| dc.description.abstract | This dissertation consists of three essays on applied econometrics.
The first essay is entitled A Structural Analysis on US Spectrum Auctions. The spectrum auction allocates spectrum licenses to companies. This paper provides a structural analysis on US spectrum auctions to estimate bidders’ values, which is essential for auction policy evaluations. I first perform a theoretical analysis, then construct a structural model to rationalize bidders’ bidding behaviors as a bundle choice problem. I propose a multiple-step estimation to recover the parameters in bidders’ value function from the model. In the estimation, I develop a framework to handle the high-dimensionality issue in the bundle choice model with individual-level data. This paper analyzes the 1995-1996 spectrum auction in the US. I find evidence of complementarity in this auction, as well as heterogeneity in the complementarity valuation across bidders.
The second essay is entitled Optimal Model Averaging of Mixed-Data Kernel-Weighted Spline Regressions, and it is coauthored with Qi Li and Jeffrey S. Racine. Model averaging has a rich history dating from its use for combining forecasts from time-series models and presents a compelling alternative to model selection methods. We propose a model average procedure defined over categorical regression splines. Theoretical underpinnings are provided, finite-sample performance is evaluated, and an empirical illustration reveals that the method is capable of outperforming its nonparametric peers in applied settings.
The third essay is entitled Multivariate Density Forecast Combination. Density forecasts are able to convey the uncertainty in addition to the point forecasts, and multivariate density forecasts further allow people to capture the interdependency among different variables of interest. This paper develops a class of combination schemes for multivariate density forecasts, in view of that the forecast combination could effectively improve the forecast performance upon single forecasts. I prove the asymptotic optimality of the estimated combination weight. Monte-Carlo simulations are provided to demonstrate the theoretical results. | en |
