Estimation of Treatment Effects based on Kernel Matching

Abstract: The treatment effect represents the average causal impact or outcome difference between treatment and control groups. Treatment effects can be estimated through social experiments, regression models, matching estimators, and instrumental variables. In this paper, we introduce a novel kernel-matching estimator for treatment effect estimation. This method is particularly beneficial in observational studies where randomized control trials are not feasible, as it uses the full sample to increase the efficiency and robustness of treatment effect estimates. We demonstrate that the proposed estimator is consistent and asymptotically efficient under certain conditions. Through Monte Carlo simulations, we show that the estimator performs favorably against other estimators in the literature. Finally, we apply our method to data from the National Supported Work Demonstration to illustrate its practical application.