Abstract

We study wage distribution in the regions of the European Union (EU) or NUTS-2 over the period 2000-2021. Geographic income inequality remains highly persistent in the EU and poses potentially negative consequences for both the equity and efficiency of the economic system. Our primary focus is on wage distribution because in-work income by far constitutes the predominant source of gross market income for households. Moreover, the increasing economic integration within the EU has resulted in a diminished capacity for individual countries to set labour policies, making the choice of region as the territorial unit of study more appropriate than the country level.
Methodologically, we rely on the decomposition of Theil's T statistic for the population, which comprises two components: the between-regions component (Tb) and the within-region component (Tw). Whereas Tb represents the sum of the contributions of each region to total wage inequality in the EU, Tw accounts for total EU regional inequality within the regions, attributed to productive sectors (differences in salaries across sectors). For this analysis, we use data on employee compensation (gross wages) and employment classified by 10 economic activity sectors (NACE Rev. 2) sourced from various Eurostat databases, covering 235 EU regions across the 27 Member States alongside 2000-2021.
The key outputs of this study include (1) the programming of Theil's T statistic decomposition in Stata, (2) the development of comprehensive graphical analyses to present our results, and (3) the estimation of regression models to explore the correlation between wage inequality and socioeconomic factors, using panel data estimators with 5,170 observations.
This approach enables the examination of spatial income inequality within a society by considering its constituent groups (regions and productive sectors), offering notable advantages.
First, we provide an open-access, homogeneous, and reliable dataset of gross wage inequality indicators that will facilitate the analysis of inequality trends at sub-national levels (regional level). To our knowledge, this information is not available in Eurostat or other well-known international datasets (OECD or World Bank). Second, the coverage in terms of countries, regions, and years (2000-2021) surpasses existing datasets, since the income Gini coefficient for Croatia is only available from 2010 onwards. Third, the uniformity of the method (Theil’s T decomposition) and its statistical properties yield values that are comparable over time and across regions.
Finally, it is worth noting that we provide information on income inequality within EU regions by considering the EU as a whole; that is, we compare each region with all other regions in the EU. To effectively model economic policies aimed at improving labour markets, reducing poverty, or mitigating economic inequality, it is crucial to have information at a disaggregated level by regions or smaller geographic units than countries. Our approach is suitable for making comparisons of economic inequality between EU regions and states within the United States.