MATCHING AS AN ECONOMETRIC EVALUATION ESTIMATOR - EVIDENCE FROM EVALUATING A JOB-TRAINING PROGRAM

This paper considers whether it is possible to devise a nonexperimenta l procedure for evaluating a prototypical job training programme. Usin g rich nonexperimental data, we examine the performance of a two-stage evaluation methodology that (a) estimates the probability that a pers on participates in a programme and (b) uses the estimated probability in extensions of the classical method of matching. We decompose the co nventional measure of programme evaluation bias into several component s and find that bias due to selection on unobservables, commonly calle d selection bias in econometrics, is empirically less important than o ther components, although it is still a sizeable fraction of the estim ated programme impact. Matching methods applied to comparison groups l ocated in the same labour markets as participants and administered the same questionnaire eliminate much of the bias as conventionally measu red, but the remaining bias is a considerable fraction of experimental ly-determined programme impact estimates. We test and reject the ident ifying assumptions that justify the classical method of matching. We p resent a nonparametric conditional difference-in-differences extension of the method of matching that is consistent with the classical index -sufficient sample selection model and is not rejected by our tests of identifying assumptions. This estimator is effective in eliminating b ias, especially when it is due to temporally-invariant omitted variabl es.