued) 126 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH TABLE 6-1 THE MAIN DIFFERENCES BETWEEN THE CZECH REGIONAL POLICY AND THE EU COHESION POLICY (continued) Sphere Czech regional policy EU cohesion policy Remarks on Czech policy Public Huge instability (14 new Different systems Senous disadvantage administration regions, planned dissolution given large expected role of districts and creation of of regions, towns, and smaller distncts in 2003, municipalities large honzontal fragmenta- tion of local govemment and unprecedented instability of their financing systems) Volume of Small but increasing Many times higher financial resources Source Bia2ek 2002 The Future The forthcoming accession has several immediate implications for reonentation of the Czech regional policy. Specifically, by the time of accession into the EU the very relevance of the existing Czech regional policy would be in question. The primary task of national regional policy should be to eliminate the leverage effect of support from the SFs in the form of matching grants, as subjects from poorer regions would not be able to provide sufficient financial resources for cofinancing of eligible proj- ects. Therefore, the Czech regional policy might provide, for example, an additional 15 percent cofinancing of projects implemented in the neediest regions so that local subjects would be able to obtain support from SFs. The reorientation of national regional policy toward the EU cohesion policy would also require a change in its time horizon, from the current annual programs to a multiyear approach. The Czech Republic is also missing an evaluation culture to guarantee effective and efficient use of public resources, not only in the sphere of regional policy but also in the public sector in general. In addition to these mostly technical changes of national regional policy, conceptual questions must be addressed. One of the big challenges facing future EU members is a gradual switch from a low-road to a high-road strategy of competitiveness (see also Porter 1999). The cur- rent advantage of low costs does not offer a sound basis for catching up with the West. Therefore, for example, the current emphasis of state policy for internal investors should be refocused from traditional investment incentives, first to after- care programs aimed at maximizing the positive effects of existing foreign invest- ments and second to improving the structure of incoming FDIs toward the industrial SOME COMMENTS ON THE BOLDRIN AND CANOVA STUDY 127 branches with higher added value and more sophisticated production requiring high- quality human capital. The immediate goal should not necessarily be high-tech indus- tries, but medium-tech would be a good start (including service or customer software centers and audit or consultancy firms operating on a global scale). Second, from a regional point of view, it would be desirable to promote more even spatial distribu- tion of FDIs. Along with hard measures like the provision of adequate infrastructure, this promotion could take the form of soft measures, such as an application of the concepts of complex territorial marketing. These changes would help the Czech Republic to switch from a low-road to a high-road competitiveness strategy and thus facilitate its real integration into the European economy. Notes I. The model of social market economy "seeks to combine a system of economic organiza- tion based on market forces, freedom of opportunity and enterprise with a commitrnent to the values of internal solidanty and mutual support which ensures open access for all members of sciciety to services of general benefit and protection" (European Commission 1996, p 13) W. Molle (1997) defines cohesion as "the degree to which disparities in social and economic welfare between different regions or groups within the Community are politically and socially tolerable" (p 429) B,ibliography Bachtler, John, Ruth Downes, and Grzegorz Gorzelak, eds. 2000. Transition, Cohe- sion and Regional Policy in Central and Eastern Europe London: Ashgate. Blazek, Jiri. 1999. "Regional Development and Regional Policy in Central East European Countries in the Perspective of EU Enlargement." In M. Hampl, ed., Geography of Societal Transformation in the Czech Republic Prague: Charles University, Faculty of Science. - . 2002. "Regional Impacts of the Czech Accession into the EU: An Attempt of Qualitative Analysis." In M Hampl, ed., Regional Development Specifics of the Czech Transformation, European Integration and a General Theory (in Czech). Prague: Charles University, Faculty of Science. Blazek, Jiri, and Sjaak Boeckhout. 2000. "Regional Policy in the Czech Republic and the EU Accession." In John Bachtler, Ruth Downes, and Grzegorz Gorzelak, eds Transition, Cohesion and Regional Policy in Central and Eastern Europe London: Ashgate. 128 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH European Commission. 1996. "The First Report on Economic and Social Cohesion." Brussels. Hampl, M., ed. 1999. Geography of Societal Transformation in the Czech Republic. Prague: Charles University, Faculty of Science. Hirschman, Albert. 1958. The Strategy of Economic Development New Haven, Conn.: Yale University Press. Illner, M. 2001. "Czech Regions Facing European Integration." In Grzegorz Gorze- lak, E. Ehlrich, L. Faltan, and M. Illner, eds., Central Europe in Transition, Towards EUMembership. Warsaw: Regional Studies Association, Polish Section, Wydawnictwo Naukove. Krugman, Paul. 1991. Geography and Trade. Cambridge, Mass.: MIT Press. Malmberg, Anders. 1997. "Industrial Geography. Location and Learning." Progress in Human Geography 21(4):573-82. Molle, Willem. 1997. The Economics of European Integration: Theory, Practice, Policy, 3rd ed. Aldershot, U.K.: Ashgate. Porter, Michael. 1999. The Competitive Advantage of Nations New York: The Free Press. CHAPTER 7 Rtegional Policy Experience in Southern Italy Fabrizio Barca W e should stop discussmg European Union (EU) regional policy or Structural Funds, arid, as a matter of fact, national regional policies, as separate from state intervention at large. Let us consider two very different types of state services: the provision of public goods, like safety, roads, communications, and training (type A services); and state transfers to finrs and households (type B services). If we raised hands around this table, there would probably be a great deal of support for type A services, though we might have differences of opinion about how much should be provided. There would proba- bly be little support for type B services. But, whatever our views are, I find it extremely difficult to take part in discussions where we talk about the regional policy of Europe or regional policy at the national level as separate from type A and type B services. Whenever a state provides those services, whether the services are labeled "regional" or not, it is influencing and changing the attractiveness of the sites where those services are provided. So regional policy is inherent in any public action. Most regional policy is implemented by state departments or regional depart- ments or municipalities, and the tariff levels of municipalities also influence the attrac- tiveness of metropolitan towns. That is regional policy. But it is not labeled "regional." So then, what is "regional policy" about? It is about putting a label on public action that has territorial effects. Regional policyt is about targeting some regions (backward regions, regions with undercapacity, or politically sensitive regions) and devising and implementing policies aimed at those regions. 129 130 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH This is also the case with EU regional policy. What then is the specific purpose of EU regional policy? There is first one very clear purpose, a political one. It is the same purpose that explams why, in a different field- that of short-term countercyclical policy-the U S. federal government compensates for differences in cycle among U.S. states. Federal funds in the United States represent one of the major sources of adjustment of the economic cycle, together with labor movement and with rents coming from portfolio holdings in other states; federal funds perform the role of increasing the political support for the federal government. In the area of struc- tural adjustment, EU Structural Funds-however small their amount-serve primarily the same political purpose: to increase consensus regarding the EU, in Slovenia as well as Hungary's eastern areas, m southern Italy, in the hills of France, and so on. But there is also another reason for regional policy to be enacted at the EU level. This is the part of the story that interests me most. I will not refer to type B service policies-to policies delivering subsidies and incentives. I will refer only to the provision of public goods to areas with under- capacity, and I will argue that EU Structural Funds can improve the effectiveness of type A service policies-training, infrastructure, roads, water, electricity, or the mamtenance of archaeological sites. Does Brussels add something to countries providing public goods to territories that are lagging behind and show clear underutilization of resources? This is the ques- tion. In order to provide an answer I will consider the Italian case. Figure 7-1 shows the growth of Mezzogiomo (which has about one-third of Italy's population and accounts for about one-quarter of Italy's gross domestic prod- uct [GDP]) since 1989 and the nsing contribution of Structural Funds. Structural Funds do not seem to have helped growth much. But at some point, apparently independently of policies, something happened in the southern economy; this change is easier to see on figures 7-2 and 7-3. Halfway through the 1990s, Mezzogiorno was still an area with 13.5 percent of employment in the agricultural sector. In 2002 the share is only 8.6 percent. Around 1995 the employment in the service and industry sectors started picking up, together with a very strong rise of net intiation rate of firmns and a continuous rise in export and tourism The shift (which is described in Barca 2001a and 200 lb) seems to be linked to a change in policy, or more appropriately to the end of a bad policy. In 1992-93, thanks to an agreement with Brussels, Cassa del Mezzogiorno, a centralized state agency managing both subsidies and infrastructures, was brought to an end, while transfers to state-owned corporations in the south were interrupted drastically. To the southem people, these events indicated that their incomes might drop. Suddenly, their saving propensity increased, their propensity to work in the public sector dropped, and their propensity to become self-employed and start new enterprises jumped. So the problem for policy was very clear. The first goal was not to make mistakes: not to bnng back poor incentives for the youth of Mezzogiomo. But was "doing noth- ing" the answer? Unfortunately, the south has an extremely low provision of public goods. The infrastructure is terrible. The quality of training is appalling. Security and REGIONAL POLICY EXPERIENCE IN SOUTHERN ITALY 131 FIGURE 7-1 GDP GROWTH AND STRUCTURAL FUNDS CDP (percentage changes) Structural Funds as (percentage of GDP) 3.0 - 20 0 2.5 - _ - ---.3- ---------- - 15.0 2.0 -- - -- .- 5 -1.5 - - - 12.3 ______ - - -- - 100 1.5 - 10 0 - A eae D go t --- 9---- ----- -- -- -------- GD\rwh1920 5.0 05 _- -- 0.0 - 0.0 -015 ---____ - - _--- _- - 11) - 19-9-0 - - -- -5 -------0---- - - - -1.5 ~~~~~~~~~~~~~~~-10 0 -1.5 19901991 1992 1993 1994 1995 1996 1997 1998 19'99 20'00 2001 2002 Structural funds as a percentage of GDP - GDP growvth - - - - Average GDP growth 1989-94 ---Average GOP growvth 1995-2002 Souirce Italian Ministry of Economy and [stat FIGURE 7-2 NONFARM EMPLOYMENT (1995 = 1 00, seasonally adjusted quarterly data) 114 - 113 ---------- ----- -------------------- 112 -_ -- ----_- ill ------------ -- -- ------ ..-- -----------__---- 110 -- --- -- 1 )9 ----- -------------_ -South --- --- 11)8 -----_---------------------- -- ----- - - ---- I 07 - - -------------- --------- ------ 10)6 -- --- ---------- -------------- 11)5 ----- ---- ---------- --enterNorth-- 11)4 -- 1 9)3 - 1994 19 1------ -7 19 1 2 11)2 - --___ --a--- ----- 1101 ----- - ----- --- -- ---- --- 11)0 -- -- -- - - - -- - -- -- -- - 9 9 - --- --- --- -- _---- --- --- -- _- -- - --- --- 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Note Unitofmeasure onthe y-axis is thousands of employees So ifnonfarm employment in 1995 is 100,000,110 equals I1 0,000 employees Source Istat 132 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH FIGURE 7-3 EMPLOYMENT (1995 100; seasonally adjusted quarterly data) 114 113 112 - 110 108 - _ Center North 107 106 - - -- - 105 ------ 104 \ South 103- 102 - - - 101 100 99 _ 98 I I h h h, I 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Note Unit ofmeasurc on the y-axis Is thousands of employees So if nonfanrm employment in 1995 is 100,000, I lequals I 10,000 employees Souirce Istat safety, for both citizens and finns, is inadequate Many extraordinary natural and cul- tural sites are inaccessible. This situation keeps profit expectations low. State provision of public goods is therefore indispensable for a breakthrough to take place. The question then was-and is-how to bring about a change, how to have the state coming back into the picture without making major mistakes and actually producing public services, public local goods, and not making transfers to firms, or restarting the old game, or leaving everybody feeling safer at home? Which level of government should provide these public goods? What information should be used to decide what to do? The information is obviously very local, especially in a country like Italy, which-it should be kept in mind-is not like Ireland. There are not many green fields. Both urban and rural areas are thickly inhabited, and many cultural, natural, and human resources are waiting to be used to their full capacity. The local agents know what to do, but they cannot do it. They do not have the institutions to do it, or a system through which they can evaluate whlch project should be done, which road should be built, how Pompeii should be turned into a great source of development. The way to approach the issue is rather straightforward. If the knowledge is local, then the local municipalities, not Rome, need to be given the means and the REGIONAL POLICY EXPERIENCE IN SOUTHERN ITALY 133 incentives to bring about the projects and the power to select the projects But there is a problem: regions are terribly inefficient The further and final question then becomes how to create the right incentives for the political leadership of regions to drastically modernize their public adminis- trations and to liberalize the markets for the provision of local public services (water, transportation, and so forth) so as to allow regions to properly select and implement lo^al projects? The idea came about to use as an incentive tough and credible rules set by an enlightened central administration, which happens to be the one I head at the Italian Ministry of Economy and Finance. We were going to give regions a list of 20 insti- tutional changes to enact before they would be given additional authority "If you- region-don't have employment services, e-government, or evaluation units, you can't evaluate projects. You need institutional reforms that change completely the working of public administration You can neither provide water nor privatize the water system if you don't have an administration that is able to write a tender offer or a programming document, to negotiate well with the private sector." The institutional changes were going to be turned into 20 targets to which 10 per- cent of all funds were going to be linked, to be distributed in 2004. Here is where Europe finally comes into the picture My administration would not have had the strength to set this mechanism in place, nor the credibility to carry it to the end-as we eventually did, with remarkable results, about which I will elaborate else- where-had we not had the rules sealed in a contract with Brussels. And that is true for many other institutional adjustments that I will not have space to mention here As Joseph Stiglitz would say, the EU regional policy gave us the credibility we lacked. And so it happened in many other states and regions all over our old (EU 15) Europe; and so it is happening now all over our new (EU25) Europe. The EU method, which results from the coordination and "institutional melting pot" of many European countnes, gave us and many other administrations not only cred- ibility but ideas and principles that are becoming part of a shared vision of economic policy. Many bureaucratic mistakes and misconceptions came, too, but we can address them-and indeed urgently remove them in the reform of EU regional policy-once we have made clear that EU regional policy is indeed an indispensable tool of our Union. Bibliography Barca, Fabrizio 2001 a. "The European Challenge." Italj Resilient and Vulnerable. Vol I (Spring/Summer). - -. 2001b. "Rethinking Partnership in Development Policies. Lessons from a European Policy Experiment." In Mark Drabenstott and Katharine H. Sheaff, eds., Exploring Policy Options for a New Rural America-A Conference Summary. Kansas City, Mo.. Federal Reserve Bank of Kansas City. CFIAPTER 8 Eiuropean Union Regional Aid atid Irish Economic Development Frank Barry Regional aid does not guarantee a real convergence of living standards in the recip- ient region. This is obvious from the experience of the Italian Mezzogiorno, to a lesser extent from the experience of Greece, and perhaps to some extent from the expenence of Spain. Regional aid is likely to be of greatest benefit when the other requirements for real convergence are satisfied. From this point of view the Irish and Portuguese expenences are of particular interest. These countries have followed very different industrial strategies Ireland's has been based substantially on a policy of attracting inward foreign direct invest- merit (FDI) in high-tech manufacturing sectors, while Portugal has expenenced sub- stantial real convergence on average European Union (EU) incomes with a manufacturing sector that remains dominated by indigenous low-tech industry (fig- ure 8-1). (Finland represents another interesting case of a geographically peripheral though not histoncally poor country that has prospered through indigenous high-tech industry.) I will have something to say later about why I think the Portuguese and Spanish experiences differ. Mostly, however, I will be concerned with the Insh experience The penod of substantially increased EU regional aid overlapped with the Irish boom, which saw unemployment fall from a high of 17 percent in the late 1980s to around 4 percent today, numbers at work increase spectacularly (by more than 50 per- cent); and average incomes rise from less than 65 percent of the EU average, a level around which they had hovered since 1960, to reach parity by the decade's end.' 135 136 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH FIGURE 8-1 THE CONVERGENCE EXPERIENCES OF GREECE, IRELAND, PORTUGAL, AND SPAIN EU average = 100 120 100 80 60 ; - <-c __- * * - - - - - - - [ - - *Ireland (GNP) 40 . - - --_______________ 1 - Spain --Greece 20 - - Pordugal 0 Sozurce Eurostat To some extent the boom can be seen as an episode of "delayed convergence," simply making up ground that had been lost during decades of poor economic man- agement. The failure of the 1950s resulted from eschewing free trade as the rest of Western Europe shed its protectionist policies. The underperformance of the 1960s and 1970s was a consequence of the failure to increase student numbers in line with the rest of Western Europe. The 1980s were lost as the country struggled with prob- lems bequeathed by a period of undisciplined fiscal policy What explains the timing and rapidity of the boom, however? Both were undoubtedly influenced by the country's strong FDI orientation. The amount of FDI from both within Europe and outside Europe exploded in the late 1980s because of the Single Market and the extent of investment funds made available by the long U.S boom Ireland's share of these flows also