3. Convergence in socioeconomic outcomes within member states
Understanding patterns of regional convergence within countries can better clarify trends in socioeconomic outcomes. Convergence at EU level (across countries or regions as analysed in section 2) does not necessarily imply regional convergence within countries, for example when capital and metropolitan regions strongly outperform rural and peripheral regions within a country. It is important to look beyond convergence at EU level to capture potential challenges to territorial and social cohesion within specific Member States. This section provides regional convergence analysis within each of the 19 Member States with more than four NUTS2 regions. (85) It focuses on the indicators analysed in section 2 for which regional data are available (GDP per capita, (un)employment and NEET rates, tertiary and adult education rates, and AROPE rates), covering 2007 to 2022. It also provides an in-depth look at within-country convergence in median incomes across very small territorial units (NUTS3 regions) based on unique administrative data collected in eight Member States (Box 2.2(European Commission) (Cedefop, )).
Upward divergence in economic performance at EU level was observed within most EU Member States. Sixteen of the 19 Member States analysed saw their national GDP per capita increase since 2007 amid growing regional disparities (Table 2.2). The growth in regional disparities was particularly strong (more than 50%) in Denmark and several central and eastern European countries (Bulgaria, Czechia, Poland, Romania). In central and eastern Member States, divergence partly resulted from capital regions outperforming other regions. (86) This was not always the case in other countries, e.g.: increases in regional variation in France were linked to particularly sluggish growth in regions with low levels of GDP per capita. Only Portugal experienced upward convergence in GDP per capita, but this was largely due to low growth in some developed, previously dynamic, regions. (87)
Convergence patterns in labour market outcomes varied from country to country. Employment rates grew in most Member States since 2007, sometimes accompanied by regional convergence and sometimes by divergence (notably in Romania). Similarly, regional NEET rates converged in some countries and diverged in others, while average national rates largely declined. Concerningly, Denmark, France, and Romania all saw strong regional divergence in NEET rates. Within-country developments in unemployment rates are more reassuring, with 10 of the 18 countries experiencing upward convergence, notably Bulgaria, Czechia, Germany, and Portugal. Overall, within-country variation in labour market outcomes was highly sensitive to specific national and regional factors, but a detailed analysis is beyond the scope of this report.
Most countries experienced growth in tertiary education attainment accompanied by regional divergence. This was the case for 16 of the 18 countries analysed. In several eastern and central European Member States (Czechia, Hungary, Poland, Romania) and Portugal, regional differences grew particularly strongly. These developments stemmed from sharp increases in tertiary education attainment in capital regions, reflecting a combination of factors including the concentration of universities, high demand for tertiary-educated workers and associated wage premiums. By contrast, a lack of tertiary education opportunities and an outflow of highly qualified workforce posed challenges for some less urban regions, contributing to (risks of) talent development traps as outlined in the European Commission’s Communication on harnessing talent in Europe's regions. Only Finland saw regional differences in tertiary education attainment fall, though this was accompanied by a mild decline in attainment at national level.
Table 2.2
Convergence patterns in socio-economic outcomes vary by Member State
Within-country convergence/divergence across NUTS 2 regions, 2007-2022
Note: Only Member States with more than four NUTS2 regions are covered. Substantial change in national outcome average over time is defined either as 5% (GDP per capita) or 0.5pp (other indicators). * Developments for 2007-2021, due to data availability ** Developments for 2015-2022, due to data availability
Source: Analysis covers the same indicators (for the same age groups) as section 2.
More than half of the Member States experienced upward divergence in adult education participation, reflecting similar developments at EU level. The growth in regional differences was particularly strong in Austria, Czechia, Greece, France, Italy, Poland, Portugal, Romania, and Sweden. However, considerable increases in regional variation often reflected the initial situation in 2007 when adult learning participation was very low in many countries, irrespective of the region. Although some countries saw regional variation more than double, it remained relatively modest in most countries in 2022. Three countries (Finland, the Netherlands, and Spain) saw upward convergence in adult education participation.
National declines in social exclusion and poverty risks were typically accompanied by stable or declining regional differences. The evidence of within-country variation in AROPE rates is more tentative due to severe data limitations and changes in the AROPE definition over time, allowing analysis in just 10 Member States between 2015 and 2022. In most of these countries, national AROPE rates declined since 2015 and regional differences either declined (Czechia, Denmark, Finland) or remained stable (Hungary, Italy, Romania, Spain, and Sweden). Regional variation grew in Slovakia and Bulgaria alone during this period.
Box 2.2: Convergence in median incomes across small regions (NUTS 3) within selected Member States
This box provides an in-depth look at developments in incomes in small regions within nine EU Member States (1) with sufficiently detailed income data from administrative sources over longer periods of time (typically 2007-2022). (2) Compared to the rest of the analysis presented in this chapter, this provides a more granular analysis of geographical disparities in incomes for a large number of NUTS3 regional units. The limitation of these administrative data is that they are not fully harmonised, preventing inter-country comparison of specific results. Following other analyses of this data, (3) regions are compared in terms of median incomes that approximate the earnings of a typical person living in a given region.
