Trends in local government in Europe

Abstract

The article explores which reform trends are expected to play a significant role in the further development of local government in European countries. Analysis is based on an adopted Delphi survey of (mainly academic) local government experts from thirty-two countries. Results include a list of trends structured by the perceived importance for the coming ten years. Redistributing responsibilities and duties across levels of government constitute the highest ratings out of all trends. Furthermore, results do not provide clear evidence to support a shift from public management to governance as has recently been proclaimed in academic debate and literature. Management topics continue to be of great importance. Last but not least, clusters extracted point to different reform groups, which highlight more reluctant or enthusiastic attitudes to the relevance of the suggested trends of the study.

Key words:

• Europe
• local governance
• local government
• new public management
• trend analysis

Acknowledgements

This article is part of a research programme at the University of St Gallen funded by the United Bank of Switzerland (UBS AG).

Notes

1 http://www.inpuma.net/IPMNQuery.html

2 Countries included in the study: Albania, Austria, Belarus, Belgium, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, Finland, France, Germany, Great Britain, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Malta, Moldova, Netherlands, Norway, Poland, Portugal, Romania, Russia, Slovenia, Spain, Sweden, Switzerland, Ukraine.

3 Country-specific responses relied on only one to two questionnaires, while results for single variables relied on the total n of thirty-two data sets. Replacement of missing variables with means would therefore have no impact on the mean of the variables, but it would strongly bias the country-related data sets. Looking at each data set as a country profile, replacing missing variables with the mean would generally lead to a value indicating either a slightly positive or negative valuation of the trend for the respective country. Since the data set includes only one set of values per country, it is important not to bias the country-specific valuations.

4 As an ordinal scale was used, the appropriate measure was Spearman's rho.

5 ** indicates that the correlation is significant at the 0.05 level (2-tailed), * indicates that the correlation is significant at the 0.1 level (2-tailed).

6 Analysis was conducted using all available variables, while K-means cluster analysis aiming for a higher number of clusters (more than two) results in at least one cluster containing only one or two cases.

7 Analysis to identify variables discriminating between the two groups revealed by cluster analysis leads to a discriminant function that predicts group membership by 100 per cent in the original group as well as in cross-validation. The function has a high canonical correlation (0.966) and a low Wilks' lambda value (Λ = 0.067, p = 0.000), which points to an extremely high discriminatory ability of the function.