Agenda

Titolo completo: Estimating Average Treatment Effects in Evaluation Studies: Using Dirichlet Process Mixtures

Relatore: Zizhong Yan, University of Warwick

Abstract: This paper focuses on the estimation of the average treatment effect on the treated (ATT) in evaluation studies under unconfoundedness. As an alternative to traditional matching and reweighting methods, I propose a constrained Dirichlet process mixture of normals (DPMN) model to consistently estimate the covariates distribution in the treatment group and match the control units to the treated so that the distributions of covariates are stochastically equivalent. I use this approach to build a matching estimator and a reweighting estimator with desirable properties. First, the DPMN matching estimator meets the balancing property by construction. Second, since DPMN yields consistent estimates of the propensity score, the reweighting estimator is semi-parametrically efficient. Traditional matching and propensity-score based methods are two-step approaches, which may result in incorrect standard errors. In this paper, the whole algorithm is integrated into a single efficient Markov Chain Monte Carlo scheme: the resulting marginal standard errors can account for errors arising from the first step estimation. I illustrate this new method with Monte Carlo experiments and an empirical application of the LaLonde(1986) data. The DPMN reweighting estimator is found to have a performance comparable to conventional reweighting estimators. I also find that the DPMN matching estimator is less biased and more efficient than traditional matching estimators, as a result of improved balance.