Treatment Choice Based on Semiparametric Evaluation Methods

Abstract:     
Most evaluation studies focus on estimating average treatment effects and largely neglect heterogeneous responses in the population. However, the effectiveness of a policy does not only depend on the programmes themselves but also on the selection of participants. Hence for improving overall effectiveness of the policy it would be important to know for each individual which is his optimal programme, not only which programmes are effective on average. In this paper a semiparametric approach to estimating optimal treatment choices is proposed and applied to Swedish rehabilitation programmes. It seems that a much higher re-employment rate could have been achieved through better participant allocation.

Number of Pages in PDF File: 36

Keywords: Optimal treatment choice, statistical decision rules, heterogeneous response, targeting, profiling, policy evaluation, heterogeneous treatment effect, causal effect, matching

JEL Classification: C44, C14, H43, J24, J60

working papers series