Implications of dataset choice in comparative welfare state research

Abstract

In the debate on welfare politics replacement rates have been hailed as a measure far more appropriate to identify retrenchment than expenditure. However, this contribution demonstrates that there is a considerable dependent variable problem on the entitlement side as well. By means of a comparison between replacement rates for three major programmes from the two most prominent datasets, Scruggs's CWED and Korpi and Palme's SCIP, we show that not only diagnoses on the occurrence and intensity of welfare cutbacks vary considerably, but also the results on the determinants of welfare state change. After tracing some causes of the divergences, we conclude that future research needs to be more programme-specific, more data-conscious and more humble in its claims.