Abstract
In a 2008 article, Per Pettersson-Lidbom uses regression discontinuity to test for Swedish party effects on municipal taxation, spending and employment. His study is based on the assumptions that local coalitions only include nationally represented parties and are stable over the studied time period, 1974-1994. We have instead used a full population data set of actual ruling coalitions, allowing us to relax these assumptions and to include 935 previously undefined cases in the sample. Thereby, we can remove a systematic bias towards excluding center-right coalitions. Using the same estimator as Pettersson-Lidbom we are largely unable to replicate the results in his study. Where Pettersson-Lidbom finds significant party effects, we have found either no, or only weakly significant results.