The proposed research consists of the development of econometric tools for the analysis of economic data using nonparametric and/or semiparametric methods in quantile models. Each of the three projects comprising the proposal improves on existing methodology in a different manner. The objective of the first project is to develop semiparametric methods that yield reliable inference in instrumental quantile models when there is potentially insufficient exogenous variation, a desideratum absent from classical methods. The second project seeks to produce tighter bounds in triangular (quantile) models with discrete endogenous regressors; tighter bounds lead to qualitatively more meaningful conclusions. The third and final project aims to produce a new semiparametric estimator for a general binary choice model which is easier to compute, for which confidence intervals are more straightforward to obtain, and which can be extended to allow for endogenous regressors. All three projects entail extensions of existing econometric methodology. The first and third projects involve discontinuity problems, which require novel applications of empirical process theory. The firstand second projects moreover require extensions of the asymptotic theory for semiparametric and nonparametric estimators. The results of all procedures will be established rigorously and their usefulness demonstrated by means of simulations and empirical exercises.
Broader Impact The proposed research will be submitted for publication to academic journals and presented at conferences and departmental seminars. It is of relevance in a wide range of empirical applications using cross-sectional data, especially in industrial organization and labor economics. Efforts will be made to persuade empiricists to use the methodology developed here. Graduate students will coauthor most papers arising out of this proposal.
