## Regional Economic Benefits from Deep Integration ## Source details **Canonical URL:** [Regional Economic Benefits from Deep Integration](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025047-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025047-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025047-print-pdf.pdf.json) --- ### I. Introduction — scope and headline findings - Context: - Europe’s income gap to the United States narrowed by about 10 percentage points in the early 2000s; higher growth in countries that joined the EU in 2004 (“new EU member states”) accounted almost entirely for the partial convergence. - Ten countries are currently candidates for EU membership, with many in the Western Balkans. - EU accession involves access to the EU single market and negotiation across 35 chapters covering economic and social conditions (including institutions and the rule of law); this “deep” integration can generate larger benefits through efficiency and scale economy gains. - Key contributions: - Use of a synthetic difference-in-difference estimator (Arkhangelsky et al. 2021) on regional data. - Factor and sectoral decomposition of gains. - Analysis of regional heterogeneity and initial conditions. - Main headline estimates and patterns: - Accession is estimated to have added more than 30 percent to income per capita (highly significant and robust). - Income gains transmitted equally through productivity catch-up and capital deepening. - Industry drove gains initially; services contributed significantly after a few years. - All regions in new member states gained, with within-country differences explaining about 40 percent of the variation (example range 5 to 47 percent in Hungary). - Regions with better access to finance and those more integrated through value chains prior to accession experienced higher growth and productivity gains. - Old member states gained too—on average close to 10 percent at the end of the sample (when dropping Greece)—as firms expanded production and reaped efficiency gains. ### II. Methodology and data - Estimator and implementation: - Synthetic difference-in-difference estimator (Arkhangelsky et al. 2021) applied to regional (NUTS2) data; implemented following Clarke et al. (2023). - Estimator weights untreated regions in a donor pool to match pre-treatment trends of treated regions and employs bootstrapping for inference. - Permits violation of parallel trends in aggregate data and allows constant level differences between treated and untreated groups. - 95% confidence intervals and p-values are based on Large-Sample approximations as in Arkhangelsky et al. (2021). - Formal setup: - The estimated average treatment effect on the treated, 휏̂ ௌ஽ூ஽ா , is generated from a two-way (훼 ௜ and 훽 ௧ ) fixed effect regression with optimally chosen weights 휔ෝ ௜ ௌ஽ூ஽ா across regions and 휑ො ௧ ௌ஽ூ஽ா across time periods. - Unit-fixed effects imply matching on pre-treatment trends. - Donor-pool selection: - Baseline donor pool: 179 NUTS2 regions in 14 old EU members (Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden). - Robustness checks include alternative donor pools (non-European OECD regions). - Drivers and decomposition: - Applied estimator to output and to capital and labor stock components of a Cobb-Douglas production function: ∆푦 = ∆푡푓푝 + 훼∆푘 + (1 − 훼)∆푙. - Capital share 훼 estimated as the average level of EU new member states’ capital share in 2004 and 2022: 0.57. - Estimator also applied to real GVA and subcomponents: industry, services, agriculture. - Heterogeneity analysis: split regions into below and above median characteristics and estimate separately. - Data: - Treatment: 45 NUTS2 regions in the ten new EU member states. - Baseline donor pool: 179 NUTS2 regions in 14 old EU member states. - Main data source: Annual Regional Database of the European Commission's Directorate General for Regional and Urban Policy (ARDECO); series start in 1980. - Variables: real GDP, real GVA by sectors (million EUR 2015), GDP per capita (current market prices in million PPS), total employment (thousands of persons), real capital stock (million EUR 2015). - Additional regional initial-condition measures: geographic and economic proximity (Amendolagine et al. 2024), financial depth (average ratio of long-term debt to sales of firms in each region in 2000 from Orbis). - 2003 baseline regional statistics: - Average region in old EU member states: population 1.8 million; GDP per capita 43 thousand PPS. - Average region in new member states: population 1.6 million; GDP per capita 19 thousand PPS. - Robustness donor-pool alternative: OECD regional database—covers 43 territorial level 2 regions from eight new EU Member States and 269 global non-European regions from Australia, Canada, the Republic of South Korea, New Zealand, and the United