Many countries provide extra resources to schools serving disadvantaged pupils. We exploit a discontinuity in the assignment of such personnel subsidies in Flanders to estimate the impact on cognitive outcomes via a regression discontinuity (RD) design. Because bias can be substantial in RD...