increased considerably. At the same time, in the late 1980s, the country's fiscal crisis, with which it had struggled for almost a decade, was finally resolved. And then of course there was the substantial increase in EU Structural Funds (SFs), following reform and reorganization in 1988; SF receipts per annum over the course of the 1990s were more than double the levels prevailing in the latter half of the 1980s.2 EUROPEAN UNION REGIONAL AID AND IRISH ECONOMIC DEVELOPMENT 137 Indeed, SF expenditures have been substantial throughout the EU periphery. Community support in the 1994-99 period accounted for almost 15 percent of total investment in Greece, 14 percent in Portugal, 1 0 percent in Ireland, and 6 percent in Spain. Analysis undertaken so far (by the present author and others) suggests, however, that the direct effects of these EU regional aid programs were modest. In the Irish case it is estimated that they contributed a maximum of about 0 5 percent per annum to the gross domestic product growth rate of the 1990s, whereas the boom saw Irish average real growth exceeding that of the EU 15 by around 6 percent per annum.3 Notice my use of the term direct effects By these I mean the increased demand associated with EU transfersplus the supply-side effects associated with an improved stock of human capital and physical infrastructure, evaluated on the assumption that the: response of Irnsh output is in line with estimates emerging from the international empirical literature. In this chapter I want to focus on something different-that is, on what we might call the indirect effects that can arise via interactions between SF expenditures and other concurrent developments in the economy. I begin by asking what the SFs were needed for and what they were spent on Then I discuss their interaction with three other developments that were taking place: the resolution of the fiscal crisis, the emergence of a greater degree of labor market flexibility, and the coming to fruition of the FDI-oriented development strategy. Finally, I will comment on the implica- tions for the candidate countries of Central and Eastern Europe. St]rategic Target Areas for SF Spending Unfavorable Initial Conditions in the Cohesion Economies Ireland had a number of characteristics that made it attractive as an export platform for foreign (primarily U S.) multinational companies. EU membership, of course, was crucial 4 The country also was an English-language environment, which may have been particularly important for U S. companies. Of major importance was the fact that Ireland has long had the lowest rate of corporation tax of any EU member stale Ireland shared with the other EU cohesion states a number of unfavorable char- acteristics, however. They included relatively low levels of human capital, relatively poor physical infrastructure, and a poor record in research and development (R&D) I now briefly discuss these unfavorable characteristics. Educational Attainment Table 8-1 provides some illustrative statistics in regard to the educational attainment of the population, showing that, overall, the cohesion countries remain behind the Organisation for Economic Co-operation and Development (OECD) average in this respect. 138 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH TABLE 8-1 EDUCATIONAL ATTAINMENT OF THE POPULATION AGED 25-64,1998 (Country percentages are expressed as a fraction of the OECD mean) Percentage that Percentage that Percentage that has attained at least has attained at least has attained at least Country upper secondary tertiary B (diploma level) tertiary A (degree level) Ireland/OECD 0 84 1 00 0 79 Greece/OECD 0 72 0 76 0 79 Spami/OECD 0 54 0 95 1 00 Portugal/OECD 0.33 0 43 0 50 ANote "At least tertiary B" includes "at least tertiary A Souirce OECD 2000 Physical Infrastructure and Peripherality The cohesion economies also lagged behind in terms of physical infrastructure. Indeed, the European Commission report One Market, One Money (1990) stated that firms in peripheral regions identified infrastructural deficiencies in the areas of education and training, transport and communications, and the supply and cost of energy as more important impediments to their development than geographical aspects of peripherality such as the proximity of suppliers and customers. The available data on the stock of infrastructure in peripheral regions provide supporting evidence. Table 8-2, adapted from Biehl (1986), reports relative infrastructural levels for an aggregate of transporta- tion, telecommunications, energy, and education, showing that the periphery countries had a substantial infrastructural deficit relative to the core EU countries in the mid- 1980s. Poor transportation infrastructure also affects economic distance from purchas- ing power, which is how peripherality is usually defined. Industrial Competitiveness R&D activity indicators provide one piece of evidence that can be used to track the level of development of firms and businesses in a region. One suspects that countries lower TABLE 8-2 RELATIVE INFRASTRUCTURAL LEVELS IN THE COHESION COUNTRIES AS A PROPORTION OF THE EU AVERAGE, 1985-86 Country Proportion Ireland 67 1 Greece 56 0 Spain 74 3 Portugal 38.7 Source Biehl 1986 EUROPEAN UNION REGIONAL AID AND IRISH ECONOMIC DEVELOPMENT 139 down in these rankings are also deficient in other areas of industrial and business devel- opment. Table 8-3 illustrates the position in regard to R&D over the course of the 1980s. These three indicators, therefore, serve to illustrate some of the structural weak- nesses of the cohesion economies before the emergence of the substantial SF programs. Allocation of SF Expenditures It comes as little surprise that these three areas of weakness were also the areas that attracted the bulk of SF expenditures, as shown in Table 8-4. The logic of the SF programs may therefore be seen to entail expenditures in areas in which there were strong microfoundations for public intervention and in which the cohesion countries were found to be deficient. Inidirect Effects of SFs: Interactions with Other Developments in the Economy Analysis of the impact of the SFs has found them to have quite moderate effects. I suggest, however, that their interaction with other concurrent developments in the Irish economy may have meant that they were particularly beneficial in that case. TAB-LE 8-3 BUSINESS ENTERPRISE EXPENDITURE ON R&D AS A PERCENTAGE OF DOMIESTIC PRODUCT OF INDUSTRY, RELATIVE TO THE EU AVERAGE Counitry 1981 1989 Ireland/EU 0 29 0 35 Greece/EU 0 07 0 06 Spain/EU 0 14 0 29 Portugal/EU 0 07 012 Source OECD 2001 TABLE 8-4 ALLOCATION OF STRUCTURAL FUNDS IN IRELAND Area Allocation of total SFs, 1994-99 (%) Physical infrastructure 36 3 Human resources 28 4 Production/investment aid to the private sector 25 8 Income support 9 5 Source Barry, Bradley, and Hannan 2001 140 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Fiscal Consolidation Successive Irish governments struggled throughout the 1 980s to overcome the debt cnsis that had resulted from inappropriate procyclical fiscal expansion at the end of the previous decade. The attempt to close the deficit via high taxation proved unsuc- cessful, due to the fact that it was by necessity procyclical (in a contractionary direc- tion), while workers responded to the tax hikes by raising wage demands. A new approach was tned in the late 1980s, when government expenditure, and in particular capital expenditure, was reined in as an alternative to increasing taxes still further. Barry and Devereux (1995) argue that this consolidation proved suc- cessful first because it was countercyclical, and second because it was supported by the development of the "social partnershup approach," which promoted wage mod- eration via the promise of future reductions in income taxes. The timing of the increase in SFs in 1989 was fortuitous in allowing the re- instatement of infrastructural projects that had been postponed as part of the necessary fiscal contraction. The infrastructural deficiencies that would otherwise have resulted would have made it more difficult to attract the levels of FDI achieved since then. The SFs would also have facilitated the social partnership agreements by relax- ing the government budget constraint, both directly (to the extent to which the prin- ciple of additionality can be sidestepped) and indirectly (through the tax revenues associated with the increased FDI inflows that subsequently emerged).5 Labor Market Developments The social partnership agreements begun in 1987 brought substantial competitive- ness gains and unprecedented industrial peace (Barry 2000). This brings me to an interesting point that has not been fully appreciated in the literature on the impact of the SFs to date.6 It concerns the effect of the labor market environment on the macroeconomic impacts of SF expenditures. To set the scene, consider the unemployment experience of three of the cohesion countries-Ireland, Portugal, and Spain-relative to the EU1 5 over the era from the Delors 1 Commu- nity Support Framework onward (figure 8-2). Portugal, we see, remains at low levels of unemployment throughout, Spain's unemployment remains high, and Ireland's tumbles from very high to very low. A stark charactenzation of these different situations then would be to see the Portuguese labor market as extremely flexible, with the country remaining close to full employ- ment throughout the period; Ireland's labor market might be characterized by wage ngidity, so that unemployment falls as labor demand expands, while Spain's might be characterized as an extreme insider-outsider model, where labor market insiders reap all the benefits of an increase in labor demand for themselves, so that it is all dis- sipated in the form of higher wages rather than greater employment. Let us consider the effects of SFs in this context. The impact of all these spend- ing programs can be categorized in terms of demand (short-run) and supply (longer- run) effects. The Keynesian short-run effects will be strongest where there is labor market slack (that is, Ireland and Spam rather than Portugal) and where demand can EUROPEAN UNION REGIONAL AID AND IRISH ECONOMIC DEVELOPMENT 141 FIGURE 8-2 IRELAND, PORTUGAL, AND SPAIN UNEMPLOYMENT RATES MINUS THAT OF THE EU15 Percent 15 10 - 5 - ~.....Spain-EU15 * Ireland-EU15 05 - -., ,, s _ ........ ------ Portugal-EU15 -5 - ,. . --10- 1987 1990 1993 1996 1999 Sour-ce Eurostat influence this slack (that is, Ireland rather than Spain). Thus, the short-run effects can be expected to be much larger in Ireland than in Spain or Portugal. As to the longer-run supply-side effects, let us think of them in terms of a model with three productive factors: labor, private capital, and public capital. If the increase in public capital is the same in each case, the impact on Spain will be less than on either Portugal or Ireland-in Ireland because an increase in public capital draws more labor as well as further private capital into employment, and in Portugal because with flexible labor markets there will be a higher level of employment and a higher stock of private capital relative to the initial stock of infrastructure, so the marginal product of public capital will be higher (Barry 2002b). In this scenario, the impact of SFs will be substantially higher in Portugal and Ireland than in Spain. Though this provides only a small part of the explanation, it is consistent with their respective convergence experiences over the penod EDI Inflows As mentioned earlier, Ireland's FDI-oriented strategy came to full fruition during the 1990s. It is unlikely that as much FDI could have been attracted to the economy had the extra SF-financed infrastructure not been in place I Besides the level of FDI inflows drawn to the economy, the SFs would also have affected the type ofFDI that Ireland was able to attract. Over recent decades, foreign industry in Ireland has become increasingly high-tech, as figure 8-3 illustrates, and this could have come about only in conjunction with the increasing human capital stock of the labor force.8 Consistent with this fact, the R&D orientation of both indigenous and foreign industry has been rising, as seen in table 8-6 The overall growth in the R&D onen- 142 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH FIGURE 8-3 DISTRIBUTION OF EMPLOYMENT IN FOREIGN-OWNED INDUSTRY IN IRELAND BY TECHNOLOGICAL LEVEL Percent 100- 80 60 40 20 0 | Low-tech E Medium-tech 5 High-tech Source Author's research tation of the economy, however, is primarily due to the operations of the foreign- owned sector, which carries out 64 percent of business-related R&D expenditures (BERD) In Irish industry. Progress in Terms of Strategic Target Areas I have argued that Ireland may have gained greater benefits from the availability of the SFs than the other coheston economies did because of how the effects of SF spending may have interacted with concurrent developments in the economy. Let us now see how this assumption is reflected in the various peripherality indicators employed earlier. Educational Attainment By the end of the 1990s, with the aid of the SFs, the relative position of the EU periphery had improved. In the case of educational attainment, this can be seen by focusing on the attainment of younger members of the population, as they would have been of student age at the time of the SF disbursement (table 8-5). Comparing the table 8-5 numbers with those for the whole population given in table 8-1, we see that in each case the periphery has converged on the OECD aver- age in terms of educational attainment.' To focus on Ireland in particular, we see that while the country continues to lag behind the OECD mean in terms ofthe population that has attained at least upper sec- EUIROPEAN UNION REGIONAL AID AND IRISH ECONOMIC DEVELOPMENT 143 TABLE 8-5 EDUCATIONAL ATTAINMENT OF THE POPULATION AGES 25-34,1998 (Country percentages are expressed as a fraction of the OECD mean) Percentage that Percentage that Percentage ihat has attained at least has attained at least has attained at least Country upper secondary tertiary B (diploma level) tertiary A (degree level) Ireland/OECD 0 93 1 16 1 00 Greece/OECD 0 92 0 88 0 94 Spain/OECD 0 74 1 28 1 31 Portugal/OECD 0 40 0 44 0 50 Note "At least terniary B" includes "at least tertiary A Source OECD 2000 ondary education, it has converged on the OECD average in terms of those attaining at least a university degree or equivalent, and has actually gone ahead in terms of those attaining a diploma or equivalent. This extra Irish throughput in tertiary education, furthermore, is largely concen- trated in the sciences. Thus, 1995 data from the United Nations Educational, Scien- tific, and Cultural Organization (UNESCO 1998) reveal that 40 percent of Irish tertiary graduates are in the fields of natural sciences, agriculture, and engineering, compared with an EU average of only 28 percent. Scientific degrees and diplomas are in strong demand within foreign-owned industry in [reland and, to this extent, Ireland's overall strategy can be seen to have influenced the setting of development prionties within the human capital domain. Physical Infrastructure and Peripherality While SF-supported improvements in transport infrastructure have been very sub- stantial, it is not clear whether they have allowed the periphery to converge along this dimnension. The European Commission itself suggests that "while investment in peripheral regtons has improved accessibility, it has been accompanied by similar investment in neighbouring regions and more central ones, which can counteract any relative gain" (2001, volume 1, p. 132). It goes on to caution that "the overall effect of such investment depends on what other measures are taken to stimulate economic activity in the regions concerned" (ibid.), which concurs with the point made above about the interaction with Ireland's FDI-onented industrial strategy Industrial Competitiveness I used R&D indicators earlier as an illustrative measure of industrial competitive- ness. Again we see, in table 8-6, that Ireland has continued to converge on core EU countries along this dimension, though the other penphery countnes have not Weak- ness in the R&D area is still pervasive within indigenous industry, however. 144 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH TABLE 8-6 BERD AS A PERCENTAGE OF DOMESTIC PRODUCT OF INDUSTRY, RELATIVE TO THE EU AVERAGE Country 1991 1997 Ireland/EU 0 50 0 87 Greece/EU 0.13 0 13 Spain/EU 0 38 0 33 Portugal/EU 0 13 0 13 Soturce OECD 2001 Where Do the Central and Eastern European Countries Stand in Terms of the Strategic Target Areas? The Central and Eastern European countnes (CEECs) lag behind the rest of Europe in many of the same areas that the Cohesion Funds were instituted to address. Some brief details on this are provided below. Educational Attainment While the Czech Republic and Hungary are both ahead of the OECD in terns of the proportion of the younger population with secondary education, these two countries as well as Poland lag well behind in terms of third-level degrees and diplomas (table 8-7). Some further comparisons are possible with the aid of 1995 data from UNESCO (1998) (table 8-8). They paint a slightly different picture from that emerging from the OECD. They show, for example, that the expected number of years of formnal school- ing for each EU country is higher than that for any CEEC, which is not what a com- parison of the OECD data for Portugal and Poland would suggest. The UNESCO TABLE 8-7 EDUCATIONAL ATTAINMENT OF THE POPULATION AGES 25-34, 1998 (Country percentages are expressed as a fraction of the OECD mean) Percentage that Percentage that Percentage that ha3 attained has attained has attaaied at least upper at least tertiary B at least tertiary A Country secondary (diplonia level) (degree level) Czech Republic/OECD 1 28 0 40 0 63 Hungary/OECD 1 07 0 56 0 88 Poland/OECD 0 86 0 48 0.75 Ireland/OECD 0 93 1 16 1 00 Note "At least tertiary B" includes "at least tertiary A Sozurce OECD 2000 EUFIOPEAN UNION REGIONAL AID AND IRISH ECONOMIC DEVELOPMENT 145 TABLE 8-8 EDUCATION INDICATORS, 1995 E.xpected years of Net enrollment Gross enrollment Country formal schooling ratio-secondary ratio-tertiary EUIO 15 2 92 47 8 Irelznd 13.6 85 37 0 Greece 13 8 84 38 1 Spain 15 5 94 46 1 Portugal 14 3 78 34 0 Czech Republic 13 1 88 20 8 Hungary 12 5 73 19 1 Poland 13 1 83 274 Estonia 12 5 77 38 1 Slovenia n a n a 31 9 Slovak Republic n a n a 20 2 Latvia 11 4 78 25 7 Lithuania n a n a 28 2 Rormania 11 4 73 18 3 Bulgrana 12 1 75 39 4 n a =.Not applicable Note Net enrollment ratio refers to the percentage of the population of the age group corresponding to that level of education enrolled, gross enrollment ratio refers to total enrollment divided by the population of the age group offi- cially corresponding to that level of education, EUI 0 refers to EU15 less Luxembourg and the Cohesion 4 (Greece, Irelatid, Portugal, and Spain) Source UNESCO 1998 source contains data for each of the CEEC economies, however, allowing one to see how they are ranked relative to each other. These data show that the CEECs are behind even the EU cohesion countries (with the possible exception of Portugal) in terms of educational attainment. Midelfart-Knarvik and others (2000) suggest that educational standards may be becoming tncreasingly important, finding: "The locatton of R&D-intensive indus- tries has become increasingly responsive to countries' endowments of researchers, with these industries moving into researcher-abundant locations The location of non- manual-labour intensive tndustries was and remains sensitive to the proportion of countries' labour forces with secondary and higher education" (p. 2). Phvsical Infrastructure and Peripherality Transport infrastructure plays an important role in calculations of centrality or "close- ness to purchasing power." Schiirmann and Talaat (2000) provide a recent ranking of EU countnes and CEECs in this regard. Their index is based on a measure of travel costs between points within the overall region weighted by the purchasing power that each point represents. 