States; data starts in 2001. ### III. Impact on new member state regions — magnitude, timing, and drivers - GDP per capita gains (pooled estimation): - Since accession, GDP per capita gains averaged 27 percent. - Gain exceeded 10 percent after 5 years and 40 percent at the end of the sample. - The gain is equivalent to a 2 percentage points higher GDP growth rate per year. - Prior to accession, matching on changes in GDP per capita for a decade indicates the estimator effectively isolates accession effects starting in 2004. - Donor-pool weighting and pre-treatment fit: - Non-zero weights assigned to 115 donor-pool regions, representing 64 percent of all donor regions. - Maximum weight assigned to any region is 3.7 percent; two regions exceed 3 percent; nine regions between 2 percent and 3 percent. - Weighting reduces the income gap between new and old EU members before accession by more than half. - Greek regions account for 28 percent of the weight, followed by Spain and Germany; these three countries collectively account for 60 percent of the control group. - Regional heterogeneity: - By the end of the sample, regional income gains range from 78 percent in the Central and Western Lithuania Region to 4 percent in Severozápad (Northwest) Region in the Czech Republic. - Country-level heterogeneity: highest gains in the Baltics; lowest in Cyprus. - GDP-weighted aggregate gains from separate regional estimations average 28 percent until 2022 versus 27 percent in the pooled estimation (not statistically different). - Factor contributions to real GDP level gains: - Real GDP gains are 34 percent by 2022. - Sustained capital accumulation contributed around 20 percentage points to gains in GDP by the end of the sample—slightly more than half of the GDP gains. - A large fraction of capital accumulation came through foreign direct investment (FDI); net FDI inflows averaged around 3.6 percent of GDP in new member states between 2004 and 2007. - Increased productivity (TFP) accounted for slightly less than half of the income gains: 14 percentage points. - Applying the estimator directly to the capital stock shows divergence beginning in the year of accession and culminating into a 35 percent difference. - Employment had a very small negative effect on GDP. - Sectoral contributions to real GVA: - Agriculture: negligible contribution. - Industry: initially the main driver (relocation of industrial production and construction boom); contribution began to stagnate in 2019 and declined in 2021 and 2022. - Services: began contributing in 2011 and increased over time, accounting for 16 percent of the higher GVA compared to the synthetic control group by the end of the sample period. ### IV. Initial conditions and heterogeneity of gains - Trade and value-chain integration: - Regions with stronger pre-accession trade linkages with the single market gained more after accession. - Dividing regions by median economic integration shows a nearly 10 percentage point higher average income gain over 15 years for more integrated regions. - Financial depth: - Regions with more developed financial markets prior to accession experienced larger gains. - Regions with above median initial financial depth increased income per capita almost two-fold compared to those below the median. - Cities and human capital: - Regions containing capital cities gained on average 27 percent after 15 years, while others gained on average 22 percent. - Regions with a share of tertiary education above the top 20 percent quantile benefitted on average 1.5 times as much as others after 15 years. - Migration: - Around a million people left the new EU members between 2004 and 2007; authors find little evidence that EU accession notably accelerated outward migration (results available upon request). ### V. Robustness checks and alternative estimations - Alternative donor pools: - OECD-based donor pool (non-European regions): estimator assigns non-zero weight to 239 regions (89 percent of donor regions). Maximum weight assigned to any region is 2 percent; U.S. regions have 67 percent of weight, Republic of South Korea 13 percent, Japan 11 percent. - Excluding Greece from donor pool: results indistinguishable until 2010; then suggest permanently lower but still material gains. - Excluding Germany: gains somewhat larger than baseline from 2011 until end of sample. - Accounting for pre-accession gains: - Shifting treatment year to 2000 identifies pre-accession gains. Five years after accession gains exceed 20 percent (compared to 16 percent in baseline); additional gains diminish and are insignificant six years after accession. - Different estimation strategies and reported ATTs: - Baseline Synthetic Difference-in-Differences: ATT 0.27, P>|t| 0.00, 95% conf. Interval 0.23 to 0.32. - Difference-in-Differences: ATT 0.31, P>|t| 0.00, 