EU Regional Convergence: Where do we stand?
1. Introduction: Convergence and Cohesion.
The outbreak of the financial and sovereign debt crisis has dragged the European Union (EU) into a deep economic recession. While the economic cycle has now started reversing and positive growth rates are registered across all EU Member States (MSs), there is little doubt that the economic crisis has produced severe and long-lasting consequences both at national and regional level. Against this background, this short piece aims at reviewing the status of economic convergence within the EU.
Economic convergence refers to the process according to which relatively poorer countries (or regions) grow faster than relatively richer ones, generating therefore a catching-up effect of the former to the latter. Accordingly, convergence is the fundamental economic mechanism as well as the logical precondition to achieve economic and social cohesion within the EU.
The latter lays as an explicit EU objective and finds its first formulation in the Single European Act (1986), where Article 130a reads “in order to promote its overall harmonious development, the Community shall develop and pursue its actions leading to the strengthening of its economic and social cohesion.” This constitutes the legal ground for the creation of the European Structural Funds, as well as the backbone of the EU Cohesion Policy. The latter was intended act as a backstop against the possible relocation of economic activities from the periphery to the core of the EU, following the creation of the European Single Market and the introduction of the Four Freedoms (freedom of movement of people, of goods, of capital, of establishing and providing service ).
After having been shelved for many years due to the consolidation of a strong economic cycle, both concepts of convergence and socio-economic cohesion are coming back to the centre of the debate and stimulating the interest of scholars and policymakers. It is appears clear that, as the ongoing recovery shows strong difference in MSs’ growth rates, the same concern standing behind Article 130a has been strongly remerging. In other words, the risk is that, in the absence of national barriers, difference in productivity may exacerbate the polarisation between industrialised and rural areas within the Union. Indeed, conceived by a strong economic cycle until 2007, the crisis has unveiled some structural weaknesses at national and regional level, which in turn determined lower growth rates in many peripheral countries.[RM1]
Furthermore, a more structural reason justify the revival of the debate of economic convergence and socio-economic cohesion. Besides, the relevant political implications, a more cohesive union lays as precondition for a more effective economic governance. Therefore, sustainable economic convergence is crucial for the correct functioning of both the European Union and the Eurozone. To give an example, calibration of ECB monetary policy, Euro exchange-rate, commercial imbalances across EU MSs, fiscal transfers as well as EU common expenditures are all elements which are influenced by the level economic of convergence, and that in turn may impact MSs’ macroeconomic performances.
In light of these considerations, economic convergence is an important topic that it is likely to gain more and more attention amongst EU policymakers, also in regard of the transformations envisaged in the post-Brexit era.
2. Overview on Convergence Theory
As a way of introduction, the three main strands constituting the available literature on economic convergence are briefly described hereafter.
The Neoclassical approach to convergence derives directly from the Solow’s growth model. It assumes that (1) a country’s level of technology is determined by a number of exogenous factors, and that (2) capital has diminishing marginal returns. Accordingly, capital flows downhill, from capital-intense to relatively capital-scarce countries, and determines GDP growth differentials. This dynamics implies the necessary existence of a catching-up effect. Upon this basis, neoclassical models predict the so-called unconditional convergence, as GDP growth rates are exclusively determined by the initial stock of capital available in the country.
Nevertheless, while preliminary evidence were found in the early datasets on GDP pro capita across countries (i.e. Maddison’s dataset, 1982), the hypothesis of unconditional convergence failed the test with more comprehensive databases (i.e. Heston and Summers dataset, 1991). The failure of cross-country convergence motivates the formulation of “models that drop the two central assumption of the neoclassical model” (Romer, 1994) opening, thereby, the door to the second strand of literature.
Romer and Lucas are considered the founders of the New Growth Theory (Lucas 1988; Romer 1986). [MP2] As opposed to the neoclassical models, the latter predicts the possibility for a conditional convergence, which results from application of endogenous growth models. Demonstrating that a country’s level of technology depends on the relative wealth of the country, they argue that economies tend to converge to same level of GDP pro capita if initial situations are similar.
The confrontation of the two main assumptions of the early models stimulated the creation of several modified neoclassical approach. An extensive review of these models can be found in Borsi and Metiu (2013). Within this area of research, Azariadis (1996) and Galor (1996[MP3] ) were the first to formulate the concept of club convergence, according to which a robust convergence can be noticed in groups of homogenous countries. However, if in the theory of conditional convergence argues that a country approaches its own but unique equilibrium, club convergence theory predicts the existence of multiple equilibria (Islam, 2003).