146 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH The most peripheral regions at present (according to figure 4.2 in their paper, p. 44) are the Baltic states, northern Sweden and Finland, and Bulgaria and Roma- nia Hungary, Slovenia, the Czech and Slovak Republics, and the southwest corner of Poland are no more peripheral than Ireland, Portugal, or Spain, and less peripheral than Greece Interestingly, these authors also present a projection for the year 2016 based on the assumption of EU accession (with its associated reduction in border delays) and the implementation of the huge Transportation Infrastructure Needs Assessment (TINA) transport infrastructure plans for Ccntral and Eastern Europe (TINA Secre- tariat 1999) (along with the Trans-European Energy Networks program for EU incumbents). In this scenario (which they plot in figure 4-14, p. 61, of their paper), some regions in Poland, the Czech Republic, the Slovak Republic, Hungary, eastern Germany, and Portugal move ahead of Ireland, with Greece left even further behind. Industrial Competitiveness The Czech Republic and the Slovak Republic lie between Ireland and Spain in the R&D rankings, while Hungary enters at a surprisingly low 20 percent of the EU aver- age (table 8-9). This indicates that there is still a substantial amount of ground to be made up along this dimension. Concluding Comments We see that the CEECs share many of the unfavorable charactenstics of the EU cohe- sion economies. These characteristics include relatively low levels of human capital and R&D, alongside economic peripherality. These are the kinds of disadvantages that EU regional aid sets out to redress. TABLE 8-9 BERD AS A PERCENTAGE OF DOMESTIC PRODUCT OF INDUSTRY, RELATIVE TO THE EU AVERAGE Couintry 1991 1997-99 Czech Republic/EU 1.06 0 67 Hungary/EU 0 31 0 20 Poland/EU - 0.27 Slovak Republic/EU 1 06 0 60 Ireland/EU 0 50 0 87 Greece/EU 013 0.13 Spain/EU 0.38 0 33 Portugal/EU 0 13 0.13 - = not available Source OECD 2001 EUROPEAN UNION REGIONAL AID AND IRISH ECONOMIC DEVELOPMENT 147 We have seen that the cohesion countries have converged in terms of some of these structural charactenstics at least. I have argued that a benign macroeconomic environment is also important, however, specifically in terms of labor market flexi- bility. The tripartite Social Partnership agreements instituted in 1987 in Ireland facil- itated industrial peace and a return to labor market equilibrium. Without some such steps to promote labor market equilibrium, it is doubtful that the boom of the 1 990s, and the macroeconomic effects of the SFs themselves, would have been as strong. I showed specifically that an insider-dominated labor market can reduce substantially the macroeconomic benefits of SF spending Labor market rigidities also hinder the possibilities of real convergence more directly (Barry and others 2000; Daveri and Tabellini 2000). The chapter also illustrated how the SFs interacted strongly and positively with Ireland's FDI-oriented strategy to generate very rapid convergence when the cir- cunistances were auspicious-that is, during the era of the Single Market and the sus- tained U.S. boom. Ireland adopted a low rate of corporation tax and fostered other characteristics favorable to the attraction of FDI With the aid of SFs, educational attainment improved considerably, facilitating the shift into high-tech (though largely foreign-owned) industry This in turn, along with targeted SF-funded aids to indus- try, raised the level of competitiveness of the Irish business sector The economy would have run into infrastructural constraints much sooner, which would have impinged on its ability to attract FDI, had the SF-funded infrastructural programs not been in place. Portugal's relatively successful convergence performance of recent years shows, however, that a reliance on FDI is not the only path to development for EU penphery economies Again, what appears to have been crucial in the Portuguese case, relative to Spain at least, is the degree of labor market flexibility that the economy exhibits. This point has been made in a number of papers, including Barry (forthcoming a), Bover, Garcia-Perez, and Portugal (2000), and Daven and Tabellini (2000). Thus, Por- tuguese convergence has been impressive, even though, consistent with its relatively low human-capital stock, the economy has specialized in low-tech production. This discussion suggests that there is no one route to economic development for the CEEC economies. Some, such as Hungary and Estonia, appear to be following the Irish development model, using low corporation-tax rates to attract export- oriented FDI. Others, such as the Czech Republic, though it has done well so far in attracting preaccession FDI, have not adopted the low corporation-tax strategy and may have a different development model in mind. The Portuguese experience sug- gests that this will not necessanly hinder real convergence possibilities (particularly since indigenous Czech industry appears to account for a substantial proportion of Czech BERD), though our simulations of a model of the Czech transition suggest that care must be taken to ensure labor market flexibility A further consequence of the SFs, emphasized elsewhere in this volume, is, as FitzGerald (1998) states, related to expenditure effectiveness evaluation: "The need 148 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH to satisfy the donor countries, through the EU Commission, that their money is well spent has resulted in the introduction of a set of evaluation procedures which has helped change the way the administration approaches public expenditure. In the past the only question, once money had been voted by parliament, was whether it had been spent in accordance with regulations. Now there is increasing interest in assessing how effective the expenditure has been." The introduction of more rigorous controls on the probity with which public funds are spent, as well as a more careful evaluation of the programs on which they are spent, may prove to be even more crucial in the case of the CEEC economies than it was in Ireland and the other cohesion economies. As to regional developments in Ireland, the strength of the boom ensured that all regions of the economy expanded, though the expansion in the Greater Dublin area was much more substantial than the expansion elsewhere. By the end of the 1990s this region exhibited higher wages, lower unemployment, and greater labor force par- ticipation than elsewhere in the economy-as well as greater congestion, of course. Once the economy reached full employment at the end of the 1990s, regional considerations came more to the fore in the deliberations of the industrial develop- ment agencies. Their recently updated Project Appraisal system places greater emphasis than heretofore on industrial dispersion across the regions (Barry, Walsh, and Murphy 2002). Notes I I use gross national product (GNP) rather than gross domestic product (GDP) in this cal- culation to exclude the profits of foreign companies located in Ireland. GDP per capita is sub- stantially higher. 2. 1 use the term Structural Funds rather loosely to include the Community Support Frame- work (CSF), the Cohesion Fund, and Community Initiatives. In the 1994-99 penod the CSF composed over 75 percent of the total allocated to Ireland. 3. As the Organisation for Economic Co-operation and Development (1999, footnote 32) points out, however, even this apparently modest effect nevertheless represents quite a respectable intemal rate of return, of 6 to 7 percent per annum, on the funds invested. 4. Before EU accession in 1973, Ireland's FDI inflows came from Europe rather than the United States, were market seeking rather than export onented, and were relatively low-tech, characteristics that descnbe most Central and Eastern Europe-bound FDI uiflows today (Barry forthcoming b) By the mid-1980s all these characteristics had been reversed 5 It would nevertheless be incorrect to conclude that the SFs generated the Irish boom through facihtating income tax reductions As mentioned earlier, corporation tax is the most important tax EUFIOPEAN UNION REGIONAL AID AND IRISH ECONOMIC DEVELOPMENT 149 relevant to the country's ability to attract FDI This tax has actually increased over time, from the zero rating on profits stemming from manufactunng exports that was introduced in the late l 950s 6 The literature I am refernng to includes econometnc work and calibration studies such as those ofthe Economic Social Research Institute (ESRI 1997) and Barry, Bradley, and Hannan (2001), which use the HERMIN model, the Commission's work using the Quest model (Vol- ume 2 ofthe Second Report on Economic and Social Cohesion, European Commission 2001), and various papers by Vitor and Pereira. 7 As it was, the boom created tremendous congestion This alone would have led to its ulti- mate dissipation (Barry 2002a, Dascher 2000) 8 The categonzation of high-, medium-, and low-tech industries is from OECD (1994) 9. SF expenditures affect these indicators directly Thus, apprenticeship programs affect the nunmbers attaining at least upper secondary education, diploma courses affect those attaining tertiary level B, and conversion courses are included in tertiary level A Bibliography The word processed describes informally reproduced works that may not be com- monly available through libraries Barry, Frank 1999 "Irish Growth in Historical and Theoretical Perspective." In Frank B3arry, ed., Understanding Ireland 's Economic Growth London: Macmillan - . 2000. "Convergence Is Not Automatic Lessons from Ireland for Central and Eastern Europe " World Economy 23(10).1379-94 - 2002a. "FDI, Infrastructure and the Welfare Effects of Labour Migration." jMfanchester School 70(3):364-79 -. 2002b. "Labour Market Structures and the Effects of EU Regional Aid." Work in progress. Processed. -. Forthcoming a "Economic Policy, Income Convergence and Structural Change in the EU Periphery " In Henryk Kierzkowski, ed., Europe and Globall- sation. London: Palgrave-Macmillan. - . Forthcoming b "EU Accession and Prospective FDI Flows to CEE Coun- Ines." In Robert Lipsey, ed., Real and Monetary Aspects of FDI in Industrial Countries Frankfurt: Deutsche Bundesbank/Springer Verlag. 150 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Barry, Frank, and Michael B. Devereux. 1995. "The Expansionary Fiscal Contraction Hypothesis: A Neo-Keynesian Analysis." Oxford Economic Papers 47 249-64. Barry, Frank, John Bradley, and Aoife Hannan. 2001. "The Single Market, The Structural Funds and Ireland's Recent Economic Growth." Journal of Common Market Studies 39(3):537-52. Barry, Frank, John Bradley, Michal Kejak, and David Vavra. 2000. "The Czech Eco- nomic Transition: Exploring Options Using a Macrosectoral Model " Available at http://www.ucd.ie/-economic/staff/barry/research.htmrl Barry, Frank, Brendan Walsh, and Anthony Murphy. 2002. "The Rationale for Sub- sidizing Jobs in a Fully Employed Economy." Irish Banking Review (Autumn): 38-50 Available at http.//www.ucd.ie/-economic/staff/barry/. Biehl, Dieter. 1986. The Contribution ofInfrastructu7re to Regional Development. No. 2 Luxembourg. Office for Official Publications of the European Communities. Bover, Olympia, Pilar Garcia-Perez, and Pedro Portugal 2000. "Labour Market Out- liers: Lessons from Portugal and Spain " Economic Policy 30:381-428. Dascher, Kristof. 2000. "Trade, FDI and Congestion The Small and Very Open Economy." CEPR Working Paper 2526. London: Centre for Economic Policy Research Available at http.//www.ucd ie/-economic/workingpapers/2000.htm. Daveri, Francesco, and Guido Tabellini 2000 "Unemployment, Growth and Taxa- tion in Industrial Countries." Economic Policy 30:47-104. ESRI (Economic Social Research Institute). 1997 The Single Market Review: Aggre- gate and Regional Impact-The Cases of Greece, Spain, Ireland and Portugal Lon- don: Kogan Page in association with the Commission of the European Communities. European Commission. 1990. One Market, One Money Luxembourg. . 2001. Second Report on Economic and Social Cohesion. Luxcmbourg. FitzGerald, J. 1998. "An Irish Perspective on the Structural Funds." European Plan- ning Studies 6(6):677-94. Midelfart-Knarvik, K. H., H. Overman, S. Reading, and A. Venables. 2000. The Location of European Industry Economic Papers 142. Luxembourg: European Commission EUROPEAN UNION REGIONAL AID AND IRISH ECONOMIC DEVELOPMENT 151 OECD (Organisation for Economic Co-operation and Development). 1994. Manu- fiacturing Performance A Scoreboard of Jndicators. Paris. 1999. OECD Economic Surveys Ireland Paris. 2000. Main Economic Indicators Paris 2001. Main Economic Indicators. Paris. Pereira, Alfredo, and Gaspara Vitor. 1999. "An Intertemporal Analysis of Develop- ment Policies in the EU." Journal of Policy Modeling 27(7) 799-822 Schinrmann, C, and A Talaat 2000. Towards a European Peripherality Index Report for European Commission DGXVI Regional Policy Available at http:// irpud raumplanung.uni-dortmund.de/irpud/pro/peri/ber53-gesamt pdf. TINA Secretariat 1999 Transport InfrastrulcturXe Needs Assessment (TINA) Identi- jfication of the Network Components for a Futuire Trans-European Transport Net- work in Bulgaria, Cypris, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia Final Report Vienna UNESCO (United Nations Educational, Scientific, and Cultural Organization). 1998 World Education Report New York Vitor, Gaspara, and Alfredo Pereira 1995. "The Impact of Financial Interpreta- Lion and Unilateral Public Transfers on Investment and Growth in EC Capital- Exporting Countries." Journal of Developnment Economics 48(1):43-66. CHIAPTER 9 Does Cohesion Policy Work? Some General Considerations and Evidence from Spain Angel de la Fuente Initroduction Over the last two decades, the European Union (EU) has adopted an active co- hesion policy aimed at reducing income disparities by subsidizing various types of investment programs in the Union's poorest regions through the so-called Structural Funds. This policy has often been questioned on at least two different grounds Perhaps the most common argument is that it has not worked. since most of the assisted regions continue to be relatively poor in spite of these pro- grams, EU grants are mostly a waste and should therefore be scrapped, or at least sexerely curtailed. The second objection, which is seldom explicitly stated but often lurks behind calls for cuts in structural programs, is based on the view that there is no reason why the EU should engage in redistribution across its con- stituent territories. The European Commission's view on this second issue seems to be that such redistribution is necessary because economic integration will tend to hurt the poorer regions of the EU by facilitating the concentration of economic activity in certain core areas. As has already been said in previous chapters, this prediction seems to be based on an implicit assumption-that there are sharply increasing returns to scale- for which there is very little empirical support. Hence, I do not think one can build a solid case for cohesion policies on the basis of the divergence predictions of the "new" growth and trade theories 153 154 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH But I do not think that assuming agglomeration is necessary either. In my view, the case for redistribution must necessarily be based on political and equity consid- erations that have to do with what a typical European citizen would consider fair and would be willing to support when it comes to EU budgetary policy. In this regard, I think we can validly extrapolate to the Community level the preferences of European electorates as manifested in the policies of national governments-provided we keep in mind that the typical taxpayer's willingness to pay for redistribution drops rapidly as his or her distancc from the beneficianes increases. While views about the desired level of redistribution vary widely across member countries, my impression is that there is fairly broad support in Europe for a moderate amount of budgetary solidar- ity. This consensus has been clearly visible in EU budget practices, which have con- sistently resulted in sizable net transfers to the poorer member countnes,' and has been incorporated into the EU's goverming treaties in the form of an explicit com- mitment to economic cohesion and financial solidarity among member states. Hence, I will take it as given that a certain amount of redistribution within the EU is desirable. Given this, it probably makes sense that at least some of this redis- tribution should be achieved through conditional investment grants with "addition- ality" requirements and some sort of quality filter to make sure that the funds flowing into the poorer territories are effectively used to promote the territories' development and are not diverted for consumption purposes. This leaves me with two questions, to which I will devote the bulk of this chap- ter. The first is whether we can reasonably expect that EU cofinancing of infrastruc- ture and training programs wll contribute to growth and convergence, and the second has to do with the level at which redistribution should be conducted. On the first issue, I am cautiously optimistic. I will argue that supply-onented regional policies can work in principle and have actually worked quite well in Spain, at least when judged in terms of their stated objectives. My argument will be based on a brief review of the empirical evidence on the growth effects of investment in infrastruc- ture and education, and on some estimates of the impact of the Structural Funds in Spain. On the second issue, I will argue that EU cohesion policy should be formu- lated at the national rather than at the regional level, essentially because member states already have adequate systems for internal redistribution. Can Regional Policy Work? There is considerable disagreement among both academics and policymakers con- cerning the effectiveness of the Structural Funds (and of regional policies in general) as instruments for the reduction of income disparities. Many critics of these programs argue that they cannot be very effective on the grounds that billion-dollar expendi- tures over two decades have not translated into clear progress in terms of regional convergence. An academic exposition of this view can be found in a recent paper by Boldrin and Canova (2001). These authors examine the evolution of the distribution DOES COHESION POLICY WORK? 155 of income across the EU regions over the past two decades and find no evidence that convergence is taking place or that recipients of EU transfers (with the exception of Ireland) have performed better than other regions Like the less formal versions of the same argument, however, their analysis has the serious shortcoming that it fails to control for any factors other than EU aid. A recent paper by Ederveen, Gorter, and Nahuis (2001) illustrates why results obi ained in this manner can be extremely misleading. These authors estimate a series of convergence equations relating growth in the European regions to initial income per capita and Structural Fund transfers. When no additional variables are included in ihe equation, the estimated coefficient on the transfers variable is negative and sig- nificant. When regional fixed effects are introduced, however, the coefficient of EU transfers becomes positive and significant. Upon reflection, these results should not be surprising Since the recipients of EU aid are by definition poor regions, the vol- um-e of aid works as a proxy for the omitted vanables that presumably explain why these regions have below-average incomes. The estimated coefficient on the volume of aid is negative because this is the only way the specification allows one to assign to these territories a low steady-state level of income. But as soon as other factors are controlled for, even by the simple expedient of introducing a set of regional dum- mies, the positive impact of aid on growth becomes apparent. A more sophisticated, although indirect, case against regional policies can be fotnd in the regional convergence literature (see Barro and Sala I Martin 1991 and especially Sala I Martin 1996). While these authors find that the speed of regional convergence is very low in Europe and in other samples, they are also skeptical about government's ability to accelerate the process The main piece of evidence