Quantile Regression with Unobserved Heterogeneity: Advancing Policy Evaluation and Asset Pricing

The central theme of this thesis is developing new methods for estimating and forecasting the distributional structure for a set of high-dimensional economics and financial data. This thesis advances the literature by extending the quantile random coefficient models and quantile varying-coefficient models to panel data. We adopt a quantile latent factor structure to flexibly capture the unobserved serial and cross-sectional heterogeneities embedded in the panel data. Applying these methods, we contribute to the empirical literature on policy evaluation and asset pricing.