B1228
Title: Estimation and inference for partially linear models with estimated outcomes using high-dimensional data
Authors: Jing Tao - University of Washington (United States) [presenting]
Abstract: Methods are given for estimating and conducting inference on partially linear models with estimated outcomes using high-dimensional data. Our new estimator allows for but does not require many more regressors than the number of observations for the parametric part. The first stage allows a general set of machine learning methods to be used to form the estimated outcomes. In the second stage, we derive the convergence rates for the linear parameters and the nonparametric function under a partially linear specification for the outcome equation, respectively. We also provide bias correction procedures to allow for valid pointwise and uniform inference for both the linear parameters and the nonparametric function. We evaluate the finite sample performance with simulation studies. Additionally, a real data analysis of the effect of the Fair Minimum Wage Act on the unemployment rate is performed as an illustration of our methods.