Estimation and Inference of Distributional Partial Effects: Theory and Application

ABSTRACT

This article considers nonparametric and semiparametric estimation and inference of the effects of a covariate, either discrete or continuous, on the conditional distribution of a response outcome. It also proposes various uniform tests following estimation. This type of analysis is useful in situations where the econometrician or policy-maker is interested in knowing the effect of a variable or policy on the whole distribution of the response outcome conditional on covariates and is not willing to make parametric functional form assumptions. Monte Carlo experiments show that the proposed estimators and tests are well-behaved in small samples. The empirical section studies the effect of minimum wage hikes on household labor earnings. It is found that the minimum wage has a heterogenous impact on household earnings in the U.S. and that small hikes in the minimum wage are more effective in improving the household earnings distribution.

KEY WORDS:

• Distributional partial effects
• Minimum wage
• Nonparametric
• Semiparametric
• Stochastic dominance.

ACKNOWLEDGMENTS

This article is based on the first chapter of the author’s dissertation in the University of Texas at Austin. The author is deeply indebted to her advisor Stephen Donald for his support. The author thanks Jason Abrevaya, Maxwell Stinchcombe, Yu-chin Hsu, two anonymous referees, the associate editor, participants in the 2011 All California Econometrics Conference, and seminar participants in Emory University, University of California, Davis and University of California, Berkeley for many valuable suggestions. All errors are the author’s.