Political agency models suggest that elected public officials choose different policies than appointed officials do. This paper is the first (a) to apply a clean empirical design to study whether the selection rule has a causal effect on public officials’ policy choices and (b) to investigate transmission channels. I exploit a unique setting in Germany, where a reform has created quasi-experimental variation in the selection rule for mayors. As the outcome variable, I use data on grant receipts for highly visible investment projects for which mayors must apply to the state government. Elected mayors attract 7% to 7.4% more grants in election years; for appointed mayors, there is no cycle. Using hand-collected data on mayor characteristics, I find suggestive evidence that although the selection of mayors changes following the reform studied, a likely reason for the observed cycle is that elected mayors have stronger electoral incentives.