Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Based on administrative data on unemployed in Belgium, we estimate the labour market effects of three training programmes at various aggregation levels using Modified Causal Forests, a causal machine learning estimator. While all programmes have positive effects after the lock-in period, we find substantial heterogeneity in effectiveness across programmes and unemployed. Simulations show that “black-box” reassignment rules that respect capacity constraints on average, increase, respectively decrease, the time spent in employment, respectively unemployment, by more than one month within 30 months of programme start. A shallow policy tree delivers a simple rule that realizes about 85% of this gain.