I proceed in five steps. First, I discuss the research design and the test strategy, in particular the cabinet design, the control variables, and the fixed-effects model. Second, the main effects of the ideology indicators representing different explanations – arguments based on group representation as represented by party labels and the framing argument by programmatic positions – are assessed as regards the two explananda: the generosity and the conditionality of unemployment insurance. This section also includes regressions with alternative dependent variables for generosity and conditionality as a rough first measure for the robustness of the analysis. Moreover, the section discusses the extent to which single cabinets and countries drive the results. In a third step, I investigate whether the results are robust when alternative and complementary aspects such as opposition ideology are considered. Fourth, the specific framing argument, which implies that the effects of (economic) pressures are conditional on government ideology, is investigated via interaction analyses. Finally, the results of the analysis are summarized with regard to the competing hypotheses H1, H2, H3, H4, H5, H4, H5b as well as the three metaexpectations.

Research Design: Case Selection, Data, and Model Specification

Case Selection: “Ideologically Consistent Quasi Cabinets”

The approach underlying the selection of countries is conventional and leads to a group of eighteen OECD countries, as in most of the reference studies assembled in table 3.1. The general rationale of a most similar systems design, no matter whether in a small-n or large-n context, is to reduce external variance by choosing a country group as homogenous as possible with regard to factors outside of the theoretical framework (Jahn 2006: 234-235). The universe or population of countries for which the effect of government ideology on retrenchment policies is to be investigated consists of countries that are all highly developed industrial countries, full-fledged established democracies for the entire period under study (1971-2009), and mature welfare states; although the latter criterion does not further reduce the intersection of countries. The intersection of these criteria (e.g., according to the country groups assembled in Jahn 2006: 225-227) leaves us with Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, the Netherlands, Norway, Sweden, Switzerland, the United Kingdom, Australia, Canada, New Zealand, the United States, and Japan.