The last set of models proposed in the literature identify the difference in the quality of the institutions as the most important variable in order to cross-county difference in GDP pro capita. North (1990) and Easterly and Levine (1997) were the firsts to explore the link between long-term growth and quality of institutions. On this track, an increasing number of theoretical and empirical works have started analysing the economic convergence across the EU. The common objective of these researches was to identify the driver behind productivity growth and investigate the impact of the introduction of the Euro (Monfort, 2008; del Hoyo et al., 2017; ECB, 2015).
While the identification of the drivers of productivity growth will be explored in the following pieces of the series, the present study will give an overview of the current status of convergence, using the two main indicators in the available literature. Sigma and beta convergence find their formulation and first applications in the studies of Baumol [MP4] (1986), Barro and Sala-i-Martin (1992), and Mankiw et al. (1992).[MP5]
This paper looks at the different layers of economic convergence. The analysis of convergence will take place both at EU28 Member State level and at NUTS 2 regional level. The key variable of interest is GDP per capita in purchasing power standards (PPS) for the period 2000-2015 (Eurostat). The aim is to assess whether looking at the national aggregates hides some regional dynamics that should instead be taken into account. To assess convergence trends, this paper will first look at the relationship between the speed of GDP per capita growth relative to the EU average and the relative wealth of the region in year 2000 (beta convergence). Secondly, it will explore the evolution of the coefficient of variation (standard deviation divided by the mean) of GDP per capita in PPS among regions (sigma convergence).
Clusters of regions are constructed in order to add a level of analysis that goes beyond national borders and seek to find geographic commonalities among countries. Three clusters are identified. Central Eastern Europe includes Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia. North Western Europe is composed of Austria, Denmark, Germany, Finland, France, Luxembourg, Netherlands, Sweden, and the United Kingdom. Finally, Southern Europe includes Cyprus, Greece, Italy, Malta, Portugal, and Spain.
4.1 Member States
Convergence has been taking place among EU28 MS since year 2000. The top graph “EU Member States” plots MS according to their relative wealth compared to the EU average in year 2000 and the change of their relative position between 2000 and 2015 (beta convergence). As the theory predicts and as already presented by some (see for example ECB, 2015), Member States with a lower initial relative wealth are those that have experienced the larger positive change (top left quadrant), thus converging towards the EU average. Accordingly, the position of richer Member States has deteriorated in relation to the EU average (bottom right quadrant). Interestingly but not surprisingly, Central Eastern European (CEE) countries are those that have grown the most, thus converging faster to the position of other EU MS. Among this group, Lithuania, Estonia, Romania, Latvia and Slovakia are those that had the most impressive performance, gaining between 30% and 40% compared to their relative wealth position in year 2000. On the other side of the spectrum, Italy, Greece, Cyprus and the Netherlands are those that have lost the most compared to their situation in year 2000.
4.2 EU Regions
National aggregates may however mask other dynamics at regional level. The bottom graph “EU Regions (NUTS 2)” presents the growth of GDP per capita since 2000 relative to the EU average for NUTS 2 European regions. The distribution of regions follows the one of MS, although with more dispersion. Concerning the change since 2000 (y axis), one can easily identify two regions showing an astonishingly high rate of GDP per capita growth compared to the EU average. These are the capital regions of some Eastern European countries, namely Bucharest and Bratislava[MP6] , which have experienced much stronger growth the other regions in their own country. Conversely, there are regions that performed much worse than their national average. At MS level, only Italy has had a negative change larger than -20, while there are plenty of regions that have experienced a negative change that is larger than this.
The largest dispersion however is on the x axis. Compared to the EU average, in year 2000 there were many regions that were much richer than their national aggregate. At the same time, many regions belonging to the North Western cluster were below the EU average in 2000, while their national figure was above average. This shows that national aggregates may hide some internal dynamics, such as the presence of large disparities among regions of one country. By selecting a country in the filter on the right of the graph, the reader can see the relative position of the regions belonging to the chosen MS.
The majority of the regions with GDP per capita above the EU average has experienced negative change since 2000, while most of the poorest regions (below 60% of the EU average) have experienced positive change. These dynamics show that convergence is taking place. However, a comparison of the slopes of the convergence lines reveals that the speed of convergence is much slower at regional level (bottom graph) than when looking at Member States (top graph). Indeed while the beta coefficient for MS is -0.314, the one for regions is -0.203. The speed of convergence among regions is around 30% slower than among MS.
