Explaining Coalitions : Evidence and Lessons From Studying Coalition Formation in Swedish Local Government

Abstract: The aim of this thesis is to better understand coalition formation. To date, a vast number of theories have been presented with the goal to explain coalition outcomes. For example, early coalition theories assumed that parties were motivated solely by the goal of attaining the power and prestige that come with being in office, and predicted that minimal winning coalitions would form. Later coalition theories are instead often based on the assumption that parties are interested in implementing a policy program.Three problems with much of the existing coalition research are identified and solved in this study. First, many coalition theories have only been evaluated on the data set of Western European national governments despite the fact that these theories often emerged from consideration of this same data set, which implies that these theories have not been subjected to proper evaluations. In this study, a new data set consisting of a large number of formation opportunities in Swedish local government is used to evaluate coalition theories. Second, few empirical investigations have focused on comparing alternative coalition explanations using careful controls. In this study several different statistical techniques are used to evaluate the relative importance of a number of important coalition variables. Third, coalition researchers have not focused on giving adequate causal explanations to coalition formation in terms of measuring both the causal effects and the causal mechanisms underlying these effects. In this thesis, a large-n statistical study is combined with in-depth case studies, which makes it possible to measure and isolate effects, and to investigate the mechanisms that explain these effects.The analyses performed here indicate that when explaining coalition formation we should take into account that parties are neither pure office-seekers nor pure policy-seekers. Instead, parties are driven by multiple goals. The results also indicate that we should consider that parties are concerned with the effects that coalitional choices may have on future election results. The results found here show that parties should not be treated as unitary actors, and that some parties may be less likely to be in government due to the fact that they use highly democratic decision-making procedures, or because they are highly factionalized. Another conclusion drawn here is that when explaining coalitions, we should consider that the history of interaction between parties matters and that parties are concerned with minimizing the transaction costs with forming a coalition. Many other theories are either backed up or discredited by the evidence.