they offer to back up this conclusion is a remarkable empirical regularity: the apparent stabil- ity of the rate of convergence, which has been found to be close to 2 percent a year in a variety of samples According to Sala I Martin, the fact that convergence takes place at practically the same speed within groups of territories supposedly charac- terized by very different levels of redistributive effort implies that such policies can- not be very effective. This conclusion seems, however, much too hasty. Governments can certainly influence the rate at which regions accumulate various productive factors-particu- larly infrastructures and human capital. To the extent that these factors have an effect on productivity, and on the location of mobile private inputs, there will be room for supply-side policies to influence the dispersion of regional incomes and to promote or accelerate income convergence. From this point of view, the stability of the conver- ge nce coefficient across different samples may indicate that the level of redistributive eflbrt has been too low to have a noticeable effect on the evolution of income dispar- ities, and/or that the policies adopted in the past have not been very effective, but it cannot be taken as evidence that regional policy per se is necessarily ineffective. Since EU regional policy has essentially taken the form of conditional grants for the financing of training and infrastructure projects, the discussion about its effectiveness 156 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH should begin with an analysis of the contribution of these two types of investment expen- ditures to productivity growth. Although the issue is, as we will see, somewhat contro- versial, I believe that the existing evidence provides reasonable support for the view that expenditure on education can have a considerable effect on productivity growth, and that the same holds true for infrastructure investment, at least in regions where the endow- ment of this factor is relatively low. Academic economists have traditionally been inclined to consider educational expenditure a key component of national investment, with a substantial economic pay- off in terms of output growth, and have often assigned the accumulation of human cap- ital a central role in formal models, particularly in the recent literature on endogenous growth This optimism seemed to be confirmed by a first round of cross-country empirical studies of the determinants of growth, where a variety of educational indi- cators were consistently found to have the expected positive effect. A second round of such studies (using panel data techniques), however, produced rather disappoint- ing results and even led some researchers to question the link between education and productivity.2 In recent years, the evidence seemed to be accumulating that such neg- ative results were largely due to poor data and vanous econometric problems. The cur- rent thinking about this issue is probably well summanzed by Temple (2000), who, after surveying the relevant micro- and macroeconomic evidence, concludes that "the weight of the evidence points to significant productivity effects" of educational invest- ment. Some recent work by de la Fuente and Domenech (2002) helps support this con- clusion. They find, in particular, that the amount of measurement error in the educational data sets that have been used in most growth studies is very considerable and that this error mduces a large downward bias in the estimated coefficient of human capital in the aggregate production function. When this bias is corrected using an extension of the classical errors-rn-variables model, the results suggest that the con- tribution of educational investment to productivity growth is quite sizable. The degree of consensus on the productivity effects of infrastructure investment is probably much smaller. The issue has been the subject of a debate that is still ongo- ing in the literature. The available empirical evidence is problematic, and its inter- pretation is complicated by econometric problems that have not been fully solved. Early work on the subject, notably by Aschauer (1989), concluded that the elasticity of national or regional output with respect to public capital is large and very signifi- cant, and that the rate of return on public investment is exceedingly high. A number of more recent studies, however, have questioned these results on the basis of vari- ous econometric problems. Some of these studies find that the significance of public capital disappears when a specification in first differences is used or fixed effects are introduced to control for unobserved national or regional specificities; these studies conclude that the accumulation of public capital does not appreciably contnbute to productivity growth. Other recent papers, by contrast, confirm the significance of infrastructure indicators using cointegration or panel data techniques that should in principle take care of some of the main objections to Aschauer's results. Some of DOES COHESION POLICY WORK? 157 them (especially Femald 1999) also provide rather convilcing evidence that causa- tion runs from infrastructure investment to productivity growth, and not the other way around De la Fuente (2002b) surveys the available evidence and concludes that there are sufficient indications that public infrastructure investment contributes significantly to productivity growth, at least in countries or regions where a saturation point has not been reached. The returns to such investment are probably quite high when infra- structures are scarce and basic networks have not been completed, but fall sharply thereafter Hence, appropriate infrastructure provision is probably a basic ingredient for a successful (regional or national) development policy, even if it does not hold the key to rapid productivity growth in advanced countries where transportation and communications needs are already adequately served This conclusion is based in part on a comparison of results for the regions of Spain and the U.S. states. Public capital variables are almost always significant in panel data specifications for the Spanish regions and often insignificant in similar exercises conducted with U.S data. Oiie possible explanation for this difference is that, as Fernald (1999) notes, the data for the U.S states start in 1970-that is, at approximately the time when the inter- state highway system was completed-whereas the Spanish data refer to a sample where the stock of infrastructures is still clearly insufficient. Some Impact Estimates for Spain Even though the evidence on the subject is not as clear as one would like, on the whole, the literature that I have briefly surveyed in the previous section suggests that investment in education and infrastructure is an important source of productivity growth. It follows that a policy aimed at reducing regional disparities by supporting the accumulation of these factors in poor regions can work in principle In this section I will provide some estimates of the impact of regional policies on growth and convergence in Spain. These estimates are based on a simple supply- oriented model that has been estimated with regional panel data covering 30 years. The model has two basic ingredients The first is an aggregate production function that relates regional output to the level of employment, the stocks of productive fac- tors (infrastructures, other physical capital, and the educational attainment of the work force), and the level of technical efficiency The second is an employment equa- tion that describes the evolution of this variable as a function of changes in factor stocks and in wage rates, allowing in an ad hoc fashion for adjustment costs that gen- erate sluggish dynamics I will also use an investment function estimated with national data for a sample of Organisation for Economic Co-operation and Develop- ment (OECD) countries to try to approximate the response of private investment to the measures financed by the Structural Funds 4 Before tuning to the Structural Funds per se, I want to take a quick look at the evolution of Spanish infrastructure policy over the past four decades The model I 158 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH have sketched above can be used to estimate the contribution of infrastructure invest- ment to convergence in income per capita across the Spanish regions. Figure 9-1 summarizes the results of this calculation for each quinquennium between 1955 and 1995.5 It shows that Spanish infrastructure investmnent was not redistributive at all pnor to 1980. After this date, by contrast, the redistributive pattern is clear and the contribution to regional convergence becomes positive and sizable. Although the pol- icy shift actually starts a bit before Spain's accession to the EU (in 1986), there is lit- tle question that the Structural Funds have played a key role by channeling a large volume of infrastructure investment into lagging regions. In the remainder of this section I will use the same model to estimate the contri- bution of the last completed Community Support Framework (CSF) to the growth of output and employment in the poorer Spanish regions.6 The exercise is based on the assumption that investment projects that are cofinanced by the EU are no different from others of the same nature. This assumption may be a bit too optimistic because, by reducing marginal costs, EU subsidies may have made for somewhat laxer proj- ect selection standards than otherwise, but I am reasonably confident that it is not a bad approximation The calculations that follow attempt to quantify the contribution of all the pub- lic resources channeled through the CSF (including national cofmancing as well as EU grants) and of the induced change in private investment to growth in output and employment during the period 1994-2000. The calculation involves adding these flows of resources to observed 1993 factor stocks and using the estimated production FIGURE 9-1 BETA CONVERGENCE/DIVERGENCE IN RELATIVE INCOME PER CAPITA INDUCED BY INVESTMENT IN PRODUCTIVE INFRASTRUCTURES Percent 0.60 0.40 - 0.20 - 0.00 \ -0.20 - -040 - -0.60- 1955-60 1960-65 1965-70 1970-75 1975-80 1980-85 1985-90 1990-95 Souirce De la Fuente 2001 DOIES COHESION POLICY WORK? 159 an(I employment functions to calculate the resulting increase in the variables of inter- est over their observed values in the reference year The results should be interpreted wilh caution because (among many other things) they do not provide a valid response to the question of what would have happened if the Structural Funds had not existed. To answer this question, we would need to know how the Spanish administrations would have reacted to the loss of these funds. It is almost certain that they would have made up at least part of the loss using their own budgets, but it is hard to be more pre- cise. As a rough adjustment for this and for the fact that the CSF also includes national resources, I suggest multiplying my impact estimates by around one-half to estimate the true marginal contribution of EU cohesion policy Figures 9-2 and 9-3 show the cumulative impact of the CSF on the stocks of pro- ductive factors and on the levels of output and employment of the entire set of Objec- tive 1 regions (excluding Ceuta and Melilla) during the period 1994-2015. Figure 9-2 shows that the CSF can be seen as a large positive "shock" that, over a period of seven years, raises aggregate factor stocks significantly above their starting levels (up to 20 percent in the case of infrastructures). Once the CSF has been executed (and assum- ing there are no new interventions), the stocks of physical capital and infrastructures are allowed to gradually return to their onginal levels as CSF-financed investments depreciate. The impact on the stock of human capital, by contrast, remains constant FIGURE 9-2 CUMULATIVE IMPACT OF THE 1994-99 COMMUNITY SUPPORT FRAMEWORK ON FACTOR STOCKS OF THE OBJECTIVE 1 TERRITORY Percent 25 20 15 1 0 5 1994 1999 2004 2009 2014 | - Intrastructures Other capital ----- Years of tralning Souwce Author's calculations 160 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH FIGURE 9-3 CUMULATIVE IMPACT OF THE 1994-99 COMMUNITY SUPPORT FRAME- WORK ON OUTPUT AND EMPLOYMENT IN THE OBJECTIVE 1 TERRITORY Percent 8- 7 - 6 - -- - - - - - - 5- 4- 01 . . . . . . . , . . . . . . . . . . . 1994 1999 2004 2009 2014 Output ------ Employment Sozurce Author's calculations until the end of the working life of the beneficiaries of training programs, which, on average, will take place after the end of the penod covered in the figure. Figure 9-3 traces the impact of these shocks on the evolution of output and employment. As may be expected, the output effect has approximately the same pro- file as factor stocks and begins to decline as soon as the CSF has been completely executed. The time path of employment, on the other hand, is very different from that in figure 9-2. Since this variable adjusts sluggishly over time, netjob creation remains positive until about 15 years after the conclusion of the programming period. Figure 9-4 summarizes the cumulative impact of the CSF on the output and employment of each of the Objective I regions in 2000. The figure shows that the growth effects of the CSF vary significantly across territories, reflecting differences in both the volume of investment and its rate of return. For the Objective I regions as a whole, the CSF adds 6.9 percentage points to output and 3.4 points to employ- ment in 2000 When we take as our reference the entire country, the CSF's cumula- tive contributions to Spanish output and employment in the same year are 3.5 and 1.85 points, respectively. Figure 9-5 quantifies the CSF's contribution to convergence in income per capita between Objective I regions and the rest of the country. It shows a convergence ratio that measures the fraction of the original income gap that would have disappeared as a result of the execution of the CSF (if the population of the different regions had DOES COHESION POLICY WORK? 161 FIGURE 9-4 CUMULATIVE IMPACT OF THE COMMUNITY SUPPORT FRAMEWORK IN 2000 Percent 10 7- .5 '3- .2- I) Ga Mu As Cant CyL Cana An Ex C-M Va Obj. 1 Spain I * Output 0 Employment An = Andalucia, As = Astuna, Cana = Cananas, Cant = Cantabria, CGM = Ceuta y Melilla, CyL = Castilla y Leon, Ex = Extremadura, Ga = Galicia, Mu = Murcia, Obj I = Objective 1, Va = Valencia Souirce Author's research FIGURE 9-5 CONVERGENCE RATIOS INDUCED BY THE COMMUNITY SUPPORT FFIAMEWORK Percent 35 - 30) 25- 15 10 1 5 Ga Cana Cant CyL As Va Mu C-M An Ex Obj.1 An = Andalucia, As = Astuna, Cana = Canarias, Cant = Cantabria, C-M = Ceuta y Mehila, CyL = Castilla y Leon, Ex = Extremadura, Ga = Galicia, Mu = Murcia, Obj I = Objective I, Va = Valencia Sou(rce Author's research 162 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROwTH remained constant over the sample period and growth performance had been uniform across them except for the effects of the CSF). For the whole of the Objective 1 ter- ritory, this coefficient is a bit over 20 percent and reaches values above 30 percent for Canarias, Cantabria, and Galicia. ShouWd There B3e Cohesion aeross Countries or across IRegions? The estimates in the previous section suggest that structural policies have worked quite well in Spain. They have contributed significantly to the growth of the poorer regions and to the reduction of regional disparities. It must be recognized, how- ever, that focusing on lagging regions entails a sizable efficiency cost and may not be optimal from a national perspective. Figure 9-6 shows why. The estimated returns on public investment are much higher in some of the richest Spanish regions than in most of the territories that are eligible for assistance under Objective 1. It follows that the overall impact of EU aid would have been considerably higher (and Spain's convergence toward average EU income correspondingly faster) if efficiency considerations had been given greater weight in the allocation of these funds. I am not sure that shifting structural assistance toward the richer regions of the cohesion countries is necessarily optimal, as it would certainly entail some FIGURE 9-6 RELATIVE MARGINAL PRODUCT OF INFRASTRUCTURES IN THE SPANISH REGIONS, 1995 120- 100- 80- 60- 40 20- 0 m -20- -H -40- -60 Cana Va Ga Mu Cant CyL An As Ex C-M Ma Ba Cat PV Na Ar Ri An = Andalucia, Ar = Aragon; As = Astuna; Ba = Baleares Cana = Canarias, Cant = Cantabna, Cat = Cataluna, C-M = Ceuta y Mehila, CyL = Castilla y Leon, Ex = Extremadura, Ga = Gahcia, Ma = Madnd, Mu = Muncia, Na Navarra, PV = Pais Vasco, Ri = Rioja, Va = Valencia Source Author's research DOES COHESION POLICY WORK? 163 cost in the form of greater internal inequality in output per capita. On the other hand, this cost will be substantially mitigated by the standard mechanisms for per- sonal redistribution that operate within (but not across) countries. The social pro- tection and tax systems of European countries will redirect a significant part of aily income gains from more efficient investment policies toward the poorer seg- ments of the population. For Spain, I have estimated that a policy shift in this direction would generate a net welfare gain (see de la Fuente 2002c). This may not be the case elsewhere, but I would argue that member countries should cer- tainly be free to distribute EU development funds across regions as they see fit, aiter weighing the relevant costs and benefits. Or, to put it in a slightly different way, cohesion policy should be formulated at the national rather than at the regional level because member countries have adequate mechanisms for internal redistribution. N otes This chapter was prepared for a conference on income convergence and European regional policy organized by the Bertelsmann Foundation, the World Bank, and Fundaci6n CIDOB. I gratefully acknowledge financial support from the Spanish Ministry of Science and Tech- nology under grant SEC99-1189 and from Fundaci6n Caixa Galicia 1. See de la Fuente and Domenech (2001) for an analysis of the redistributive impact of the E1J budget. 2 Positive results are reported by, among others, Mankiw, Romer, and Weil (1992) and Barro and Lee (1994), while Islam (1995), Caselli, Esquivel, and Lefort (1996), and other authors report the loss of significance of schooling indicators in fixed-effect specifications Pritchett (1995) also reports negative results and argues that we should start taking them at face value 3 See de la Fuente (2002a) for the details of the model and its estimation. 4. This function is the one estimated in de la Fuente (1997). f would have much preferred to estimate an investment function for the Spanish regions, but some of the required data are not available 5. Figure 9-1 shows the partial convergence coefficient induced by infrastructure investment in each penod. This parameter measures the rate of beta convergence that would have been observed if all regions had experienced similar growth rates except for the contribution of infrastructure investment For further details on the parameter's meaning and construction, see de la Fuente (forthcoming) 6. This section is based on de la Fuente (2002a) 164 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH l3Bibliiography The word processed describes informally reproduced works that may not be com- monly available through libraries. Aschauer, David. 1989. "Is Public Expenditure Productive?" Journal of Monetary Economics 23:177-200. Barro, Robert, and Jong-Wha Lee. 1994. "Sources of Economic Growth." Carnegle- Rochester Conference Series on Public Policy 40:1-46. Barro, Robert, and Xavier Sala i Martin. 1991. "Convergence across States and Regions." Brookings Papers on Economic Activity 1:107-82. Boldrin, Michele, and Fabio Canova. 2001. "Inequality and Convergence in Europe's Regions. Reconsidering European Regional Policy." Economic Policy 32:205-45. Caselli, Francesco, Gerardo Esquivel, and Fernando Lefort. 1996. "Reopening the Convergence Debate: A New Look at Cross-Country Growth Empirics." Journal of Economic Growth 1(3):363-89. de la Fuente, Angel 1997. "Fiscal Policy and Growth in the OECD." CEPR Discus- sion Paper 1755. London: Centre for Economic Policy Research. . 