A German policy experiment from 2012 allows me to identify the short-run and longrun causal effects of decentralization on the placement efficiency of public employment services (PES). I exploit variation over time and across districts with different types of PES in a difference-in-differences...