95% conf. Interval 0.26 to 0.35. - Synthetic Control: ATT 0.16, P>|t| 0.078, 95% conf. Interval -0.01 to 0.33. - Synthetic Control w/t Mayotte: ATT 0.32, P>|t| 0.040, 95% conf. Interval 0.015 to 0.62. - Short pre-accession (shorter pre-period): Synthetic Dif-in-Dif ATT 0.27; Manufacturing covariate ATT 0.27; Tertiary education covariate ATT 0.29 (all with P>|t| 0.00). - Sensitivity: dropping the region with highest SCM weight (Mayotte) increases SCM estimate from 16 percent to 32 percent. - Including covariates: - Controlling for manufacturing share does not change gains. - Controlling for share of population with tertiary education somewhat increases the treatment effect. ### VI. Impact on old member state regions - Aggregate and temporal patterns: - Initial decline in GDP per capita in 2004 and 2005 relative to synthetic control due to broad-based growth slowdown among old members. - Improvement relative to control group up until the Global Financial Crisis. - Gains hovered around five percent from 2008 to 2016; about 10 percent by end of sample when excluding countries affected by the European Debt Crisis. - Average impact on all old member states hovers around zero through most of the sample, with the gain reaching 6.7 percent at the end of the sample. - Overall estimated gains from EU enlargement for old member states are around 5 to 10 percent (lowest estimate for the group including all old member states). - Regional heterogeneity: - After five years: gains in Scandinavia, Austria, Germany, and Spain. - After fifteen years: largest gains in regions in Germany and Austria; substantial gains in Scandinavia and regions farther away (e.g., Portugal). - Apparent losses in Greece and Italy largely reflect the European Debt Crisis and other country-specific developments post-2004. - GDP-weighted aggregate impacts from region-specific estimations are similar to pooled estimates but add around 1.5 percentage points to overall gain. ### VII. Key quantitative findings (selected) - Average income gains for new member states (baseline synthetic Dif-in-Dif): more than 30 percent (ATT 0.27 corresponds to 27 percent in context). - Real GDP gains by 2022: 34 percent. - Productivity (TFP) contribution to income gains: 14 percentage points. - Capital accumulation contribution to GDP gains: around 20 percentage points. - Capital stock divergence versus synthetic control: 35 percent difference. - Services sector contribution to higher GVA by end of sample: 16 percent. - Regions with capital cities: average gain 27 percent after 15 years; other regions 22 percent. - Regions with top 20 percent tertiary education share: benefitted on average 1.5 times as much after 15 years. - Net FDI inflows averaged around 3.6 percent of GDP in new member states between 2004 and 2007. - Old member states’ gains: around 5 to 10 percent; aggregate gain reaching 6.7 percent at end of sample for all old member states. ### VIII. Policy implications and recommendations - Preconditions that amplify accession gains: - Deepen initial economic integration with established markets to leverage production networks. - Improve access to long-term finance and deepen financial markets to facilitate investment and capital accumulation. - Invest in education and skills (higher tertiary attainment linked to larger gains). - Reduce costs for firms to establish cross-border production networks to capture economies of scale from a larger single market. - For candidate countries aiming to obtain similar gains as 2004: - Implement far-reaching reforms to overcome reform gaps in economic and broader institutional setups that may be larger than those in 2004. - Consider that large redistributive programs financed by the EU in 2004 may not be replicated; uncertainty exists about future EU funding and complexity in an enlarged EU. ### IX. Conclusion - Using a synthetic difference-in-differences estimator on regional data, the 2004 EU enlargement produced large income gains averaging more than 30 percent for new members; results are robust across specifications. - Gains were heterogeneous across regions and driven primarily by capital accumulation and productivity increases rather than geography alone. - Old member states also benefitted, with estimated gains around 5 to 10 percent, concentrated in regions integrated with new member states but also present farther away. - The 2004 enlargement example underscores benefits of deep integration (single market, financial transfers, political stability) but cautions that similar gains for future accession rounds are not guaranteed without substantial reforms and uncertain fiscal support. *IMF Working Paper: Regional Economic Benefits from Deep Integration (excerpt from WP/2025/047, content as provided in the source PDF)* ### References ............................................................................................................. ### Regional Economic Benefits from Deep Integration ### I. Introduction - In the early 2000s, Europe’s income gap to the United States narrowed by about 10 percentage points; higher growth in countries that joined the European Union (EU) in 2004—referred to as “new EU member states”—accounted almost entirely for the partial convergence. - Ten countries are currently candidates for EU membership, with many of them in the Western Balkans. - EU accession entails access to the EU single market and negotiation across 35 chapters covering economic and social conditions (including institutions and the rule of law); this “deep” integration can generate larger benefits through efficiency and scale economy gains. - The 2004 enlargement provides a strong empirical setting for quantifying accession impacts because of long pre- and post-accession periods and no concurrent major shocks affecting treatment or donor groups around accession. - Key contributions of the paper: - Use of a synthetic difference-in-difference estimator (Arkhangelsky et al. 2021) on regional data. - Factor and sectoral decomposition of gains. - Analysis of regional heterogeneity and initial conditions. - Main headline estimates and patterns: - Accession is estimated to have added more than 30 percent to income per capita (highly significant and robust). - Income gains transmitted equally through productivity catch-up and capital deepening. - Industrial sector drove gains initially; services contributed significantly after a few years. - All regions in new member states gained, but effects vary strongly (within-country differences explain about 40 percent of the variation; example range 5 to 47 percent in Hungary). - Regions with better access to finance and those more integrated through value chains prior to accession experienced higher growth and productivity gains. - Old member states gained too—on average close to 10 percent at the end of the sample (when dropping Greece)—as firms expanded production and reaped efficiency gains. ### II. Methodology and Data - Estimator: - Synthetic difference-in-difference estimator (Arkhangelsky et al. 2021) applied to regional (NUTS2) data. - The estimator weights untreated regions in a donor pool to match pre-treatment trends of treated regions and employs bootstrapping for inference. - Permits violation of parallel trends in aggregate data and allows constant level differences between treated and untreated groups. - Implemented following Clarke et al. (2023); 95% confidence intervals and p-values are based on Large-Sample approximations as in Arkhangelsky et al. (2021). - Formal setup: - The estimated average treatment effect on the treated, 휏̂ ௌ஽ூ஽ா , is generated from a two-way (훼 ௜ and 훽 ௧ ) fixed effect regression with optimally chosen weights 휔ෝ ௜ ௌ஽ூ஽ா across regions and 휑ො ௧ ௌ஽ூ஽ா across time periods. - Unit-fixed effects imply matching on pre-treatment trends. - Donor pool selection considerations: - Baseline donor pool: all old EU member regions (14 old EU members: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden). - Trade-offs: non-European regions less impacted by treatment but shorter pre-accession sample and structural differences; old member regions allow longer pre-treatment period and better matching but could bias estimates if old regions benefited from enlargement or were affected differentially by EU funds or shocks. - Robustness checks include alternative donor pools (non-European OECD regions). - Drivers and decomposition: - Applied estimator to output and to capital and labor stock components of a Cobb-Douglas production function: - ∆푦 = ∆푡푓푝 + 훼∆푘 + (1 − 훼)∆푙 - Capital share 훼 estimated as the average level of EU new member states’ capital share in 2004 and 2022: 0.57. - Estimator also applied to real Gross Value Added (GVA) and subcomponents: industry, services, agriculture. - Heterogeneity analysis: split regions in new member states into below and above median characteristics and estimate separately. - Data: - Baseline: 45 NUTS2 regions in the ten new EU member states (treatment) and 179 NUTS2 regions in 14 old EU member states (baseline donor pool). - Main data source: Annual Regional Database of the European Commission's Directorate General for Regional and Urban Policy (ARDECO), based on harmonized Eurostat and national/international sources; series start in 1980. - Variables: real GDP, real GVA by sectors (million EUR 2015), GDP per capita (current market prices in million PPS), total employment (thousands of persons), real capital stock (million EUR 2015). - Additional regional initial-condition measures: - Geographic and economic proximity (Amendolagine et al. 2024): geographic distance between major regional cities and economic linkages in 2000 using cross-regional input-output (bilateral value-added trade). - Financial depth: average ratio of long-term debt to sales of firms in each region in 2000 from Orbis firm-level data. - 2003 baseline regional statistics: - Average region in old EU member states: population 1.8 million; GDP per capita 43 thousand PPS. - Average region in new member states: population 1.6 million; GDP per capita 19 thousand PPS. - Robustness donor-pool alternative: OECD regional database—covers 43 territorial level 2 regions from eight new EU Member States and 269 global non-European regions from Australia, Canada, the Republic of South Korea, New Zealand, and the United States; data starts in 2001. ### III. Impact on New Member State Regions - GDP per capita gains (pooled estimation): - Since accession, GDP per capita gains averaged 27 percent. - Gain exceeded 10 percent after 5 years and 40 percent at the end of the sample. - The gain is equivalent to a 2 percentage points higher GDP growth rate per year. - Prior to accession, matching on changes in GDP per capita for a decade indicates the estimator effectively isolates accession effects starting in 2004. - Donor pool weighting details: - The estimator assigns non-zero weights to 115 donor pool regions, representing 64 percent of all donor regions. - Maximum weight assigned to any region is 3.7 percent; two regions exceed 3 percent; nine regions between 2 percent and 3 percent. - Weighting reduces the income gap between new and old EU members before accession by more than half. - Greek regions account for 28 percent of the weight, followed by Spain and Germany; collectively these three countries account for 60 percent of the control group. - Robustness checks include donor pools excluding Greece and Germany and using non-European regions. - Regional heterogeneity: - By the end of the sample, regional income gains range from 78 percent in the Central and Western Lithuania Region to 4 percent in Severozápad (Northwest) Region in the Czech Republic. - Country-level heterogeneity: highest gains in the Baltics; lowest in Cyprus. - GDP-weighted aggregate gains from separate regional estimations average 28 percent until 2022 versus 27 percent in the pooled estimation (not statistically different). - Factor contributions (to real GDP level gains): - Real GDP gains are 34 percent by 2022 (somewhat lower than GDP per capita gains, reflecting worse demographics in new members relative to controls). - Sustained capital accumulation contributed around 20 percentage points to gains in GDP by the end of the sample—slightly more than half of the GDP gains. - A large fraction of capital accumulation came through foreign direct investment (FDI), especially in the years prior to the Global Financial Crisis; net FDI inflows averaged around [figure truncated in provided content]. *IMF Working Paper (excerpt): Regional Economic Benefits from Deep Integration* ### 3.6 percent of GDP in new member states between 2004 and 2007. Another source for financing of investment ### wpiea2025047-print-pdf - 3.6 percent of GDP in new member states between 2004 and 2007. Another source for financing of investment ### Factor and sectoral decomposition of gains - Capital accumulation and productivity were the primary drivers of income gains after EU accession. - Increased productivity (TFP) accounted for slightly less than half of the income gains from EU accession (14 percentage points). - Capital stock divergence: applying the estimator directly to the capital stock shows divergence beginning in the year of accession and culminating into a 35 percent difference (results available upon request). - Employment had a very small negative effect on GDP, reflecting a combination of developments including the employment rate, labor force participation, demographics, and migration. - Sectoral contributions to real GVA: - Agriculture: contribution remains negligible. - Industry: initially the main driver (relocation of industrial production and construction boom); contribution began to stagnate in 2019 and declined in 2021 and 2022. - Services: began contributing in 2011 and increased over time, accounting for 16 percent of the higher GVA compared to the synthetic control group by the end of the sample period. ### Initial conditions and heterogeneity of gains - Regions with stronger pre-accession trade linkages with the single market gained more after accession. - Dividing regions by median economic integration shows a nearly 10 percentage point higher average income gain over 15 years for more integrated regions. - Regions with more developed financial markets prior to accession experienced larger gains. - Regions with above median initial financial depth increased income per capita almost two-fold compared to those below the median. - Other initial conditions: - Regions containing capital cities gained on average 