We use (donut) regression discontinuity design and difference-in-differences estimators to estimate the impact of a one-shot hiring subsidy targeted at low-educated unemployed youths during the Great Recession recovery in Belgium. The subsidy increases job-finding in the private sector by 10 percentage points within one year of unemployment. Six years later, high school graduates accumulated 2.8 quarters more private employment. However, because they substitute private for public and self-employment, overall employment does not increase but is still better paid. For high school dropouts, no persistent gains emerge. Moreover, the neighboring attraction pole of Luxembourg induces a complete deadweight near the border.