Are Regional Elections really ‘Second-Order’ Elections?

Abstract

Schakel A. H. and Jeffery C. Are regional elections really ‘second-order’ elections, Regional Studies. This article critically assesses the applicability of the second-order election model to regional elections. It offers first a critique of the second-order election model, arguing that the model has imported an inappropriate ‘nationalizing’ bias into the study of regional elections. Second, the article conducts an empirical analysis that shows that second-order election predictions do not appear to hold for regional elections which take place (1) in political settings where regional elections do not have the potential to signal a future alternation of government at the national level; (2) in authoritative, powerful regions; and (3) in regions where non-state-wide parties compete in regional elections.

Schakel A. H. and Jeffery C. 区域选举真的是“次级” 选举吗？区域研究。本文批判性地评价次级选举模型之于区域选举的适用性。本文首先批评次级选举模型在研究区域选举时不适当地引进了“全国化”的偏见。再者，本文的经验分析显示，次级选举模型预测并不适用于下列情况中的区域选举：(1)在区域选举不具备做为未来国家层级政府轮替指标的政治环境；(2)在强盛的威权区域；(3)在有非国家层级政党参与区域型选举的区域。

区域选举 次级模型 方法论的国家主义

Schakel A. H. et Jeffery C. Les élections régionales, sont-elles vraiment des élections de second ordre?, Regional Studies. Cet article évalue d'un oeil critique l'applicabilité du modèle des élections de second ordre aux élections régionales. Dans un premier temps il fait la critique du modèle des élections de second ordre, affirmant que le modèle a importé dans l'étude des élections régionales un préjugé inapproprié en faveur du ‘nationalisme’. Dans un deuxième temps, l'article fait une analyse empirique qui montre que les prévisions électorales de second ordre ne semblent pas tenues pour ce qui concerne les élections régionales qui ont lieu (1) dans des milieux politiques où les élections régionales ne risquent pas de signaler une alternance future au plan national; (2) dans des régions autoritaires, puissantes; et (3) dans des régions où des partis politiques qui ne fonctionnent pas au niveau de l'État présentent des candidats aux élections régionales.

Élections régionales Modèle de second ordre Nationalisme méthodologique

Schakel A. H. und Jeffery C. Sind Regionalwahlen wirklich ‘zweitrangige’ Wahlen?, Regional Studies. In diesem Beitrag wird kritisch die Frage untersucht, ob sich das Modell der zweitrangigen Wahlen auf Regionalwahlen anwenden lässt. Zunächst wird das Modell der zweitrangigen Wahlen kritisiert und argumentiert, dass dieses Modell eine unangemessene ‘nationalisierende’ Verzerrung in das Studium der Regionalwahlen importiert hat. Anschließend wird in einer empirischen Analyse gezeigt, dass die Prognosen von zweitrangigen Wahlen nicht für Regionalwahlen zu gelten scheinen, die (1) unter politischen Bedingungen stattfinden, in denen Regionalwahlen nicht das Potenzial zur Signalisierung eines künftigen Regierungswechsels auf nationaler Ebene aufweisen, (2) in autoritären und mächtigen Regionen stattfinden und (3) in Regionen stattfinden, in denen nicht auf nationaler Ebene antretende Parteien kandidieren.

Regionalwahlen Modell der Zweitrangigkeit Methodologischer Nationalismus

Schakel A. H. y Jeffery C. Son las elecciones regionales realmente elecciones de ‘segundo orden’?, Regional Studies. En este artículo evaluamos desde un punto de vista crítico la aplicabilidad del modelo de elecciones de segundo orden en las elecciones regionales. Ofrecemos primero una revisión crítica del modelo de elecciones de segundo orden, sosteniendo que el modelo ha importado un sesgo inapropiado de ‘nacionalización’ en el estudio de las elecciones regionales. En segundo lugar, en el artículo realizamos un análisis empírico en el que demostramos que las predicciones de las elecciones de segundo orden no parecen ser válidas en elecciones regionales que tienen lugar (1) en entornos políticos en los que las elecciones regionales no tienen la posibilidad de indicar un futuro cambio de gobierno a nivel nacional; (2) en regiones con poder y autoridad; y (3) en regiones donde partidos de ámbito no nacional compiten en las elecciones regionales.

Elecciones regionales Modelo de segundo orden Nacionalismo metodológico

Keywords:

• Regional elections
• Second-order model
• Methodological nationalism

JEL classifications:

• H70
• H77

Acknowledgements

The authors would like to thank Jaap Woldendorp, Hans Keman, Ian Budge, and their research assistants Krista Hoekstra, Jelle Koedam and Tim Mickler for sharing their party government dataset. The authors are grateful to Liesbet Hooghe, Gary Marks and Michael Tatham for valuable comments. Arjan H. Schakel wishes to thank the British Academy for financial support through a Newton International Fellowship for 2009–2011. Earlier versions of this article were presented at the workshop ‘Towards a Regional Political Science’ held at the University of Edinburgh, 24–25 June 2010; at an Institute of Government staff seminar at the University of Edinburgh, November 2010; and at the workshop ‘Multilevel Governance: Estimating Authority Below and Above the State’, held in Berlin, 20–21 December 2010.

Notes

The authors recognize that the terminology of ‘region’ is contested, and that some of the places termed ‘regions’ in this article consider themselves ‘nations’. But the article conforms for simplicity's sake to standard comparative terminology which deems a ‘region’ to be a unit of government bigger than local government and smaller than state-wide government, and ‘national’ to be the description for government at the state-wide scale.

Though Wyn Jones and Scully (2006, p. 191, n. 8) do at least point to subsequent SOE work on EP elections.

An alternative operationalization would be to construct an expected vote share and use that as a benchmark for the regional vote share. The assumption is that the change in national support for a party is gradual, that is, that the change is the same per unit of time during the whole election period from the first to the second national election. A caveat of this method is that it fails to account for short-term forces such as party or party leader popularity or the traits of individual candidates (Gaines and Crombez, 2004, p. 296). Related to the expected vote method is the relative vote (Dinkel, 1977; Jeffery and Hough, 2001), which is calculated by taking the average of two general election vote shares. This method significantly reduces the number of cases since a party is included only if it participates in at least two general and one regional elections in succession. The analyses were replicated according to the different operationalizations, and the results remained robust.

Models whereby regions are clustered in countries and models using panel-corrected standard errors controlling for autocorrelation (Beck and Katz, 1995), with and without country dummies, yield by and large the same results.