4.3 Geographic clusters
While convergence is taking place overall, one can identify some diverging regional dynamics. Regions in the top right quadrant are those that in year 2000 were richer than the EU average and that have become even richer in relation to the rest of the Union. They are thus diverging from the top, as their relative position has further improved over time. Geographically speaking, regions in this quadrant are overwhelmingly from North Western MS, with the exception of the regions of Praha (CZ), Bratislava (SK) and Pais Vasco (ES). Conversely, some regions are diverging from the bottom, i.e. they were poorer than the EU average in year 2000 and their position has further deteriorated over time (bottom left quadrant). Here, the worst performers are all from Southern Europe, mostly Italian, Greek and Spanish regions.
Looking at the cluster dimension of regional convergence, one can identify a few distinct developments. First, Central Eastern regions are predominantly present in the top left quadrant, indicating that those regions, which were the poorest, have been converging to the rest of the EU over the past 15 years. North Western regions have a more heterogeneous behaviour, with some of the richer regions converging towards the average and others diverging towards higher levels of GDP per capita and faster growth. Conversely, Southern regions are the worst performers, experiencing the largest negative change. While for some regions this means converging to the average, for many others this implies growing disparities towards the bottom of the scale.
4.4 Sigma Convergence
An analysis of the coefficient of variation (of GDP per inhabitant in pps) allows to take into consideration the time dimension of convergence. The top graph presented below shows that during the years 2000- 2007, convergence was taking place at both regional and national level within the EU, although at MS level the pace was faster. Since 2008 however, the variation at regional level has begun to increase, showing that, probably as a consequence of the global financial crisis, regional disparities in the EU have expanded. There has been a change of trend in 2014, but unfortunately more recent data is not available. Consequently, even if there has been convergence for several years, the level of dispersion of wealth among EU regions in 2015 was back to the level in 2000. At member state level instead, diverging trends were measured during 2009-2011 and 2013-2015, in coincidence with the aftermaths of the global financial crisis and the euro debt crisis.
Both beta and sigma computations show that in 2015 there was more divergence among regions than among member states. When breaking down the data by cluster, the same tendency appears (two bottom graphs). One can identify several interesting developments. First, Southern Europe is more uniform and shows a lower variation compared to the other two clusters. One can distinctively notice the impact of the crisis on regional divergence. Indeed, while the converging trend had slowed down already since 2006, during 2009-2012 it jumped up and has not yet returned to pre-crisis levels. Thus, the crisis has increased heterogeneity among Southern regions, as the poorest regions have become even poorer. Second, the North Western cluster is on a diverging path since the beginning of the period, both at regional and MS levels. This confirms what mentioned before, i.e. there are some ‘champion” regions booming to even higher levels of GDP per capita while others are left behind. Third, Central Eastern Europe shows two very different dynamics at MS and regional level. The bottom left graph shows CEE countries on a steady converging path, reaching a level of internal dispersion that is much lower than the of the North Western cluster. The crisis has barely influenced this trend, with a minimal slowdown between 2008 and 2009. This is apparently good news for EU convergence. However, a more careful analysis at regional level shows that such convergenc3e has not taken place at all, and the level of internal dispersion in the Central Eastern cluster is just slightly lower than the one in year 2000. This implies that national averages of CEE countries hide high levels of dispersion at regional level. One can consider for example that the capital regions of these countries (see the case of Bucharest and Bratislava mentioned above) have an outstanding growth performance that drives up the measure of the national average. However, at the same time some regions grow at much lower levels, thus explaining the increase in regional dispersion.
Barro R. J. and Sala-i-Martin X. (1992). “Convergence”. Journal of Political Economy, Vol 1000 (2): 223-251.
Borsi M. T. and Metiu N. (2014). “The evolution of economic convergence in the European Union”. Empirical Economics, Vol 48 (2): 657-681.
Diaz del Hoyo J. L., Dorrucci E., Heinz F. F. and Muzikarova S. (2017). “Real Convergence in the euro area: a long term perspective”. ECB Occasional Paper Series No 203.
ECB (2015). “Real convergence in the euro area: evidence, theory and policy implications”. ECB Economic Bulletin, Issue 5 / 2015.
Islam, N. (2003). “What have we learnt from the convergence debate?” Journal of Economic Surveys, Vol. 17 (3).
Monfort P. (2008). “Convergence of EU regions. Measures and Evolution”. Working Papers No1/2008. European Union Regional Policy.
Romer P. M. (1994). “The origins of endogenous growth”. Journal of Economic Perspectives, Vol. 8 (1): 3-22.