Quality of government in European regions: do spatial spillovers matter?

ABSTRACT

This paper examines the role played by spatial spillovers in shaping the regional distribution of quality of government across the European Union. To do so, it constructs a hybrid spatial weights matrix combining geographical, technological and social distances between the European regions. The results reveal that the quality of government in neighbouring regions has a positive and statistically significant effect on one region’s quality of government, which highlights the relevance of spatial effects in this context. This finding is robust to the inclusion in the analysis of different variables that may affect regional governance. Likewise, the observed effect of neighbouring regions does not depend on the specific dimension of governance considered, the spatial weights matrix used to describe the spatial linkages between the European regions, or the econometric specification employed to capture the nature of spatial spillovers. The results also show that policy innovations related to governance spread from regions with high and intermediate levels of quality of government.

KEYWORDS:

• quality of government
• spatial spillovers
• hybrid spatial weights matrix
• regions
• European Union

JEL:

• H11
• R11

ACKNOWLEDGEMENTS

The authors are grateful to the editor and two anonymous referees for their useful comments and suggestions made on earlier versions of the paper.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1. Following Ward and John (2013) and Bubbico et al. (2017), we focus here on the role played in this context by competitive learning. Nevertheless, the literature has identified other mechanisms that may shape the spatial distribution of quality of government, including coercion or socialization (e.g., Acemoglu, Cantoni, Johnson, & Robinson, 2009). For further details on this issue, see the overview of the literature on policy diffusion by Graham, Volden, and Shipan (2013). Moreover, neighbouring institutions may influence the economic performance of a region, which in turn may affect the quality of government in that region (Bosker & Garretsen, 2008).

2. If the competitive pressure is not sufficiently high, regional decision-makers can also learn from the rent-seeking benefits of bad governance, which would lead to the emergence of clusters of regions with low levels of quality of government.

3. The EQI is available for all EU countries at NUTS-2 regional level with the exception of Belgium, Germany, Greece, Sweden and the UK, for which the data are provided at NUTS-1 level.

4. See Charron et al. (2014, 2015) for further methodological details on the construction of the EQI.

5. Given the nature of the present study, we exclude from the analysis the French overseas departments and territories, the Spanish and Portuguese islands in the Atlantic, and Cyprus because of their geographical remoteness.

6. This average is calculated using the governance scores shown by the 15-nearest neighbours of the region in question.

7. The sectors are: agriculture; industry; construction; distribution, communications and transport; financial and business services; and non-market services. The data cover the period 2005–09 and were drawn from Cambridge Econometrics.

8. Data on these indicators were taken from the 2008 wave of the European Values Study (EVS). In particular, we consider: attention to environmental issues (code v295); cheating on tax (code v234); importance of politics (code v5); belonging to welfare organizations (code v10); voluntary activities (code v25); and voluntary work (code v43).

9. Table A2 in Appendix A in the supplemental data online presents the full definitions and sources of all the control variables used in the paper.

10. In order to minimize any potential problem of reverse causality, all the time-varying covariates enter in the model as their respective means over the period 2005–09.

11. Tables A3 and A4 in Appendix A in the supplemental data online show that the results of the paper remain qualitatively unaffected if we alternatively employ the generalized spatial two-stage least squares (GS2SLS) estimator or a quasi-maximum likelihood (QML) procedure.

Additional information

Funding

This research benefited from the financial support of the Spanish Ministry of Science, Innovation and Universities [project numbers ECO2015-64330-P, ECO-2016-76681-R and PGC2018-093542-B-I00].