27 percent after 15 years, while others gained on average 22 percent. - Regions with a share of tertiary education above the top 20 percent quantile benefitted on average 1.5 times as much as others after 15 years. - Migration: while around a million people left the new EU members between 2004 and 2007, the authors find little evidence that EU accession notably accelerated outward migration (results available upon request). ### Robustness checks and alternative estimations - Alternative donor pools: - OECD-based donor pool (non-European regions): estimator assigns non-zero weight to 239 regions (89 percent of donor regions). Maximum weight assigned to any region is 2 percent; U.S. regions have 67 percent of weight, Republic of South Korea 13 percent, Japan 11 percent. - Excluding Greece from donor pool: results indistinguishable until 2010; then suggest permanently lower but still material gains. - Excluding Germany: gains somewhat larger than baseline from 2011 until end of sample. - Accounting for pre-accession gains: - Shifting treatment year to 2000 (start of transition period with Agenda 2000) identifies pre-accession gains. Five years after accession gains exceed 20 percent (compared to 16 percent in baseline); additional gains diminish and are insignificant six years after accession. - Different estimation strategies: - Baseline Synthetic Difference-in-Differences (long pre-accession): ATT 0.27, P>|t| 0.00, 95% conf. Interval 0.23 to 0.32. - Difference-in-Differences: ATT 0.31, P>|t| 0.00, 95% conf. Interval 0.26 to 0.35. - Synthetic Control: ATT 0.16, P>|t| 0.078, 95% conf. Interval -0.01 to 0.33. - Synthetic Control w/t Mayotte: ATT 0.32, P>|t| 0.040, 95% conf. Interval 0.015 to 0.62. - Short pre-accession (shorter pre-period): Synthetic Dif-in-Dif ATT 0.27, Manufacturing covariate ATT 0.27, Tertiary education covariate ATT 0.29 (all with P>|t| 0.00 and similar 95% conf. Intervals). - Sensitivity: dropping the region with highest SCM weight (Mayotte) increases SCM estimate from 16 percent to 32 percent. - Including covariates: - Controlling for manufacturing share does not change gains. - Controlling for share of population with tertiary education somewhat increases the treatment effect. ### Impact on existing (old) member state regions - Enlargement effects on old member states: - Initial decline in GDP per capita in 2004 and 2005 relative to synthetic control due to broad-based growth slowdown among old members. - Improvement relative to control group up until the Global Financial Crisis. - Gains hovered around five percent from 2008 to 2016; about 10 percent by end of sample when excluding countries affected by the European Debt Crisis. - Average impact on all old member states hovers around zero through most of the sample, with the gain reaching 6.7 percent at the end of the sample. - Overall estimated gains from EU enlargement for old member states are around 5 to 10 percent (with lowest estimate for group including all old member states). - Heterogeneity across regions: - After five years: gains in Scandinavia, Austria, Germany, and Spain. - After fifteen years: largest gains in regions in Germany and Austria (well integrated with Central and Eastern Europe); substantial gains in Scandinavia and regions farther away (e.g., Portugal, possibly due to tourism). - Apparent losses in Greece and Italy largely reflect the European Debt Crisis and other country-specific developments post-2004. - GDP-weighted aggregate impacts from region-specific estimations are similar to pooled estimates but add around 1.5 percentage points to overall gain. ### Key quantitative findings - Average income gains for new member states (baseline synthetic Dif-in-Dif): more than 30 percent (average treatment effect 0.27 in table corresponds to 27 percent in context). - Productivity contribution: 14 percentage points (slightly less than half of income gains). - Capital stock divergence: culminating into a 35 percent difference versus synthetic control (results available upon request). - Services sector contribution to higher GVA by end of sample: 16 percent. - Regions with capital cities: average gain 27 percent after 15 years; others 22 percent. - Regions with top 20 percent tertiary education share: benefitted on average 1.5 times as much after 15 years. - Old member states’ gains: around 5 to 10 percent; aggregate gain reaching 6.7 percent at end of sample for all old member states. ### Policy implications and recommendations - Pre-conditions that amplify accession gains: - Deepen initial economic integration with established markets to leverage production networks. - Improve access to long-term finance and deepen financial markets to facilitate investment and capital accumulation. - Invest in education and skills (higher tertiary attainment linked to larger gains). - Reduce costs for