2001. "Infraestructuras y politica regional." In T. Garcia-Mila, ed., Nuevas Fronteras de la Politica Econ6mica, 2001. Barcelona: Generalitat de Cataluna and Universidad Pompeu Fabra. .2002a "The Effect of Structural Fund Spending on the Spanish Regions: An Assessment of the 1994-99 Objective 1 CSF." Instituto de Analisis Econ6mico, Barcelona. Processed. . 2002b. "Infrastructures and Productivity: A Survey." Instituto de Analisis Econ6mico, Barcelona Processed. .2002c. "Is the Allocation of Public Capital across the Spanish Regions Too Redistributive?" CEPR Discussion Paper 3138. London: Centre for Economic Policy Research. -. Forthcoming. "Convergence Equations and Income Dynamics: The Sources of OECD Convergence, 1970-95." Economica DOES COHESION POLICY WORK? 165 de: la Fuente, Angel, and Rafael Domenech. 2001 "The Redistributive Effects of the EU Budget: An Analysis and a Proposal for Reform." Journal of Common Mar- ket Studies 39(2):307-30. - . 2002. "Human Capital in Growth Regressions: How Much Difference Does Data Quality Make? An Update and Further Results." CEPR Discussion Paper 3587. London Centre for Economic Policy Research. Ederveen, S, J Gorter, and R. Nahuis. 2001. "The Wealth of Regions The Impact of Structural Funds on Convergence in the EU." Netherlands Bureau for Eco- nomic Policy Analysis, The Hague Processed Ferrnald, John 1999 "Roads to Prosperity? Assessing the Link between Public Cap- ital and Productivity " American Economic Review 89(3) 619-38. Islam, Nazrul 1995 "Growth Empirics. A Panel Data Approach." Quarterly Jour- nal ofEconomics 110.1127-70 Mankiw, Gregory, David Romer, and David Weil. 1992. "A Contribution to the Empirics of Economic Growth." Quarterly Journal of Economics 7(2):407-37. Pritchett, Lant. 1995. "Where Has All the Education Gone9" Washington, D.C.. World Bank Processed. Sala l Martin, Xavier. 1996. "Regional Cohesion. Evidence and Theories of Regional Growth and Convergence." European Economic Review 40.1325-52. Temple, J 2000. "Growth Effects of Education and Social Capital in the OECD." Oxford University. Processed. CHAPTER 10 Do Structural Actions Contribute to Reduced Regional Disparities in the European Union? Antoni Castells and Marta Espasa First of all, it is necessary to discuss some relevant points concerning regional pol- icy in the European Union (EU) Currently, there exist two instruments of regional policy the Structural Funds and the Cohesion Fund. They both are called Structural Actions. Since the last reform, the Structural Funds have had three main objectives 1. Objective 1, the most important in quantitative terms, is the development and structural adjustment of the poorest regions in the EU-that is, the regions that have a gross domestic product (GDP) per capita below 75 per- cent of the EU average in purchasing power standards (PPS) terms This objective includes the former Objectives 1 and 6. 2. Objective 2 is focused on the economic and social restructuring of regions with structural deficiencies This objective absorbs the former Objectives 2 and Sb 3. Objective 3 aims for the adaptation and modernization of education systems, training, and employment This objective is the result of the integration of the former Objectives 3 and 4. These objectives are carried out through four Structural Funds: the European Regional Development Fund (ERDF), the European Social Fund, the European 167 168 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Agricultural Guidance and Guarantee Fund, and the Financial Instrument for Fish- eries Guidance. It could be useful to remark that Objective I applies to the four funds, not only to the ERDF. So it is important not to confuse EU regional aid and the ERDF, though this one is the most important instrument for achieving the reduction of regional disparities. As will be more deeply examined in a later chapter, the implementation of the Structural Funds is relatively complex because the programming is made by objec- tives, while the management and the execution take place by funds. The other instrument of the Structural Actions is the Cohesion Fund, which sup- ports transportation and infrastructure projects. In contrast to Structural Funds, the Cohesion Fund is not allocated to the regions but to the four states with a per capita gross national product below 90 percent of the EU average: Greece, Portugal, Spain, and (temporanly) Ireland. Tables 10-1 and 10-2 contain some key data on Structural Actions: • Structural Actions represent around one-third of the EU budget. o For 2000-2006, Structural Actions will amount to 213 billion euros (that is, 2.7 percent of the 1999 EU GDP) o The yearly average amount of Structural Actions has increased from 10 6 bil- lion European currency units (ECUs) during the period 1988-92 to 30.4 bil- lion euros dunng 2000-2006. It means an increase of 5.7 percent per annum. TABLE 10-1 KEY DATA ON REGIONAL AID IN THE EU GLOBAL AMOUNT AND EVOLUTION (FINANCIAL PERSPECTIVES) 1988-92 1993-99 2000-2006 (millions (millions (tilillonis Category ECU/euro) (%) ECU/euro) (%) ECU/euro) (%) Structural Actions 53,140 21 7 176,348 33.5 213,010 33 0 Structural Funds 53,140 21 7 162,248 30 6 195,010 30 2 Cohesion Fund - - 15,150 2 9 18,000 2 8 EU budget 244,838 100 0 529,885 100 0 646,190 100 0 - Not availabie Source Europcan Commission 2001 DC) STRUCTURAL ACTIONS CONTRIBUTE TO REDUCED REGIONAL DISPARITIES? 169 TABLE 10-2 KEY DATA ON REGIONAL AID IN THE EU. MAIN BENEFICIARY COUNTRIES, 2000-2006 Country Percent of total Percent of GDP' Spain 26 4 1 4 Gennany 14 0 02 Italy 13 9 04 Greece 117 30 Portugal 10 7 31 a Yearly average of GDP 1999 Source European Commission 2001 * Structural Funds represent around 90 percent of Structural Actions, while the Cohesion Fund represents the remaining 10 percent. The ERDF is the biggest of the four Structural Funds, accounting for 58 percent of all Struc- tural Funds. * The main beneficiary countries, in absolute terms, are Spain, which receives 26.4 percent of total Structural Actions; Germany, 14.0 percent; Italy, 13.9 percent; Greece, 11.7 percent; and Portugal, 10.7 percent. * Structural Actions are specially allocated to the poorest countries of the EU. Portugal receives an amount equivalent to 3 1 percent of its GDP; Greece, 3.0 percent; and Spain, 1.4 percent. Stbuctural Actions and Reduction of Regional Inequalities In this section we discuss the impact of Structural Actions on the reduction of regional disparities in the EU, focusing on two key issues': 1. Does the regional aid reduce regional disparities in the EU? 2. Does the regional aid produce a progressive redistribution of income among regions9 Table 10-3 shows the most and the least benefited regions as a result of the direct impact of the Structural Actions, considering the net fiscal balance produced by these actions. By net fiscal balance we mean the amount of Structural Actions allocated to 170 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH TABLE 10-3 REGIONAL REDISTRIBUTIVE EFFECTS OF STRUCTURAL ACTIONS (average for the period 1995-97) Index GDP-95 Amount Net fiscal Percent (PPS/,nhabitant, Region received balance of GDP EU= 100) Most benefited regions (millions of ECUs in PPS) Azores (Portugal) 902.3 893 1 43 3 50 Madeira (Portugal) 662 0 648 8 28 1 52 Aegean and Crete Islands (Greece) 675 5 624 2 5 3 68 Alentejo (Portugal) 289 0 268 1 5 2 57 Algarve (Portugal) 176 0 159 1 3 8 70 Basilicata (Italy) 394.4 267 3 3 6 69 Ceuta-Melilla (Spain) 53 9 52 5 3 5 65 Least benefited regions (millions of ECUs in PPS) Brussels (Belgium) 4 7 -248 4 -0 9 172 Luxembourg 14 2 -91 6 -0 8 168 Rhetrland-Pfalz (Germany) 46 1 -243.4 -0.4 99 Baden-Wurttemberg (Germany) 63 9 -782 9 -0 3 126 Stockholm (Sweden) 8 6 -122 2 -0 3 124 Nordrhemn-Westfalen (Germany) 227 2 -1,160 5 -0 3 113 Schleswig-Holstein (Gernany) 45 0 -162 1 -0 3 106 Note Net fiscal balance is the amount of Structural Actions allocated to the region minus the contribution of the region to the financing of the Structural Actions Sources Espasa 2000 and Eurostat 1998 the region minus the contribution of the region to the financing of the Structural Actions-that is, the fiscal flow produced directly by the Structural Actions in this region. What does table 10-3 show? • The most benefited regions are clearly poor in communitarian terms. • Excluding Azores and Madeira, which are very particular cases, the direct Impact of Structural Actions in these regions is between 3.5 percent and 5.3 percent of GDP. • The distribution among countries of the 15 most benefited regions is as fol- lows: Spain, 6 regions; Portugal, 4; Greece, 3; and Italy, 2. DO STRUCTURAL ACTIONS CONTRIBUTE TO REDUCED REGIONAL DISPARITIES? 171 * The least benefited regions are, in general, rich regions in communitarian terms, but there is wide variation: Brussels and Luxembourg have an index of about 170, but Rheinland-Pfalz is slightly below the EU average. * The direct impact of Structural Actions is around -0.3 to -0.4 percent of the regional GDP, except for Brussels and Luxembourg where it is close to -I percent. * German regions in particular contribute heavily to Structural Actions. four out of the first seven regions in the bottom half of table 10-3 are Gernan regions. Therefore, we can conclude that Structural Actions are allocated according to regional disparities among European regions. Table 10-4 shows the reduction in the interregional disparities m GDP per capita due directly to the Structural Actions. We have elaborated different indicators of inequalities, some of them very elementary. We observe that the ratio of the regional GDP per capita between the top 10 percent and the bottom 10 percent is around 2.46 ancl becomes 2 34 after the Structural Actions. So, it is a reduction of nearly 5 per- cent. The reduction of the coefficient of variation is 3.62 percent And finally, the Girii index in GDP per capita is 0 1979 before the Structural Actions and 0.1863 after. Therefore, the inequality is reduced by nearly 6 percent. Thus, we can conclude that Structural Actions have a redistribution effect of reducing regional inequality between 3.5 percent and 6 percent, according to the ineqjuality index we use. TABLE 10-4 REDUCTION IN THE INTERREGIONAL DISPARITIES IN GDP PER CAPITA DUE TO STRUCTURAL ACTIONS Category Initial GDP per capita Final GDP per capita Percent vanation Maximum-minimum ratio One region 36931 35778 -3 12 10 percent of the regions 2 4591 2 3382 -4 92 20 percent of the regions 2 0744 1 9837 -4 37 Coefficient of variation 0 2503 0 2412 -3 62 Gini index 0 1979 0 1863 -5 87 Note Final GDP =initiai GDPplus net fiscal balance Sources Espasa 2000 and Eurostat 1998 172 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Structural Actions, lincome Redistiribution, and Economnic Growth The negative relationship observed in table 10-4 between Structural Actions and GDP per capita should be more systematically tested. Can we show the exis- tence of a negative and significant relationship between the level of GDP per capita and the net fiscal balance produced by Structural Actions in the different regions? Table 10-5 answers this question. There is a negative and strongly significant relationship between both variables. The lower the GDP per capita, the higher the positive impact of Structural Actions. Row I of table 10-5 indicates that the elas- ticity of the ratio of net fiscal balance to GDP in relation to the GDP per capita is around -6.5 percent, which means that an increase of 10 percent in GDP per capita produces a reduction of 0.65 percent in the ratio of net fiscal balance to GDP. This is an important redistributive impact, considering that funds represent only 0.4 percent of the EU GDP. Comparing, for instance, with national funds, the equivalent coefficient would be around -35 percent to -40 percent, but they rep- resent 40 percent to 50 percent of GDP (that is, 6 times more impact but 100 times greater size). In the same sense, row 2 of table 10-5 indicates that the elasticity of final GDP in relation to initial GDP is 0.935. Thus means that an increase in GDP of 10 percent becomes an increase of only 9.35 percent because the remaining 0.65 percent is redistributed toward poor regions through Structural Actions. Finally, we are especially interested in the effect of Structural Actions on regional growth. Does regional aid produce a significant effect on growth of regional TABLE 10-5 RELATIONSHIP BETWEEN NET FISCAL BALANCE AND GDP PER CAPITA (n = 122) Dependent variable Constant Independent variable In Y R2 F In NFB 0 0029 -0 0655 0 1916 (0 836) (-5 448) (0 03647) 29 7 In YF 0 0029 0 9345 0 9804 039 3 (0 836) (77 713) (0 03647) 6, Note NFB = net fiscal balance due to Structural Actions (in the regression this variable is specific as I + NFB/GDP), YF = final GDP (YF = Y + NFB), where Y = regional GDP per capita (expressed as an index with respect to the aver- age and in PPS) Source Castells and Espasa 2002 DO STRUCTURAL ACTIONS CONTRIBUTE TO REDUCED REGIONAL DISPARITIES? 173 GDP9 A preliminary result was obtained when testing this relationship Row 2 of table 10-5 shows a strongly significant and positive effect of Structural Actions on regional growth The elasticity between both variables is around 0 02 However, the results change dramatically if we take account of a convergence process If we introduce the variable GDP per capita, then the significance and the coefficient of the variable Structural Actions are sharply reduced This reduction occurs because Structural Actions may have an effect on regional growth but may actually overlap a process of convergence/divergence that would have taken place even without the Strructural Actions The reverse is also true The intensity of the process of economic convergence is dramatically reduced if we introduce the factor Structural Actions (table 10-6) The coefficient of regression (a-convergence coefficient) is reduced, in absolute tenrs, from -o 0950 (row 1) to -O 0617 (row 3), and its significance is reduced from 99 per- cent to 95 percent The conclusion is that when studying economic convergence, it is particularly important to specify the explicative function, introducing the appro- pnate variables for controlling nonconsidered effects Finally, we would like to mention that different models have been used for simulating the macroeconomic effects of Structural Actions The European Com- mission has published some of the Tnost important results It is difficult to compare. these models because they use different methodological approaches and consider different channels of the impact of the Structural Actions on GDP Some of them mainly consider the demand effects, while others pay special attention to effects on productivity (table 10-7) TABLE 10-6 EFFECTS OF THE STRUCTURAL ACTIONS ON REGIONAL GROWTH, 1995-97 (n 122) Independent Independent variable variable Dependeni variable Con srant In GDP pc95 In SApe Rz F In (GDPpc98/GDPpc95) ( 6089 (-4 77 i) (0 060400) 22 767 In (DPp98/GP1p9) 0 1960 0 0190 0 1387 2 8 In (GDPpc9S/GDPpc9S) (20 780) (4 526) (0 060889) 20485 In (GDPpc98/GDPpc95) 0 7735 -O0617 00089 0157. 12338 (2 584) (-1 930) (I 328) (0060209) Note GDPpc = GOP per capita expressed in PPS, SApce = xpenditure per capita on Structural Actions Source Castelis and Espasa 2Z2 0z 0~~~~~~ OZ 3 0~~~~~0 cc S3 a, -'T 00 r0 I D LL Rt ^>oo I E- 85 0_3 0 z o O ' E -: -.' 0 3 _ o CO 0 198 THE ROLE OF EU REGIONAL AID IN ECONOMIC CONVERGENCE IN SLOVENIA 199 rnenting, monitoring, and supervising the measures eligible for financing from the Structural Funds. There are considerable regional disparities between the most developed city, Ljubljana (the capital) and most other portions of Slovenia, which have the statisti- cal characteristics of regions with low economic development. After accession to the EU, additional development problems are expected to arise, especially in the coun- try's border areas, which will also constitute a new external border of the EU. To address these problems, at least the less developed parts of the country will have to be eligible for measures under Objective I. When preparing the experts' materials for the preparation of the National Devel- opment Plan, Slovenia had to make certain estimates as to the financial outflows and inflows between the EU's and Slovenia's budget in the period after accession. Slove- nia prepared a study that discussed a number of possible scenarios, depending on the assumptions as to the number of new member states, the outcome of negotiations on EU accession, and the interpretation of the Berlin conclusions. In the preparation of the National Development Plan 2001-2006, Slovenia presented the variant accord- ing to which on the expenditure side there would be a transitional period for all pay- ments into the EU budget in the period 2003-2006 (except for the EU's own traditional resources), whereas on the revenue side, there would be a gross inflow amounting to 3 percent of Slovenia's GDP from the EU Structural and Cohesion Funds. Slovenia expects to be on an equal footing (allocation of funds per capita) with the EU member states with a similar level of economic development-such as Portugal and Greece-during their negotiations regarding financing in 2000-2006 Bibliography lIvMAD (Institute of Macroeconomic Analysis and Development). 2001. Slovenia in the New Decade Sustainabldity, Competitiveness, EU Membership, Strategy of Economic Development of Slovenia 2001-2006 Summary. Ljubljana. - . 2002 Analysis of Economic Developments in 2001 and Prospects for 2002 and 2003. Spring Report. Ljubljana. Nlinistry of Economy. 2001. Pre-Accession Economic Programme of the Republic of Slovenia. Ljubljana. Republic of Slovenia, Ministry of the Economy 2001. National Development Plan 2001-2006 Ljubljana CHAPTER 14 Managing European Union Funds in the Candidate Countries: Administrative /Organization and Resource Distribution 'Vitalis Nakrosis The European Union (EU) cohesion policy is the second largest public policy in the E_U in terms of its budgetary size. In the penod 2000-2006 its total budget will account for one-third of the EU budget (£213 billion). The main aims of the EU cohe- sion policy are to reduce regional disparities between the levels of development of the various regions and the backwardness of least-favored regions or islands, includ- ing rural areas In order to achieve these aims, financial support is allocated to the EU member states from four Structural Funds-the European Regional Development Fund (ERDF), the European Social Fund (ESF), the Guidance Section of the European Agricultural Guidance and Guarantee Fund (EAGGF), and the Financial Instrument ior Fisheries Guidance (FIFG)-as well as the Cohesion Fund. In the current programming period, 2000-2006, the Structural Funds are chan- neled to the EU member states through national programs and four community ini- tiatives (Interreg, Urban, Equal, and Leader). National programs vary according to three main objectives: 1. Exactly 70 percent of the funding goes to regions whose development is lag- ging behind the EU average (Objective 1). 2. Exactly 11.5 percent of the funding assists economic and social conversion in areas experiencing structural difficulties (Objective 2). 