Institutions and policy change

Institutions have gained renewed attention by political scientists over the last decade. Still, three of the most central questions remain unaddressed. This study systematically investigates these questions by employing statistical techniques based on a sample of sixteen advanced industrial democracies, a twenty year time period and eight policy areas. The first question asks "How much do institutions matter?" Institutions have tended to be treated as "all or nothing" entities--typified by the popular question "do institutions matter?" They are either important or unimportant irrespective of contextual factors. This study attempts to move away from this absolute approach by investigating how much institutions matter across policy environments? The second question asks, "in what ways do institutions matter?" The ways in which institutions matter have been assumed to flow from the neat logic of institutional design. Revised theories emphasize the complexity of empirical reality and "bi-directional" institutional tendencies. Which theory provides the best understanding of institutional affects across space, time and policy areas has not been systematically examined. Moreover, the affects of institutions are frequently assumed to be structurally-fixed. The results of this study indicate that this is clearly not the case. Both the importance of institutions and their policy affects vary depending on what policy makers are trying to achieve and the policy environment in which they are operating. The third question asks "which institutions matter?" Executives receive the bulk of scholarly attention. The results presented here suggest that this attention is warranted. The neglect of many other institutional characteristics, however, is not. These omissions have proved to be a severe limitation on our understanding of institutions in generalizable terms. Examining a narrow range of institutions leads to a miscalculation of how much institutions matter and the ways in which they matter. The limitations of examining institutions in isolation of each other is illustrated by the case of trade unions: When unions are included in bivariate analyses of government expenditures or multivariate models without additional institutional variables, they assume a statistical importance that disappear when a range of other institutions are incorporated.