Regional unemployment in the EU before and after the global crisis

Abstract

This article empirically assesses the evolution of European regions in terms of both employment and unemployment during the recent financial crisis and global recession. Our specific research questions are as follows: (i) has there been a reversal in employment and unemployment dynamics at a regional level during the crisis (2007–10) compared with the previous period (2004–07)? (ii) have the western regions (‘old’ EU states) behaved differently in response to the crisis compared with the eastern regions (NMS)? Finally, (iii) are the differences between the two groups of regions related to structural or institutional variables? After a review of the literature on the key determinants of regional unemployment, we summarise our main findings concerning the impact of the global crisis on the labour market. Our econometric investigation aims to answer the questions we pose. Structural characteristics are considered in terms of sector specialisation of regional economies. In addition, we consider certain institutional characteristics, by including indicators of the share of temporary workers and of long-term unemployed. Our analysis is then targeted at sub-samples of western and eastern European regions: we show that the critical factors for labour market performance during the crisis in these two groups differ greatly. From a methodological viewpoint, we exploit a spatial filtering technique which allowed us to greatly reduce any unobserved variable bias – a significant problem in cross-sectional models – by accounting for latent unobserved spatial patterns.

Acknowledgements

We would like to thank several participants in the EACES-AISRe Workshop on ‘The Regional Labour Market Impact of the Last Crisis’ (Turin, September 2011) for their useful comments, and in particular the discussants Maurizio Baussola and Chiara Mussida. In addition, this article has greatly benefited from the suggestions and criticicism of two anonymous reviewers.

Notes

 1. We use the Eurostat definition of ‘temporary worker’, without considering the different meanings that it has in different countries (indeed this depends on the specific institutions of labour markets, on the legal definition of work contracts, on the various ‘atypical contracts’ existing in many countries etc.). It should be noted that the concrete meaning of ‘permanent contracts’ is also quite different, in various countries, according to the degree of employment protection legislation (especially the dismissal rules).

 2. Persistence and evolution of regional differences in labour market performance have been widely studied also in recent literature (for a review see Perugini and Signorelli 2010).

 3. Wage rigidities, low labour mobility and specific labour market institutional factors make the effects of the (otherwise temporary) aggregate demand shocks more persistent (see Bentolila and Bertola 1990).

 4. See, for example, European Commission (EC 2009) and Eurostat (2011).

 5. The evolution of regional labour market performance in the ‘old’ EU countries and their determinants was previously investigated in Perugini and Signorelli (2007).

 6. Indeed, the divergence of regional unemployment and wage rates is also influenced by the migration patterns of workers, by skills and by the flow of foreign capital: more educated workers and FDI flow to regions with a higher concentration of educated workers (see again Jurajda and Terrell 2009).

 7. These are the two key determinants, since empirical investigations on regional unemployment differentials have employed a wide range of variables (for a survey see Elhorst 2003). Many explanatory variables are negatively correlated with regional unemployment, such as GDP per capita, industry concentration and participation rate; while other variables, such as the weight of young people in the population or the presence of amenities in the region, are positively correlated with it.

 8. For example, Marelli (2007) found – by using different specialisation indices – that, for aggregate economies, structural convergence was a widespread phenomenon across European countries and regions but that, within the industrial sector and market services, specialisation trends were more mixed (with concentration prevailing in some industrial activities).

 9. The growing integration of European regions, extending to many EU countries – the alleged existence of a core of regions located in northern Europe with uniform employment dynamics is not corroborated – has also been discovered in employment terms. Marelli (2006) analysed a large sample of NUTS-2 EU regions. He also found that national borders were not particularly significant in singling out clusters of regions with similar patterns of employment growth (in particular, manufacturing employment seems to be more intensively correlated across regions than aggregate employment).

10. The existence of endogenous mechanisms leading to ‘real’ convergence is one of the factors underlying the success or failure of the European Monetary Union (see also Marelli and Signorelli 2010a). On the other hand, the possibility that regional economies exhibit business cycles different from the national ones was originally stressed by Fatas (1997).

11. In order to explain employment growth in a large sample of NUTS-3 regions in eight transition countries, Marelli and Signorelli (2010b) included an index of ‘progress in transition’, calculated from the European Bank for Reconstruction and Development (EBRD) statistics. Their key finding was a negative effect of the ‘transition index’ on employment growth in a first period (1990–2000), which became positive in the years after 2000 (indeed, the initial privatisations and market reforms were accompanied by rationalisation and restructuring processes causing negative effects on employment). Another important result of their study is that regional divergence within countries coexists with convergence between countries. However, this evidence, although stronger for transition countries (in which the clustering of employment and economic activities in the capital city regions is especially evident), has also been shown for western EU regions (as already found in Marelli 2007).

12. Vamvakidis (2009) provides empirical evidence for the EU regions for the period 1980–2000.

13. Throughout the world, the financial crisis initially harmed the US, the UK, Ireland, Spain and smaller countries (Iceland, Greece, the Baltic states). However, the largest output (real GDP) reductions in 2009 were recorded – among the biggest countries – in Japan, Germany and Italy (GDP fall was around or above 5% in all three countries). This was a consequence of world trade contraction, affecting industrial and export-oriented countries more deeply.

14. As highlighted above, interpretations of the regional labour market effect of the crisis are still few and far between. To our knowledge, this article is one of the first that has tried to address this topic.

15. For instance, subsidies for part-time work in Germany or extending income support for workers formally maintaining job contracts at reduced working-time or ‘zero-hours’ in Italy.

16. Moreover, these lags are longer if the recession comes together with a financial crisis. It should also be noted that unemployment can still rise (for a period) even after employment growth has become positive.

17. The EC (Autumn 2011) forecasts for the year 2012–13 broadly reflect this expectation.

18. The employment rate – the key labour market performance indicator of the European Employment Strategy (EES) – declined in 2009 (at 64.6%) and 2010 (at 64.2%) in the EU-27, interrupting its previous continuous rise – towards the ‘Lisbon objective’ (70%) – which started with the launch of the EES in 1997 (EU-27 employment rate at 60.7%) and culminated in 2008 (65.9%).

19. Young people have been particularly affected by the employment crisis (they have weaker contracts and lower qualifications and experience), as investigated in Choudhry et al. (2012): long-term unemployment for young workers can be harmful and may result in ‘discouraged workers’ effects and social exclusion from the labour market.

20. The grouping of regions refers either to large countries or to groups of small countries.

21. For both variables, the worst performance was found not only in Spanish regions but also in the Irish and Baltic regions.

22. Indeed, for all EU regions, the CV decreased not only in the years preceding the crisis (2004–08) but also in the crisis year (2009). The CV was calculated for all 271 EU regions and for the regions in each country. We excluded one-region countries (Cyprus, Estonia, Latvia, Lithuania, Luxembourg, Malta), two-region countries (Ireland and Slovenia) and Denmark.

23. We use data from the second quarter (30 June) for both dependent and independent variables. Although LFS data were available for the third quarter of 2010, which would have allowed us to observe some slightly later post-crisis developments, the high number of missing values made it inconvenient to use them, and indicated using second quarter data.

24. Here we refer to the growth of unemployed people in relative terms (hence the increase appears huge even in regions where the initial number of unemployed was low), which can be seen as the percentage change in the representative individuals' chance of being unemployed. For the spatial distribution of regional unemployment rates in 2010 see Figure A3.