201 202 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH 3. Exactly 12.3 percent of the funding promotes the modernization of training systems and the creation of employment (Objective 3) outside the Objec- tive 1 regions. After their accession to the EU, the candidate countries will benefit from the Struc- tural and Cohesion Funds. For instance, in the period 2004-2006, Lithuania is likely to receive about E 1.5 billion, of which £579.52 million will come from the Cohesion Fund At present, the candidate countries are involved in the preparation of pro- gramming documents (National Development Plans [NDPs] or Single Programming Documents) to receive EU support in the period 2004-2006. In addition, since 2000 the candidate countries have had access to the Poland Hungary Aid for the Reconstruction of the Economy (PHARE) Social and Economic Cohesion Program (supporting productive environment, human resources, and small- scale business-related infrastructure), the Special Accession Program for Agriculture and Rural Development (SAPARD) (supporting agnculture, rural development, and fisheries), and Structural Instrument of Pre-Accession (ISPA) (supporting envi- ronment and transport). Their annual budgets in Lithuania are about E 14-16 million, 29 million, and 50-70 million, respectively. The EU's preaccession instruments were designed to promote economic development in the candidate countries as well as facilitate their accession to the EU cohesion policy. The pnmary purpose of this chapter is to explain--drawing on the experience of Lithuania-how the candidate countries prepare for the Structural and Cohesion Funds. Two main topics are considered: administrative and territorial organization, and resource distribution. The chapter will devote more attention to the former set of issues because of their prevalence in the agenda of Lithuania's govemment. This chapter is organized into two main sections. The first section discusses administrative and territorial organization. A framework for analyzing administra- tive, economic, and financial systems for the management of the Structural Funds is presented and applied to explain the dynamics and state of Lithuania's preparations to manage the Structural Funds. The second section discusses resource distribution. The main question addressed in that section is how to balance efficiency and equity in the distnbution of resources. Administrative and Territoriall ODrganization There is no uniform approach to the management of EU funds among the EU mem- ber states. Therefore, the EU candidate countries can develop their own systems, tak- mg into consideration several alternative approaches to the management of EU funds. Alternative systems for the management of EU support can be formulated on the basis of two main principles: 1. Integrated/unintegrated management of the Structural Funds, defining the extent to which the Structural Funds are managed through existing national MANAGING EU FUNDS: ORGANIZATION AND DISTRIBUTION 203 administrative, economic, and financial systems or systems set up specifi- cally for the Structural Funds 2. Centralized/decentralized management of the Structural Funds, defining the extent to which the management of the Structural Funds is centralized or decentralized in the territorial-administrative sense. Integrated/Unintegrated Management This principle can be defined according to three main criteria 1. Low or high administrative integration'-that is, the Structural Funds are managed by existing national administrative systems or by administrative systems set up specifically for the Structural Funds (for example, project selection is carried out by competent national institutions or committees or by institutions or committees set up specifically for the Structural Funds) 2 Low or high economic integration-that is, whether existing national devel- opment programs and projects are cofinanced by the Structural Funds or new programs and projects are developed in order to absorb the Structural Funds, and whether development programs and projects cofinanced from the Struc- tural Funds are coordinated with existing national strategic documents. 3. Low and high financial integration-that is, the extent to which the Structural Funds are integrated into the national financial system (budget, treasury, financial control, and so forth) For instance, integrated administrative systems were used by Austria, Germany, Portu- gal, and Spain, unintegrated systems by Belgium, Denmark, the Netherlands, Sweden, and the United Kingdom, and mixed systems involving elements of both other types ol systems by Finland, France, and Italy (Taylor, Batchler, and Rooney 2000). C'entralized/Decentralized Management This principle can be defined according to three main criteria 1. (De)centralization of programs-that is, programs (and projects) cofi- nanced from the Structural Funds can be centralized (for example, sectoral operational programs under the Community Support Framework or the Single Programming Document for the whole country with centralized pri- orities and measures) or decentralized (regional operational programs under the Community Support Framework, Single Programming Documents for different territonal-administrative units, or one Single Programming Doc- ument with decentralized pnonties and measures). 204 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH 2. (De)centralization of decisionmaking-for example, centralized or decen- tralized project selection by central, regional, or local institutions/committees. 3. Use of central or regional/local institutions for the implementation of the Structural Funds. The application of centralized or decentralized systems for the management of the Structural Funds depends significantly on the type of program (Objectives 1, 2, and 3). Also, it depends on the territorial-admimstrative unit benefiting from the Structural Funds (the whole country or a particular territonal-administrative unit). The EU member states whose territories received assistance under Objective I of the Struc- tural Funds (for example, Greece, Ireland, and Portugal during the 1994-99 pro- gramming period) used more centralized systems. Four altemative systems can be formulated on the basis of these principles and critena (table 14-1). It should be noted that these systems are extreme altematives, unlikely to be found in practice because existing systems usually involve elements of several dif- ferent systems. However, they provide a good framework for analyzing the prepara- tion of the candidate countries for the management of the Structural Funds. TABLE 14-1 ALTERNATIVE SYSTEMS FOR THE MANAGEMENT OF THE STRUCTURAL FUNDS Centralized Decentralized * Centralized programs * Decentralized programs * Centralized decisionmaking Decentralized decisionmaking * Central institutions * Regional and local institutions Integrated * High administrative Integrated-centralized Integrated-decentralized integration * High economic integration * High financial integration Unintegrated * Low administrative Unintegrated-centralized Unintegrated-decentralized integration * Low economic integration * Low financial integration MANAGING EU FUNDS. ORGANIZATION AND DISTRIBUTION 205 Every alternative system has strengths and weaknesses. For instance, centralized systems can be more absorptive, but decentralized systems can be better suited for local community development. Integrated systems can be better at coordinating national and EU investments, but unintegrated systems can have higher visibility among the public. However, it is assumed in this chapter that the appropriateness of various alter- native systems to the candidate countries is determined by the characteristics of national administrative, economic, and financial systems. For instance, it was con- cluded that the integrated-centralized system will be the most effective in absorbing the Structural Funds in the period 2004-2006 in Lithuania, which will benefit from C)bjective I of the Structural Funds as a single Nomenclature des Unites Terntonales Statistiques (NUTS) 2 region (see Nakrosis 2002). Since about 90 percent of invest- ment expenditure that may be eligible to cofinance the Structural Funds is managed by sectoral ministries at the central level, the introduction of the integrated-centralized system would entail less administrative reorganization and lower costs. On the basis of this analytical framework, the following section attempts to determine which sSystem will be used for the management of the EU funids in Lithua- nia after its accession to the EU. It is likely that Lithuania will use the integrated- centralized system for the management of the EU funds DYNAMICS OF LITHUANIA'S ACCESSION TO THE EU FUNDS TOWARD THE INTEGRATED-CENTRALIZED SYSTEM hiI the penod 1997-2000, Lithuania made a number ofpolicy and mstitutional decisions establishing a decentralized framework for the management of the EU funds. For instance, the preparation process was coordinated by the Ministry of Public Adminis- tration Reformns and Local Authorities (responsible for the territonal administration; merged with the Ministry of Interior in 2000). According to the regional development law adopted in 2000, the EU funds were to be channeled on a decentralized basis in 10 countnes (through regional development plans implementing the NDP). However, since 2001 Lithuania has changed the orientation of its preparations for the Structural Funds toward a more integrated-centralized system In February 2001 the Lithuanian government adopted a key decision, the NDP Concept Paper, involving two major policy and institutional decisions: 1 The NDP coordinating the EU preaccession assistance (PHARE, ISPA, and SAPARD) was integrated into the budget and the State Investment Program 2. The Ministry of Finance (coordinating the budget and the State Invest- ment Program) became a coordinating authority for the Structural Funds and the Cohesion Fund, replacing the Ministry of Interior (responsible for the territorial administration). The section below assesses the present situation of Lithuania with regard to the man- agement of the Structural Funds. 206 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH INTEGRATED/UNINTEGRATED SYSTEMS Low Administrative Integration The Ministry of Finance, which is a coordinating authority for the budget and the State Investment Program, is a competent authonty for the preparation of the Single Programming Document and will become a Man- aging Authority for its implementation. Three Paying Authorities will be set up to carry out functions defined in the Structural Fund regulations. One new Paying Authority for the ESF will be set up in the Ministry of Social Security and Labour, while the National Fund (set up under the Ministry of Finance for the decentralized management of the PHARE program) and the National Paying Agency (set up under the Ministry of Agnculture for the management of SAPARD) will be reorganized to carry out functions of the ERDF and EAGGF (Guidance Section) Paymg Authorities, respectively. Functions of intermediate and implementing bodies will be carried out by existing national institutions. However, two new implementing bodies (the ESF Agency and the ISPA Implementmg Agency) remain to be established for the management of the EU preaccession instruments (the PHARE Economic and Social Cohesion and ISPA). Low-Mediun Economic Integration. Many measures of the Single Program- ming Document have been designed on the basis of existing budgetary programs and the State Investment Program in order to cofinance their implementation from the Structural Funds. On the other hand, several new measures were proposed specifi- cally for the Single Programming Document. Their implementation should be cofi- nanced from the budget, the State Investment Program, or other sources. Some sectoral inputs into the Single Programming Document were prepared on the basis of long-term sectoral documents. Also, a loose strategic framework dunng the programming process is provided by the economic development strategy and other wider national strategic documents until 2015. The extent to which the Single Pro- gramming Document will be embedded into, and coherent with, a national strategic framework remains to be seen. Low-Medium Financial Integration. The existing budgetary framework is already suitable for the management of the Structural Funds in a flexible way. The EU funds are likely to be channeled through the treasury, but methods of their inte- gration into the budget are still under consideration. The existing intemal audit system (consisting of intemal audit units and an audit department inside the Ministry of Finance) will be used to ensure sound financial management of financial support. The EU funds will be cofinanced primanly from the budget and the State Investment Program. CENTRALIZED/DECENTRALIZED SYSTEMS Centralized Programs. Since support of the Structural Funds to Lithuania will not exceed E I billion during the period 2004-2006, it will be channeled to the Single Programming Document covering the whole country. The Ministry of Finance pro- posed a centralized structure for the Single Programming Document involving no MANAGING EU FUNDS: ORGANIZATION AND DISTRIBUTION 207 regional or local priorities and measures (even in the Ignalina target region, where tile negative socioeconomic consequences of decommissioning the Ignalina nuclear power plant need to be addressed). Centralized Decisionmakng. (De)centralization of decisionmaking was still under consideration at the time of this wnting, but centralized decisionmaking is likely to be used for the management of the EU funds. For instance, centralized proj- ect evaluation and selection procedures are likely to be followed during the imple- mentation of the Single Programming Document. Centralized Implementation. Central-level institutions (sectoral ministries and other public administration institutions) dominate the preparation of the Single Pro- gramming Document However, one county administration and the Association of Local Authorities are represented in the Single Programming Document working group, and socioeconomic partners are being consulted during the programming process. All intermediate (ministries of economy, social security and labour, transport, environment, and agriculture) and implementing bodies (the National Labour Exchange, the ESF Agency, the Small and Medium Enterprise Development Agency, the Transport Investment Directorate, the ISPA Implementation Agency, the Central Financial and Contracting Unit, the National Paying Agency) proposed in the Con- cept Paper concerning the administrative system for the management of the Struc- tural Fund assistance are central-level institutions. EIJ INFLUENCE ON PREPARATION FOR STRUCTURAL FUND MANAGEMENT The assessment above does not mention the dynamics of Lithuania's accession to the Siructural Funds or the impact of various actors involved in this process of prepara- tion A combination of external and internal factors can explain the dynamics of L ithuania's preparation for the Structural Funds and the introduction of an integrated- centralized approach. One of the most important internal factors was slow prepara- tions to implement the PHARE 2000 Economic and Social Cohesion and the abolishment of a coordinating authority for the Structural Funds, the Ministry of Public Administration Reforms and Local Authorities, in 2001. However, it is the impact of the EU that is examined in the following section. This emphasis will reveal the powerful role the European Commission played in the candidate countries durng the preaccession period. The EU exerted its influence on the process of preparation for the Structural Funds in the candidate countries through several instruments, outlined below 2 Models Provision of Legislative and Institutional Templates. In 2000 the European Commission provided an informal working document to the candidate countries. The document, which defined the main requirements for the adminis- trative capacity of the candidate countries, provided a guiding framework for the candidate countries during the process of preparation for the Structural Funds. These requirements are presented in box 14-1. Lithuania has already implemented 208 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH BOX 14-1 ADMINISTRATIVE CAPACITY REQUIREMENTS FOR STRUCTURAL FUNDS o Capacity to prepare adequate statistical data (gross national product/cap/pur- chasing power standards [PPS], unemployment rates) at NUTS 2 and 3 levels for the determination of eligible areas by the Commission; and data required for programming, monitoring, and evaluation • Clear ministerial responsibilities and responsibilities of other state bodies for Structural Funds and the Cohesion Fund, in particular for preparation of Objective 1 programs o Establishment of an interministerial coordination body and elaboration of coordination procedures o Designation of a Managing Authority for each program o Partnership: involvement of regional/local authorities, and socioeconomic and other partners o Existence of adequate budgetary procedures, including procedure for multi- annual commitments and cofinancing procedure o Requirement that member state can show that additionality has been respected o Designation of a payment agency and elaboration of payment procedures o Establishment of monitoring committees o Elaboration and appraisal of indicators for monitoring and evaluation o Capacity to perform independent (ex ante) evaluation of programs o Functioning financial control, independent of final beneficiaries o Compliance with other community policies (competition, state aids, public procurement, environment, equal opportunities) o Elaboration of procedures for the certification of expenses and for correcting irregularities o Independent auditing capacity o Capacity to prepare projects for the implementation of Structural Funds and the Cohesion Fund ("project pipeline") some requirements. Other requirements should be implemented before or after Lithuania joins the EU. Agenda Setting andMonitoring. The European Commission mfluenced the prepa- ration for the Structural Funds by setting the agenda of the candidate countries and monitoring their accession progress. The most important tools of agenda setting and monitoring are the European Commission's Regular Reports and the Accession Part- nerships. In its annual Regular Reports the European Commission identifies institu- tional or policy problems in the candidate countnes and recommends solutions. For instance, in its 2000 Regular Report the European Commission emphasized that "it is of the utmost importance that the necessary structures for coordinated pro- gramming, management, monitoring, evaluation and financial management and con- MANAGING EU FUNDS: ORGANIZATION AND DISTRIBUTION 209 trol of Structural Fund assistance are established at a central level, before a stand is taken on whether a further decentralization is feasible or advisable" (Commission of the European Communities 2000, p. 77). Two main factors affected the Commission's selection of a centralized model. iimited administrative capacities of the candidate countries on the regional and local level, and the short duration of the programming period (2004-2006). In order to implement this recommendation, the Lithuanian government prepared a number of policy and institutional decisions, including the NDP Concept Paper. Gatekeeping. Access to Negotiations. The EU can influence the preparation of the candidate countries for the Structural Funds by giving access at different stages in the accession process, in particular starting and concluding negotiations on Chap- ter 21 of the Structural Funds For instance, in February 2001, during technical con- sultations with the second-wave candidates, the European Commission requested information about the institutional setup of programming and implementing the Structural Funds in Lithuania More specifically, the European Commission (in its comments on the NDP Concept Paper) welcomed a clear distinction between Lithua- nia's own regional policy, put under the responsibility of the Ministry of Interior, and the programming and preparation for the Structural Funds, put under the authonty of the Ministry of Finance. TechnicalAssistance and Twinning Technical assistance and twinning projects financed by the PHARE program and other bilateral programs are important sources of information, advice, and technical assistance to the candidate countries. For instance, to prepare candidate countries for the management of the EU funds, the European Commission proposed a Special Preparatory Program (SPP) The most sig- nificant outputs of the PHARE SPP I in Lithuania included the preparation of a National Paying Agency for accreditation, the development of a training program on the Structural Funds, and the preparation of the NDP. A few Irish experts employed in the Lithuanian SPP I promoted a centralized method of managing the EU funds. Their advice was denved from the successful expe- nence of Ireland, which followed a centralized approach in the period 1994-99. Under the 1994-99 Community Support Framework, funding from the Structural Funds was allocated to eight sectoral and only one regional operational program in Ireland. Preaccession Assistance Instruments. From 2000, the candidate countries became eligible to receive assistance