The effect of partisanship is disputed in the literature on welfare state retrenchment. The ‘new politics’ school argues that partisan conflicts are irrelevant to the understanding of retrenchment, but the second generation of retrenchment research concludes that such conflicts are still important. We engage in this debate by introducing a new empirical approach. Our method provides a necessary but currently missing link in the second generation of retrenchment studies which theorize on the input side of welfare state reform but conduct empirical studies on the output side. Our empirical approach entails a new type of data, compiled on the basis of content analysis of adopted laws, and we analyze the intentions pursued by incumbent governments in social policy-making. Based on an empirical study, we find partisan effects in programmes protecting against social risks that are disproportionally distributed among social strata.