firms to establish cross-border production networks to capture economies of scale from a larger single market. - For candidate countries aiming to obtain similar gains as 2004: - Implement far-reaching reforms to overcome reform gaps in economic and broader institutional setups that may be larger than those in 2004. - Consider that large redistributive programs financed by the EU in 2004 may not be replicated; uncertainty exists about future EU funding and complexity in an enlarged EU. ### Conclusion summary - Using a synthetic difference-in-differences estimator on regional data, the 2004 EU enlargement produced large income gains averaging more than 30 percent for new members; results are robust across specifications. - Gains were heterogeneous across regions and driven by capital accumulation and productivity increases rather than geography alone. - Old member states also benefitted, with estimated gains around 5 to 10 percent, concentrated in regions integrated with new member states but also present farther away. - The 2004 enlargement example underscores benefits of deep integration (single market, financial transfers, political stability) but cautions that similar gains for future accession rounds are not guaranteed without substantial reforms and uncertain fiscal support. *IMF WORKING PAPERS — Regional Economic Benefits from Deep Integration (content as provided in the source PDF)* ### References ### References ### Cited works - Amendolagine, V., Prota, F., & Serlenga, L. (2024). The impact of European Cohesion Policy: a spatial perspective. Journal of Economic Geography, 24(4), 477–494. - Arkhangelsky, D., Athey, S., Hirshberg, D. A., Imbens, G. W., & Wager, S. (2021). Synthetic difference-in-differences. American Economic Review, 111(12), 4088-4118. - Baldwin and Wyplosz (2012). The Economics of European Integration. McGraw Hill, New York. - Barwick, P. J., Li, S., Lin, L., & Zou, E. Y. (2024). From fog to smog: The value of pollution information. American Economic Review, 114(5), 1338-1381. - Brou, D., & Ruta, M. (2011). Economic integration, political integration or both?. Journal of the European Economic Association, 9(6), 1143-1167. - Bruno, R., Petrakou, M., & Phelps, N. (2023). The Impact of FDI and Financial Depth on EU Regional Growth: Income and Spatial Heterogeneity. Economic and Business Review, 25(2), 64-78. - Campos, N. F., Coricelli, F., & Moretti, L. (2019). Institutional integration and economic growth in Europe. Journal of Monetary Economics, 103, 88-104. - Campos, N. F., Coricelli, F., & Franceschi, E. (2022). Institutional integration and productivity growth: Evidence from the 1995 enlargement of the European Union. European Economic Review, 142, 104014. - Cerqua, A., Ferrante, C., & Letta, M. (2023). Electoral earthquake: Local shocks and authoritarian voting. European Economic Review, 156, 104464. - Chupilkin, M., Kóczán, Z., and Plekhanov, A. (2024). 20 years of EU membership. Office of the Chief Economist, European Bank for Reconstruction and Development. - Clarke, D., Pailañir, D., Athey, S., & Imbens, G. (2023). Synthetic difference in differences estimation (No. 159070). IZA Discussion Paper. - Cuaresma, J. C., Doppelhofer, G., & Feldkircher, M. (2012). The Determinants of Economic Growth in European Regions. Regional Studies, 48(1), 44–67. - Dench, D., Pineda-Torres, M., & Myers, C. (2024). The effects of post-Dobbs abortion bans on fertility. Journal of Public Economics, 234, 105124. - Deng, L., Lu, Y., & Tang, Y. (2024). Does FDI increase product innovation of domestic firms? Evidence from China. Journal of Economic Behavior & Organization, 222, 1-24. - Elsner, B. (2013). Emigration and wages: The EU enlargement experiment. Journal of International Economics, 91(1), 154-163. - Felbermayr, G., Gröschl, J., & Heiland, I. (2022) Complex Europe: quantifying the cost of disintegration. Journal of International Economics, 138 - Fleuriet, V., & Vertier, P. (2024). Does the G20 compact with Africa foster economic growth: Evidence from synthetic differences. Economics Letters, 236, 111570. - Grassi, B. (2024). The EU Miracle: When 75 Million Reach High Income (No. 709). IGIER Working Paper. - Jarrett, U., & Mohtadi, H. (2023). Understanding the globalization-crisis linkage: A “differenced” approach. Review of International Economics, 31(5), 1616-16410. - Sandkamp, A. (2020). The trade effects of antidumping duties: Evidence from the 2004 EU enlargement. Journal of International Economics, 123, 103307. - Walz, U. (1998). Does an enlargement of a common market stimulate growth and convergence?. Journal of International Economics, 45(2), 297-321. *Regional Economic Benefits from Deep Integration: 20 years after the 2004 EU Enlargement Working Paper No. WP/2025/047* --- _Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025047-print-pdf.pdf_