처치집단에 대한 평균 처치 효과의 준모수적 추정

Abstract

In this study, Abadie(2005)’s semiparametric difference-in-difference(DID) and propensity score matching DID methods are implemented to estimate the average treatment effect on the treated(ATET) of the Anti-Graft Act on the expenditure on congratulations and condolences(ECC). Both approaches are semiparametric since they do not require a specific functional form specification of the outcome variable. By following Abadie(2005), the ATET can be estimated by using observed outcomes at before and after policy intervention, treatment indicator and propensity score of all observations. On the other hand, the propensity score matching DID is a very intuitive method to estimate ATET since it can find a counterfactual control for each treated unit. Abadie’s semiparametric DID estimate is -0.06 while the propensity score matching estimate is -0.12. However, the statistical significance is found only in the latter estimation result. It implies that the treatment group are expected to reduce the expenditure on congratulations and condolences by 12.8 percent less than the control group if the control group would be exposed to the treatment.