Regional median incomes, averaged across small regions, have risen in real terms in almost all of the countries analysed (Chart 1). Annual increases range from 0.7% in Sweden between 2011 and 2021 to 2.8% in Latvia between 2007 and 2022. Regional median incomes declined only in Finland, by 1.1% per year on average. Most of this decline occurred in the aftermath of the 2007-2008 financial crisis and in 2022, following Russia’s war of aggression against Ukraine and the associated sharp increase in prices. (4)
Chart 1
Regional median incomes increased, while regional disparities declined in some countries and rose in others
Average (left axis) and standard deviation (right axis) of regional median real disposable incomes, 2007-2022
Notes: Different start and end periods across countries reflect differences in data availability. Averages of regional median incomes in a country are not population-weighted. Nominal regional median incomes deflated bu national CPI. Standard deviation is a measre of cross-region (within-country) variation, the higher the standard deviation, the higher the cross-region variation.
Source: For nominal median incomes: administrative data provided by national authorities; for CPI: OECD (2024)
Absolute differences in regional median incomes declined in some countries but grew in others (Chart 1). Five countries experienced a decrease in absolute income disparities, i.e. convergence (Austria, Finland, France, Portugal, Spain), while the other four showed a rise in income disparities, i.e. divergence (Czechia, Denmark, Latvia, Sweden). In Czechia and Finland, the magnitudes of the changes were particularly large, with standard deviation increasing by 20% between 2007 and 2022 in Czechia and decreasing by 36% between 2007 and 2022 in Finland.
A stronger picture of convergence emerges when focusing on relative differences in median incomes across regions. (5) Regional median incomes converged in seven of the nine countries analysed, with Denmark and Sweden the exceptions. For Czechia and Latvia, the convergence patterns differ depending on whether absolute or relative measures are used: absolute income dispersion increased, but to a lesser extent than the cross-regional average of median incomes, implying a decline in relative income dispersion.
Regional income convergence closely relates to differences in income growth between metropolitan (generally higher-income) and non-metropolitan (generally lower-income) regions. In countries where regional incomes converged over time, metropolitan regions experienced slower median income growth (on average) than non-metropolitan regions. In Austria, median income growth was comparatively slow in the major metropolitan regions of Graz, Innsbruck, Linz-Wels, Salzburg und Umgebung, and Vienna. Similarly, Portugal experienced slow growth in the Área Metropolitana de Lisboa, and Spain saw slow median income growth in Madrid. (6) In countries where regional incomes diverged, the opposite pattern often emerged. In Denmark, median regional income growth was highest in the City of Copenhagen and lowest in the non-metropolitan regions of Bornholm and Fyn. In Sweden, Stockholm was among the regions with the highest median income growth, while income growth was lowest in the non-metropolitan regions of Södermanland and Kronoberg.
Since 2014, median incomes in low-income regions have tended to catch up with high income regions in some countries but not others (Table 1). The strongest catch-up effects were recorded in Spain and Portugal, followed by Austria, Finland and France. Other countries showed no significant evidence of catching up. In a few countries, notably Latvia and Denmark, the number of observations is very small, reflecting the small number of regions, which may account for the lack of significant results. Despite these limitations, the analysis provides broad evidence in favour of a catching-up process of the lowest-income regions during the recovery from the 2007-2008 crisis, the COVID-19 pandemic, and the subsequent period.
Table 1
Low-income regions are catching up with high income regions in some countries, but not others
Beta convergence patterns and regression coefficients for regional median incomes, by country
Notes: Statistically significant logarithmic regressions coefficient in green, with coefficient in brackets. Initial period is 2014 for all countries except Portugal and Spain (lack of data for 2014). Final period is the latest year for which data are available, i.e. between 2020 and 2022. Nominal regional median incomes are deflated by national CPI. Observations are unweighted.
Source: For nominal median incomes: administrative data provided by national authorities; for CPI: OECD statistics available at https://stats.oecd.org/.
- 1. This section presents preliminary findings from OECD analysis of variation in incomes across small regions over time.
- 2. The administrative data on incomes collected by the OECD provides uniquely granular data across small (NUTS 3) regions. While the data are more granular and timelier than survey data on incomes, they are less harmonised across countries. The analysis thus focuses on within-country regional convergence patterns without attempting to provide cross-country comparisons. Administrative data on incomes within small regions cover nine Member States: Austria, Czechia. Denmark, Finland, France, Latvia, Portugal, Spain, Sweden. The same analysis was done with GDP per capita, as a robustness check, and results were broadly the same.
- 3. (European Commission, 2021a), (Königs, S. et al., (Forthcoming)).
- 4. Nominal incomes increased in Finland over the period, decreasing only in 2022. The decrease in real incomes is therefore purely the result of an increase in the Consumer Price Index (CPI).
- 5. Measured by coefficient of variation.
- 6. In Finland, where median incomes declined in real terms over the period, the decline was comparatively strong for all three metropolitan regions (Helsinki-Uusimaa, Pirkanmaa and Southwest Finland).