The cash divide: the allocation of European Union regional grants

Abstract

To promote economic and social cohesion, the European Union (EU) structural funds part-finance public investment programmes in European regions with about €30 billion per year. This article develops an explanation for the apportionment of structural funds across EU regions. It is argued that the Commission's decisions on regional transfer levels reflect its bureaucratic interest and potentially undermine EU goals. Using a new data set on regional transfer payments in the EU-15 from 2000 to 2006, and qualitative interviews with decision-makers, this argument is tested and corroborated. In doing so, it is shown that the recipient regions' level of economic affluence is necessary, but no sufficient explanatory factor for regional transfer levels. In contrast to previous findings in the literature, the empirical record does not suggest that regional partisan politics has an effect on the size of regional transfer levels.

Keywords:

• Budget
• EU executive politics
• European Commission
• fiscal transfers
• structural funds

ACKNOWLEDGEMENTS

I would like to thank Thomas Bräuninger, Thomas Conzelmann, Tanja Dannwolf, Fabio Franchino, Slava Mikhaylov, Berthold Rittberger and three anonymous reviewers for their comments on earlier versions of this paper. I also acknowledge the financial support of the German Academic Exchange Foundation (DAAD).

Notes

Owing to the sensitivity of the information, all interviewees prefer to stay anonymous. The numbers in brackets are the identification numbers for the different interview partners.

The classification of NUTS regions is made on the basis of population figures. I use the classification according to Council Regulation No 1059/2003.

The codings of the tax-spending dimension were used since the codings on the less specifically defined left–right dimension are not available for France. When running the analyses by using the left–right dimension, I did not obtain significantly different results.

Between the time period in which the elections were held (1998–2000) and the time period in which the expert survey was conducted (2002), some parties have split or merged. In case parties have competed separately in elections before the experts estimated them, I use this estimate for both parties. In case they have split, I take the average of the expert rankings for both parties. Only those parties that have been ranked by the experts are included in the data.

It may be objected that the sample size on the regional level, which varies from 11 (Molise in Italy) to 743 (Southern and Eastern in Ireland), may be too small for some regions. However, the samples are representative at NUTS 2 level, since they result from a random selection of sampling points within each of the NUTS 2 regions (Codebook of the Eurobarometer 52.0).

The data set that contains the above discussed variables is available on http://www.lisadellmuth.net/.

The problem that OLS estimates are downward biased in case of censored data may not occur when there are only few zero-value observations (cf. Greene 1981). As only six regions, that is less than 4 per cent of all observations, do not receive structural funding, I use OLS.