under the PHARE Social and Economic Cohe- sion component (for business and human resource development), ISPA (for envi- ronment and transport infrastructure), and SAPARD (for rural development) programs The European Commission set a framework for the programming of the PHARE Economic and Social Cohesion component (precursor of the SERDF and the ESF) by issuing guidance notes to the candidate countries According to the first guidance note, support of the PHARE Economic and Social Cohesion 2000 component was to be concentrated in a few target regions In Lithuania, three target regions (Klaipeda-Taurage, Utena, and Marijampole) were 210 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH identified to receive assistance from this component in business and human resource development. However, in the PHARE 2000 Review, the European Commission pro- posed to introduce more sectoral approaches similar to the Objective 1 approaches of the Structural Funds (European Commission 2001 a). Therefore, support of the PHARE 2001 in Lithuania was concentrated primarily on a sectoral basis in the busi- ness and human resource sectors. In addition to the PHARE 2001 programming exer- cise, this decision facilitated the reorientation of Lithuania's preparations for the Structural Funds toward a more integrated-centralized model. IResource Distribution Economic Development Policies in Candidate Countries: Plentiful Ends, Limited Means After the reestablishment of independence, Lithuania, together with other transitional economies, started to dismantle the system of public interventions in the economy and accorded higher pnonty to macroeconomic stability, liberalization, privatization, and other important issues of economic transition. A combination of overloaded agendas and fiscal deficits prevented the introduction of active economic develop- ment policies in the transition economies. The stabilization of the economy and the increasing pace of accession to the EU allowed the candidate countries to start introducing economic development policies. In the past three years there has been a proliferation of various development strate- gies and programs in Lithuania. One can distinguish between two sets of strategic documents that overlap to a certain extent. Since 1998 Lithuania has adopted national development programs and strategies for such sectors as industry, small and medium enterprises, exports, quality management, business innovation, the information soci- ety, tourism, and research and development There are about 80 strategic documents in Lithuania, most of them of national ongin. At the end of 2001 the govemrnment attempted to establish a hierarchy of strate- gic documents. The Government Resolution divided all strategic documents into long-tern planning documents (the state long-tenm development strategy, the spatial development plan, sectoral long-term strategies); midterrn intersectoral, sectoral, and honzontal planning documents (including the economic development strategy, the NDP); and short-tenm planning documents (strategic business plans). Also, the gov- ernment prepared the long-term economic development strategy until 2015. How- ever, because of loose interministerial or intergovernmental coordination it represents only a collection of sectoral development strategies with 19 sets of com- peting objectives rather than a coherent development strategy. A second set of development plans and strategies was created as a result of Lithuania's preparation to manage the EU preaccession support instruments. The Lithuanian government prepared three multiannual strategic documnents (the NDP for PHARE, the Rural Development Plan for SAPARD, and Transport and Environment MANAGING EU FUNDS. ORGANIZATION AND DISTRIBUTION 211 Strategies for ISPA). In order to launch annual programming of the PHARE Eco- nomic and Social Cohesion component, the Lithuanian government prepared three preliminary NDPs for 2000-2002, 2001-2003, and 2002-2004. Although the Struc- tural Funds represented an important learming process, their preparation revealed sev- eral important weaknesses: * Owing to their formal nature, the preliminary NDPs mainly served the pur- pose of unlocking annual assistance under the PHARE Economic and Social Cohesion component. * Preliminary NDPs were not linked to national processes of budgeting and investment planning in the candidate countries. * Preliminary NDPs in the candidate countries failed to meet some require- ments of the Structural Funds, in terms of both process (for example, involvement of socioeconomic partners) and content (for example, absence of ex ante evaluation, quantified targets, and indicators) (European Com- mission 2001b). Two sets of strategic documents illustrate that the Lithuanian government is running two parallel investment management processes, one for national investments and another for EU funds. Because of growing amounts of national expenditure to meet EU funds requirements, having two parallel processes is not sustainable. Although in 2001 the Lithuanian government decided to integrate the preparation of the NDP as well as the budget and the State Investment Program into a single framework, this decision has not been fully implemented Despite the proliferation of strategic planning documents, Lithuania's economic development policies are lacking well-defined support instruments, public interven- tions, and implementation structures through which the Structural Funds can be chan- neled. Support instruments and interventions, which can be supported by the Structural Funds and the Cohesion Fund, are particularly underdeveloped in the busi- ness sector. Industry- and business-directed expenditure incurred in 1999-2001 and eligible for the EU funds amounted to only 1.5 percent of all national eligible expen- diture (Brozaitis and Nakrosis 2002).3 There is a need to redirect business support from interventions ineligible under the Structural Funds (for example, tax support) to eligible interventions. The candidate countnes are in the process of designing state aid schemes to be cofinanced by the Structural Funds The EU preaccession instruments facilitated Lithuania's creation of new devel- opment instruments. Some new instruments were downloaded from the EU docu- ments (for example, several fields of intervention most appropriate to Lithuania from the SA1FARD regulation); other instruments resembled those of other EU member states. In addition, the preaccession instruments allowed the country to develop appropriate monitoring and evaluation systems (for example, the Rural Development 212 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH Plan for SAPARD was subject to ex ante evaluation; its midterm evaluation is being orgamzed). However, in contrast to the programs cofinanced by the EU, the remain- ing national programs are not subject to monitoring and evaluation procedures. STRATEGIC CONTENT OF NATIONAL PROGRAMMING DOCUMENTS- EFFICIENT, EQUITABLE RESOURCE DISTRIBUTION The Lithuanian government committed to presenting the Single Programming Doc- ument to the European Commission at the beginning of 2003. Its preparation is being coordinated by the Ministry of Finance through the Single Programming Document working group, which involves not only sectoral ministnes but also socioeconomic partners. Unlike previous NDPs, the Single Programming Document will be made operational in a program complementing and implemented with support of the Struc- tural Funds. An essential part of the programming documents is a development strat- egy defining development objectives, priorities, and measures. During the programming process, the candidate countries need to balance effi- ciency and equity in the distnbution of resources (Pires 2002). On the one hand, the candidate countries can concentrate support of the Structural Funds in "motor" regions and expanding sectors of the economy (for instance, the informnation technology sec- tor of the Lithuanian economy, which is concentrated in the capital region). This strat- egy would aim at reducing the development gap vis-A-vis the EU average On the other hand, the candidate countries can concentrate support from the Structural Funds in regions that are lagging behind and in contracting sectors of the economy (for example, Lithuanian rural areas, whose economic structure is dominated by agricultural activities). This strategy would aim at reducing internal regional devel- opment disparities as well as the backwardness of least-favored areas (see table 14-2). It is likely that the programming documents of most candidate countnes will pro- pose mixed development strategies biased toward equity and distributing benefits from the Structural Funds to various territorial entities, sectors of the economy, or socioeconomic groups. Although efficiency is favored by the Structural Funds' phi- losophy aiming at the reduction of development disparities between poor member states and the EU average, the philosophy's impact is offset by several factors favor- ing equity in the distribution of resources. First, in the preparation of programming documents, the candidate countries should take into account the priorities outlined in the European Commission's guide- TABLE 14-2 TERRITORIAL AND SECTORAL CONCENTRATION OF THE STRUCTURAL FUNDS. DISTRIBUTION OF RESOURCES Type of regions Expanding sectors of the economy Contracting sectors of the economy Motor regions Growth poles Mixed Regions lagging behind Mixed Disadvantaged areas MANAGING EU FUNDS- ORGANIZATION AND DISTRIBUTION 213 lines (Commission of the European Communities 1999). Some common policies and finds of the EU favor specific sectors of the economy (the Common Agricultural Pol- icy and the EAGGF, favoring farmers and processing plants, respectively) or terri- tonal areas (the Common Fisheries Policy and the FIFG, favoring zones of fisheries). Second, the partnership principle of the Structural Funds brings the development needs of different sectors (infrastructure, human resources, and productive sectors), authorities (central, regional, and local); and socioeconomic partners to the process c,f programming for the Structural Funds. The involvement of various minstitutions in the consultation process favors equity rather than efficiency in the distribution of resources The bias toward equity in the allocation of resources is likely to produce shop- ping lists of interventions eligible under the Structural Funds, rather than coherent clevelopment strategies. For instance, the European Commission criticized the Lithuanian programming documents (both preliminary NDPs and a first draft of the Single Programming Document) as lacking a strategic focus. There is a risk that insufficient attention to efficiency in the distnbution of resources may produce the phenomenon of "chasing the money" and limit the degree to which EU cohesion policy objectives can be achieved. Finally, the short programming period (2004-2006) and the limited experience of the candidate countries in managing the EU funds constrain their capacity to implement very wide and ambitious strategies. Therefore, the European Commission .suggested that the candidate countries design a clear set of priorities and measures for the period 2004-2006. Also, Lithuania is considering the introduction of various pilot projects and technical assistance measures to build necessary capacity for the next programming period RISK OF NEGATIVE DISTRIBUTIONAL EFFECTS AND MARKET DISTORTIONS A considerable amount of EU funds will be available to the candidate countnes after their accession to the EU. As mentioned above, in its draft financial proposal the European Commission estimated that total assistance to Lithuania in 2004-2006 will amount to E 1,485 billion, or about E500 million annually. The annual volume of EU support will be about twice as large as the total Lithuanian State Investment Program. However, in the preaccession period the candidate countries are preoccupied pri- marily with the following issues. * How to complete all preparations needed to receive EU assistance * How to draw down as much support from the EU funds as possible and man- age the funding without fraud, abuse, and irregularities. The effective and efficient management of EU support is not very high on the agenda of the candidate countries. Therefore, high volumes of EU support coupled with 214 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH insufficient attention paid to their effective and efficient management may produce negative distributional effects and market distortions during the implementation period. Negative distnbutional effects and market distortions of the Structural Funds can be minimized by designing public interventions aimed at creating positive externalities or public goods whose benefits are widely distributed (for example, public infrastruc- ture), addressing common business development needs (as opposed to development needs of individual enterpnses), reducing entry barriers, and minimizing the amount of imperfect information and uncertainty (Nakrosis and Vilpisauskas 2000). A good list of rationales for public interventions was proposed in the midterm evaluation of Ire- land's Community Support Framework: • Spending on public goods (for example, infrastructure) Corrective subsidies (interventions designed to alter relative prices so as to correct for general ongoing externalities [for example, grants and subsidies for job creation]) o Targeted subsidies designed to overcome specific externalities (such as information barners) or to alter behavior (for example, in-company training and research and development) o Spending with a distributional motivation (for example, social housing) (Honohan 1997). However, although the distortionary effects of the Structural Funds can be mini- mized, it is very difficult to avoid them altogether. CAPACITY TO COFINANCE THE STRUCTURAL FUNDS After the accession of the candidate countries to the EU, large sums of public or equivalent expenditure will be required to cofinance support of the Structural Funds and the Cohesion Fund. Since an average rate of cofinancing in the cohesion countries amounts to about two-thirds of the total eligible cost, Lithuania would need to allocate about C731.44 million of national cofinancing to absorb the EU funds in 2004-2006. Some authors argue that there seems to be a clear risk of the lack of national cofinancing potentially leading to sizable underuse of the preaccession funds available from the EU. However, recent analysis of Lithuania's capacity to absorb the EU funds found that the amount of state and municipal expenditure incurred in 1999-2001 and potentially eligible as cofinancing for the EU funds stands at -1,556 million, exceeding the minimal cofinancing requirement by about two times (Brozaltis and Nakrosis 2002). MANAGING EU FUNDS. ORGANIZATION AND DISTRIBUTION 215 However, the distribution of eligible expenditure according to sectors of the economy is not in line with the situation of the EU member states benefiting from the C'ohesion Fund. For instance, the cofinancing capacity in Lithuania is least sufficient in the sectors of industry and business as well as the environment (Brozaitis and Nakrosis 2002). These discrepancies can result in the need to redistribute public expenditure after EU membership. Two main types of redistribution are likely to occur 1. Redistribution among sectors of the economy ineligible (public adminis- tration, law and order, social security, defense, and so on) and eligible (the productive sector and services, human resource development, and socio- economic infrastructure) under the Structural Funds 2. Redistribution from ineligible to eligible expenditure inside eligible sectors of the economy (for example, from ineligible expenditure of maintaining existing roads to eligible expenditure of constructing new roads in the trans- port sector). T he redistribution of resources, although not socially or economically optimal, may be inevitable for the absorption of the EU funds. In order to absorb the EU funds, Por- tugal needed to shift public resources in its first years of EU membership from social areas to areas covered by the Structural Funds (Pires 2002). Finally, in order to match the EU funds with national expenditure, the candidate countries need to overcome at least two formidable problems. designing measures in the programming documents matching their existing expenditure and developing a sufficient "project pipeline." For instance, in order to draw down assistance from the Structural Funds programmed in the Single Programming Document, Lithuania will need to prepare about 1,000 projects. However, the country's capacity to prepare projects for the Structural Funds is insufficient (Commission of the European Com- munities 2000); the culture of project management in the Lithuanian public sector is still in its infancy. For instance, considerable support is allocated to finance various public administration institutions rather than separate economic development projects Conclusion In the period 2001-2002, the agenda of the candidate countnes was dominated pnmar- ily by the establishment of adequate administrative structures to manage the Structural Funds. The European Commission affected the preparation of the candidate countries for the management of the Structural Funds through several instruments discussed above. Lithuania will use the integrated-centralized system to manage the EU funds. It was argued that a centralized model promoted by the European Commission to all candidate countries, regardless of their territorial-administrative and other 216 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH characteristics, can have negative effects (for example, on the processes of decen- tralization) (Pires 2002). However, it should be stressed that the introduction of the integrated-centralized approach in Lithuania has been relatively successful so far. It significantly contributed to increasing the scope and speed of preparations for the Structural Funds. For instance, Lithuania was among the first few candidate coun- tries to close preliminary negotiations on Chapter 21. This impressive scope and speed were possible because the European Commis- sion's recommendations were quite similar to the existing characteristics of admin- istrative, economic, and financial systems in Lithuania. The integrated-centralized system of managing the Structural Funds was more appropriate than the alternatives because 90 percent of investment expenditure that may be used to cofinance the Structural Funds is managed by sectoral ministries at the central level. During a dis- cussion with me in October 2002, in explaining the recent progress of Lithuania on Chapter 21, one Lithuanian official concluded that the function of coordinating the EU funds had finally "found its home " However, resource distribution received less attention in the candidate countries in the preaccession period. Despite plentiful objectives of economic development policies, the candidate countries need to develop support instruments and public interventions through which the Structural Funds can be channeled. Also, the candidate countries need to balance efficiency and equity in the distri- bution of resources in their programming documents for the Structural Funds. Most candidate countries are likely to propose mixed development strategies biased toward equity rather than efficiency. Equity in the distribution of resources is favored by the fact that the candidate countries should take into consideration various national pri- orities as well as EU pnorities in the preparation of their programming documents. Notes 1. This criterion was borrowed from Taylor, Batchler, and Rooney (2000). According to the degree of administrative additionality, this article distinguished between "differentiated" and "subsumed" systems I have defined the other critena. 2 Instruments are drawn from Grabbe 2001. 3 To compare, in Portugal assistance to private businesses only under state aid schemes amounts to about 25 percent of all expenditure by the EU funds Bibliography Brozaitis, H., and Vitalis Nakrosis. 2002 Lithuania's Capacity to Absorb the EU Structural and Cohesion Funds Assistance. Summary Vilnius: National Regional Development Agency. MANAGING EU FUNDS: ORGANIZATION AND DISTRIBUTION 217 Commission of the European Communities. 