Regional Government Formation in Varying Multilevel Contexts: A Comparison of Eight European Countries

Abstract

Bäck H., Debus M., Müller J. and Bäck H. Regional government formation in varying multilevel contexts: a comparison of eight European countries, Regional Studies. Although governance in multilevel settings has become a prominent research field in political science, there are few comparative studies that focus on explaining sub-national coalition outcomes in such settings. This paper sets out to study regional government formation in eight European countries and it builds on a dataset that covers information on the policy preferences of parties drawn from regional election manifestos. The results show that parties at the regional level are likely to form congruent coalitions, that is, ‘copying’ the patterns of national government formation, and that they are more likely to do so in specific regional contexts.

Bäck H., Debus M., Müller J. and Bäck H. 各种多重层级脉络下的区域政府形成：欧洲八国的比较研究，区域研究。仅管多重层级环境中的治理已成为政治学中重要的研究领域，却鲜少有比较研究聚焦解释在此般环境中次国家层级结盟的结果。本文着手研究八个欧洲国家的区域政府形成，并以一个包含在区域选举宣言中各政党政策倾向资料的数据集为基础。研究结果显示，区域层级的政党倾向 “复制” 国家层级政府形成的模式，进行相同的结盟，并更有可能在特定的区域脉络中进行。

政府形成 结盟理论 多重层级系统 政策文件的内容分析 区域治权

Bäck H., Debus M., Müller J. et Bäck H. La formation de la gouvernance régionale dans divers contextes multiniveaux: une comparaison de huit pays européens, Regional Studies. Bien que la gouvernance multiniveaux soit devenue une filière importante de la science politique, rares sont les études comparatives qui visent principalement à expliquer les résultats des coalitions établies à l'échelon infranational dans de tels contextes. L'article cherche à étudier la formation de la gouvernance régionale dans huit pays européens et se fonde sur un ensemble de données qui capte des informations sur les préférences politiques des partis puisées dans leurs manifestes aux élections régionales. Les résultats laissent voir que les partis sont susceptibles sur le plan régional de constituer des coalitions harmonisées, c'est-à-dire ‘copier’ des modèles de gouvernance au niveau national, et qu'ils sont plus susceptibles de le faire dans des contextes régionaux spécifiques.

Formation de la gouvernance Théories sur la création des coalitions Systèmes multiniveaux Analyse du contenu des documents de politique générale Collectivité régionale

Bäck H., Debus M., Müller J. und Bäck H. Regionale Regierungsbildung in institutionell variierenden Mehrebenensystemen: ein Vergleich acht europäischer Staaten, Regional Studies. Obwohl das Regieren in Mehrebenensystemen mittlerweile ein zentrales Forschungsgebiet der Politikwissenschaft ist, liegen bislang nur wenige vergleichende Studien zur Regierungsbildung in diesem Kontext vor. Im vorliegenden Beitrag analysieren wir die Regierungsbildung in den Regionen acht europäischer Staaten. Dabei greifen wir auf einen Datensatz zurück, der die Policy-Positionen regionaler Parteien sowie die Eigenschaften der potentiell möglichen Regierungen umfasst. Unsere Ergebnisse zeigen, dass Parteien dazu neigen, über die Ebenen hinweg parteipolitisch kongruente Koalitionen zu bilden. Zudem haben die programmatischen Positionen der regionalen Parteien und die institutionell vorgegebenen Kompetenzen regionaler Regierungen einen Einfluss auf das Ergebnis des Regierungsbildungsprozesses.

Regierungsbildung Koalitionstheorien Mehrebenensysteme Inhaltsanalyse programmatischer Dokumente Regionale Autorität

Bäck H., Debus M., Müller J. y Bäck H. Formación de gobiernos regionales en diferentes contextos de varios niveles: una comparación entre ocho países europeos, Regional Studies. Aunque la gobernanza en entornos de varios niveles se ha convertido en un destacado campo de investigación en ciencias políticas, existen poco estudios comparativos sobre los resultados de una coalición subnacional en tales entornos. La finalidad de este artículo es estudiar la formación de gobiernos regionales en ocho países europeos a partir de un grupo de datos que abarquen la información sobre las preferencias políticas de los partidos extraídas de los programas electorales regionales. Los resultados indican que los partidos de ámbito regional tienden a formar coaliciones congruentes, es decir, ‘copiando’ los modelos de la formación del gobierno nacional, y suelen hacerlo en contextos regionales específicos.

