Search Results
Title: Two Essays in Social Program Evaluation
Resulting in 1 citation.
| 1. |
Zhao, Zhong |
|
Two Essays in Social Program Evaluation Ph.D. Dissertation, The Johns Hopkins University, 2002. DAI-A 62/10, p. 3511, Apr 2002 Cohort(s): NLSY79 Publisher: UMI - University Microfilms, Bell and Howell Information and Learning Keyword(s): College Graduates; Educational Returns; Employment, In-School; Employment, Part-Time; Treatment Response: Monotone, Semimonotone, or Concave-monotone; Variables, Independent - Covariate This dissertation consists of two essays. The first essay studies using matching estimators to estimate models of treatment effects. The second essay is an empirical study of the impact of the Federal Work-Study program (FWS) on post college earnings. In the first essay, I compare propensity score matching methods to covariate matching estimators. I first clarify the misperception in the literature that propensity score matching estimators are less data hungry (require fewer observations) than covariate matching estimators. Next I propose a new matching metric incorporating the treatment response information. Then I examine the sensitivity of the matching estimate to the propensity score specification. Finally I use the National Supported Work Demonstration data set to demonstrate that, like any non-experimental estimator, the behavior of matching estimators crucially depends on the data structure at hand. In the second essay, I use the National Longitudinal Survey of Youth 1979 Cohort data set to study the effect of the FWS on post-college earnings. I estimate different types of treatment effects of the FWS, namely, the average treatment effect (ATE), the treatment effect on the treated (TT), and the marginal treatment effect (MTE). These effects are conceptually different and have different policy implications. I find that the FWS has a zero or insignificant negative impact on post-college earnings for the whole population. But for the Black population, the FWS has a positive effect on its participants. My conclusion is that the FWS coverage rate is too high, and that it would be better to divert some of the FWS resources to other financial aid programs. |
|
|
Bibliography Citation
Zhao, Zhong. Two Essays in Social Program Evaluation. Ph.D. Dissertation, The Johns Hopkins University, 2002. DAI-A 62/10, p. 3511, Apr 2002. |