Panel Data Quantile Regression for Treatment Effect Models

Document pages: 47 pages

Abstract: In this study, we develop a novel estimation method of the quantile treatmenteffects (QTE) under the rank invariance and rank stationarity assumptions.Ishihara (2020) explores identification of the nonseparable panel data modelunder these assumptions and propose a parametric estimation based on theminimum distance method. However, the minimum distance estimation using thisprocess is computationally demanding when the dimensionality of covariates islarge. To overcome this problem, we propose a two-step estimation method basedon the quantile regression and minimum distance method. We then showconsistency and asymptotic normality of our estimator. Monte Carlo studiesindicate that our estimator performs well in finite samples. Last, we presenttwo empirical illustrations, to estimate the distributional effects ofinsurance provision on household production and of TV watching on childcognitive development.