Formación de gobierno Teorías de coalición Sistemas de varios niveles Análisis de contenido de documentos políticos Autoridad regional

Keywords:

• Government formation
• Coalition theories
• Multilevel systems
• Content analysis of policy documents
• Regional authority

JEL classifications:

• R
• R5
• R50

Acknowledgements

The authors would like to thank Daniela Beyer, Tristan Klingelhöfer, Peter Obert and Ines Wickenheiser for research assistance. Grant support is gratefully acknowledged from the Mannheim Centre for European Social Research, the University of Mannheim, and the Ministry for Science and Research of the state of Baden-Württemberg. In addition, the authors would like to thank Charlie Jeffery, Arjan Schakel, and the anonymous referees for their valuable comments and suggestions on earlier versions of this paper.

Notes

It should be made clear here that even though variation across countries is mainly presented in this case, the unit of analysis in the multivariate analysis is a so-called ‘formation opportunity’, which refers to situations after regional elections (that is, the variation within countries, across regions and across time is being studied).

The only missing manifesto in the present analysis is that of the Scottish Labour Party for the 1999 parliamentary elections.

Party competition in Germany and the Netherlands is mainly structured by an economic and a societal conflict line, which differentiates between progressive or libertarian policy preferences, on the one side, and conservative or authoritarian positions, on the other side. In Spain, Belgium and the UK, decentralization issues structure party competition additionally to the economic left–right conflict. Austrian, Czech and Swedish party competition is mainly determined by conflicts in economic policy only.

There is also a similar, positive and significant relationship between the left–right measure and positions generated using data collected within the framework of the Swiss Electoral Studies and a comprehensive survey among Swiss local parties conducted by several Swiss social scientists (see http://www.socio.ch/par/).

However, estimating positions using automated text analysis seems to be only reasonable if the chosen policy dimensions do reflect preferences on a number of related issues, rather than on individual controversies.

The elections held in Prague are not included, since the regional and municipal spheres are very hard to disentangle in the Czech capital.

Similarly, in Swedish regions (and municipalities) all or most parties are represented (on a proportional basis) in the formal executive, which suggests that no coalitions form in these systems. However, a majority party or coalition appoints committee leaders and full-time posts, which form a kind of informal executive, and previous research is followed that has characterized the coalitions that form when these posts are elected as ‘government coalitions’ (for example, Bäck, 2003). The data on the governing coalitions in the Swedish regions were provided by Bo Per Larsson at the Swedish Association of Local and Regional Authorities.

In the case of Germany, party positions from the expert survey by Laver and Hunt (1992) were also used in order to take account of the longer period under study.

In the case of Austria and the Czech Republic, this paper refers to the 2002 federal election manifestos. For the Netherlands and Belgium, this paper refers to the programmatic documents written before the 2003 elections. In Spain and Sweden, the reference points are the 2004 and 2002 national election manifestos, respectively. For the UK, this paper refers to the 2001 general election manifestos. A special case is Germany, since the reference points are the 1990 and 2002 federal elections.

In the conditional logit model, the probability that individual i chooses alternative j is:

where j = 1, 2, …, J for a total of J alternatives. In this specific application, i equals the system and j equals the potential government, or the choice (for example, Greene, 2000, p. 862; Long, 1997, pp. 178–186).

The conditional logit model assumes the independence of irrelevant alternatives (IIA). That is,

the odds of choosing one alternative over another do not depend on any other alternatives in the choice set or on the values of the covariates associated with those alternatives.

 (Martin and Stevenson, 2001, p. 39)

It is checked whether the IIA assumption is violated by applying the test procedure developed by Martin and Stevenson. The IIA assumption is violated if the IIA test value given in Table 3 is lower than 0.05, which is not the case in any regression model presented here.

This is because the conditional logit model cannot include features that do not vary across the choices, that is, the potential governments that could form after an election (Martin and Stevenson, 2001, 2010; Brambor et al., 2006; Long and Freese, 2006). This makes it impossible to include country or election dummies as fixed effects in the regression models. When, for instance, interacting ‘cross-cutting coalition’ (which varies across choices) with the degree of regional authority, the regional authority variable cannot be included by itself (which would be standard within an ordinary least squares (OLS) regression framework), since it varies only across countries or regions, but not across the choice alternatives (that is here, the potential coalitions).

The results of the conditional logit analyses remain stable even when single countries are excluded randomly.