Sequential Matching Estimation of Dynamic Causal Models

This paper proposes sequential matching and inverse selection probability weighting to estimate dynamic casual effects. The sequential matching estimators extend simple, matching estimators based on propensity scores for static causal analysis that have been frequently applied in the evaluation literature. A Monte Carlo study shows that the suggested estimators perform well in small and medium seize samples. Based on the application of the sequential matching estimators to an empirical problem - an evaluation study of the Swiss active labour market policies - some implementational issues are discussed and results are provided.