1999. The Structural Funds and Their Co-ordination with the Cohesion Fund Guidelines for Programmes in the Period of 2000-2006. Working Paper. Luxembourg. - . 2000. Regular Report on Lithuania's Progress towards Accession. Luxem- bourg. European Commission. 2000. The Main Administrative Structures Required for Implementing the Acquis Luxembourg. - 2001 a. PHARE 2000 Review-Strengthening Preparations for Membership. Luxembourg. - . 2001b. "Preparations for the Structural Funds in the Candidate Countries." Synthesis paper presented at the Twinners Seminar, Brussels, March 15 and 16. Grabbe, H 2001. "How Does Europeanisation Affect CEE Governance? Condition- ality, Diffusion and Diversity " Journal of European Public Policy 8(4): 1013-31. F[onohan, P., ed. 1997. EU Structural Funds in Ireland: A Mid-Term Evaluation of the CSF, 1994-1999. ESRI Policy Research Series, No. 3, July Nakrosis, Vitalis, and R. Vilpisauskas. 2000. "Adapting to EU Transfers: The Case of Lithuania " Paper presented at the conference on EU Structural Support: Its Macroeconomic and Distributional Effects and Social Environment, Prague, November. Nakrosis, Vitalis. 2002. Assessment of the Effects of the EU Regional Policy on Lithuania's Public Administration (in Lithuanian). Vilnius: National Regional Development Agency. Plres, L. M. 2002. "Discussion Paper for the Session 'Managing the Increased EU Aid in the Central and Eastern European Countries."' Barcelona, October 11. Taylor, S., J. Batchler, and M L. Rooney. 2000. "Implementing the New Generation of Programmes: Project Development, Appraisal and Selection." IQ-Net Bulletin 7(September) :5. CHAPTER 15 Managing European Union Regional Aid in iCentral and Eastern European Countries: Do the Countries Need Development Aid? Jan Szomburg In the past 12 years of transformations, Central and Eastern European Countries (CEECs) have had to cope with the historically unprecedented culmination of four rmajor challenges: 1. Systemic transformation-a transition to utterly new market rules that required CEECs to modify their behaviors and philosophy as well as to acquire new skills 2. Far-reaching structural changes 3. Swift opening of economies and globalization 4. A technological revolution (information technologies). Confronted with those four challenges, CEECs proved their enormous adap- tive capacities and learming skills. However, this very intense adjustment period has been accompanied by high social costs. Unemployment in most CEECs is much higher than in the existing member states. A part of their human capital has become depreciated. Societies accustomed to the system of socialist egalitarian- ism now face increasing social and regional differentiation (World Bank 2002). Transformation-related fatigue, and disappointment with the market economy and 219 220 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH democracy, are deepening. They find expression in the relatively common convic- tion that in the old days (that is, in the times of socialism) people were better off, although-in objective terms-living standards have much improved in all CEECs since socialism ended. The years 2003-2006 appear very difficult for CEECs. They can expect the following: • Very demanding conditions of European Union (EU) membership • High adjustment costs to the Acquis Communitaire * Strong competition from abroad, ensuing from-among other things- support (subsidies) to investments in EU member states from national budgets and Structural Funds, which are many times higher than appro- priations for investment support in CEECs • High cost of a nominal convergence policy in preparation for the European Monetary Union-reduction of inflation, budget deficit, and public debt. Comnparison with Cohesion Countries The situation of CEECs on the eve of accession is frequently compared with the sit- uations of the so-called cohesion countnes-the current EU members Greece, Por- tugal, and Spain. Some have argued that CEECs are slightly better off than the cohesion countries were before accession and that the new enlargement will not be substan- tially different from previous ones (chapter 3) However, regardless of all econo- metric analyses, it is worthwhile to consider how certain aspects of the CEECs' history imght make their experience different. One aspect is related to human capi- tal, even more so to social capital, the sphere of mentality and culture as well as insti- tutions. The former German Democratic Republic provides the most telling example of the significance of that historical background. The fundamental difference between the CEECs and the triad Greece, Portugal, and Spain is that the former had a 50-year break in their market economies. Therefore, formal educational back- ground does not reflect the actual level of human capital expressed in terms of very practical skills and behaviors of entrepreneurs, managers, employees, bureaucrats, and households. The break in continuity of functioning under a market economy has also a very significant adverse impact on the functioning of various public institu- tions that greatly affect an economy's efficiency and investment-related attractive- ness, including courts, notary offices, and vanous bodies regulating and monitoring market operations (offices regulating the telecommunications market, securities and exchange commissions, and so forth). They are either utterly new institutions or old MANAGING EUROPEAN UNION REGIONAL AID 221 institutions fulfilling new functions. Quite naturally, those institutions learn from their own mistakes and try to build up their ethos and operating models. The process of reforms and maturing of such institutions is not yet complete, which adversely affects the certainty of business operations, protection of ownership rights, and abid- ance by contracts-so-called systemic competitiveness in general. Central and local government administrations are only learning to get by in the new market economy- they are not yet excelling. The distinct features and experiences of CEECs have also led to an absence of continuity in the sphere of ownership. Although a certain process of primary accu- nmulation of capital has been completed, it is not the solid backbone of national cap- italism that exists in the cohesion countnes. The histoncally unprecedented scale of sale to foreign investors of enterprises-including such entire key sectors as banking- is reflected in the subsidiarization of CEECs' economies, where major undertakings in given economic sectors become subsidiaries of foreign companies. As a result, there are adverse changes in the economies' functional structure: research and devel- 0pment,.marketing, strategic planning, and so on are frequently transferred to foreign seats of strategic investors, with only production-related functions remaining in the c;ountry. At the same time, the endogenous development of small and medium enterpnses (SMEs) encounters various obstacles, including capital weakness (absence of conti- nuity in capital accumulation) Financial markets are not mature enough to facilitate this development. Paradoxically, privatization of the banking sector with the partic- ipation of large strategic investors impaired the support provided to SMEs As an example, in Poland there is a clear gap between the financial market's supply and (lemand (loans or ownership capital) at the level of I to 5 million euro A thorough analysis of the differences between CEECs and cohesion countnes in the preaccession period is beyond the scope of this chapter There are many more distinct features, including the hentage of socialism in the form of costly public social services, whose share of gross domestic product (GDP) is much higher than in the cohesion countries in the preaccession period. One should also mention the fundamental fact that CEECs are poorer than'the cohesion countries. There is an obvious conclusion CEECs need development aid, although the aid itself cannot guarantee success. Adequate internal policies will need ito be in place to put that aid to good use. Four Perspectives on Structural Fund Use Structural Funds can be viewed from four different perspectives: 1. The absorption perspective 2. The impact perspective 222 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH 3. The institutional perspective 4. The macroeconomic perspective. From the absorption perspective, the major question concerns the utilization rate of Structural Funds appropnated to a given country. It takes little effort to find out how much of the appropnated money a given country or region used. This information is easily accessible and easily understood by the general public. Therefore, this perspec- tive is of utmost importance for politicians and bureaucrats If appropnated funds are not used, the leaders have to explain why. From this perspective, it is most important to develop the largest possible supply of projects and focus on big, sinple projects that can safely ensure use of the allocated resources. If the absorption perspective dominates, less importance is attached to programs ensuring the coherence of interventions and their possibly large impact upon social and economic cohesion (long-term supply effects). From the perspective of Structural Funds' impact upon an economy, the ques- tion of whether all appropriated resources have been used is less important. The major issue becomes the impact of Structural Funds on economic competitiveness and productivity and/or on bridging regional development gaps-that is, medium- and long-term supply and structural effects. Cohesion and relevance of entire pro- grams count much more than individual projects. Because the effects of Structural Funds with regard to their impact upon social and economic cohesion are not as eas- ily evaluated as the rate of funds utilization, and hence they are less comprehensible for the public at large and also outlast the political cycle, there is much less pressure on politicians and bureaucrats to adopt this perspective. The institutional perspective has two aspects. One concerns the direct impact of institutional arrangements upon allocation of control over Structural Funds (the deci- sionmaking process). This is a very sensitive issue for politicians and bureaucrats since it directly defines the scope of their influence and the strength of their position A more or less open struggle concerning the form to be assumed by Structural Funds' management system is raging between ministries or between central and regional lev- els, and so forth. Solutions reached in this regard are difficult for the public at large to assess. The other aspect of this perspective concerns the impact of Structural Funds management on a country's institutional order (administrative system) and the shape that order will assume in the future. Structural Funds can be, for instance, a stimulus for the desirable decentralization of a country or for the undesirable disintegration of its administrative system This kind of partly indirect influence is quite significant for a country's long- term administrative efficiency and systemic competitiveness. However, there are not too many incentives for politicians to take account of this critenon. The meaning of their efforts in this respect is not easily discerned, and the effects are naturally delayed. Moreover, those effects can be incompatible with the requirements of swift and complete absorption. M4ANAGING EUROPEAN UNION REGIONAL AID 223 Given the entirety of the background conditions and all four perspectives for viewing Structural Funds, one can identify two extreme approaches. 1. The approach targeting short-tenn absorption 2. The approach targeting development. The first approach takes absorption efficiency-that is, a given country's ability to con- sume as much of the appropriated funds as possible in a given period (for example, 2004-2006)-as a dominant criterion for the organization and management of Struc- tural Funds. Attention is focused on ensuring the largest possible number of technically tnature projects. This approach naturally favors large infrastructural investments-for example, motorways-which do not necessarily yield the best possible outcome from the point of view of a country's development m a given penod (Chapter 2) Needs of local and regional development and needs of endogenous potential are rather poorly addressed under this approach. With a view to forgoing old mmistenal structures and old procedures for spending public funds, new structures and new procedures are estab- lished alongside the old ones. Thus, fast decislonmaking paths are created that are free from the "redundant" burden of the operations of legislative bodies. the parliament and regional councils as well as the central administration This "shortcut" approach can indeed bnng about the best short-term absorption outcome-for instance, in the years 2004-2006-but in a later penod, say 2007-2012, it can prove a trap, diminishing the capacities for full absorption at the level of 4 percent GDP, particularly in such areas as human resources, SMEs, or regional innovative systems. The approach targeting development from the very start is a long-term one. It is founded upon an awareness of the macroeconomic costs of using external assistance and appreciation of the fundamental role of institutional settings. It focuses not so rnuch on individual projects as on programs aiming at ensuring their cohesion. It treats the challenges of Structural Funds as an opportunity to reform administration, clecentralize it, and improve general procedures for spending public money, while avolding fragmentary organization and management. It better addresses the needs of endogenous development and "soft" investments. Paradoxically, in the long run this approach yields better absorption capacities than the first approach, not to mention a larger positive influence upon social and economic cohesion The macroeconomic perspective is usually completely ignored in considerations regarding Structural Funds However, a mass inflow of Structural Funds can decrease rnacroeconomic stability if not well used. It can stimulate increased external imbal- ance through excessive increase of internal demand and foreign exchange rates. Bibliography World Bank. 2002 Transition. The First Ten Years. Washington, D.C. CHAPTER 16 TIanaging Regional Aid in Latvia Janis Krumins Regional policy in Latvia, like regional policy in many other candidate states, is a new approach to solving the country's socioeconomic problems It demands the breaking of many stereotypes that represent traditional politicoeconomic thinking; it also demands the creation of a totally new scale of values Important groundwork has already been done in Latvia * The National Development Plan and the National Temtonal Plan have been elaborated. * The Law on Regional Development has been adopted. * Five planning regions and the institutional structure for the management of the European Union (EU) resources are currently being established Hlowever, some decisions made to date have not been good because of the existing stereotypes. Unlike most Latvians, I believe that regional policy should be oriented toward the development of lagging and depressed territories, and that the planning and implementation of the regional policy should take place at a decentralized level. Existing statistics confirm that this is the best approach Intensification of significant development differences between Riga and Ventspils-the capital and the biggest transit port in Latvia-and the rest of the state 225 226 EUROPEAN INTEGRATION, REGIONAL POLICY, AND GROWTH territory can be observed in Latvia. In 1996, incomes of the rural population were approximately 93 percent those of the urban population, but by 2000, this figure had dropped to 73 percent. This means that the growth of the national economy can be seen in only the biggest cities, while the situation in the rural areas declines more and more. A smgle working person in Ventspils pays seven tumes the income tax paid by all the inhabitants of the average small rural district. Therefore, it is obvious that Latvia is seeing the development of "economically empty" territories. Investors do not find them attractive, no new workplaces can develop in them, and their residents cannot create a satisfactory tax base to maintain or develop the temtories' infrastructure. It is no wonder then that in Latgale-the eastern region of Latvia, which shares a border with Russia-only 9 percent of residents want their children to stay in their native town or village after they graduate from school. The rest see no prospects for their offspring there. For that reason, I am convmced that the EU Structural Funds should be guided territorially, rather than in a sectoral manner, through the ministries. The financial aid needs to reach the planning regions; there should be as little mediation and bureaucracy as possible. I am a defender of the decentralized approach More than ever, regional institu- tions need to be involved in the planning and implementation of regional develop- ment policy; they need to be entrusted with higher responsibilities. The situation can be managed best at the regional level, where decisions about the support for existing projects should be made. Of course, clear programming and strict fiscal control are still needed. Let us follow the principle of subsidiarity in regional policy as well. Finally, our experience indicates that structures of political management and responsibility need to be strengthened in the regions. The local governments have established development agencies for their regions. However, coordination is still lacking There is no consensus on the responsibilities in planning the development of regions, on the level of participation in projects, and on the necessary amount of cofinancing. Therefore, two-level regional municipalities should be established that would fall within the territories of the planning regions and that could take full responsibil- ity for the development of their regions, including the use of the EU resources. Latvia has already started preparing the regional reform. A year ago, leading economists in Latvia concluded that "Latvia will become a monocentric state with a fragmented regional structure." We hope that effective help from the EU funds will prevent development of this unfavorable situation and will promote balanced long-term development for Latvia. We must contribute our wis- dom, our will, and our daily efforts to these ends. It all depends on us. CHAPTER 17 Discussion Notes LuLs Madureira Pires Three main topics were to be discussed at our session regarding management of Structural Funds after accession: institutional arrangements, financial absorption capacity, and the development policy priorities. As a background to this discussion, there are, of course, issues such as the rela- tionship between the European Union (EU) regional policy and domestic regional policies, the effectiveness of domestic regional policies and decentralization pro- cesses, the participation of regional players in allocating and managing EU funds, and the impact of additionality and fund absorption on the priorities of national devel- opment policy. The institutional topic received the most attention from our speakers. We were very lucky to hear excellent lectures on the advantages and disadvantages of centralized/ decentralized management of the EU Structural Funds and the significant and con- tradictory role played by the European Commission in this area Although I do not fully agree with Szomburg when he says that the "regional- ized concept of the Structural Funds is not fully adequate to CEEC [Central and Eastern European Countries]" just because the Central and Eastern European coun- tries' temtories are, for now, entirely eligible for Objective I of the Structural Funds, I do agree that the special arrangements imposed by the Commission on the candidate countnes for the implementation of the first programming penod (2004-2006) favor neither the strengthening of the regional and local tiers of government nor a fair involvement of regional and local players in fund allocation and management. 227 "**==>This document did not complete OCR process. <==**"