Essays in matching.

摘要

The first chapter of this dissertation uses Fisman, Iyengar, Kamenica, and Simonson's (2006, 2008) experimental speed-dating data to characterize people's preferences over dating partners. Preferences for absolute and relative attribute levels are compared under truth-telling and self-censoring equilibria. The model extends Fisman et al by incorporating incomplete information and rejection costs and applies a novel network based testing methodology. Our findings suggest that vertical preferences with self-censoring are a superior predictor of speed-dating behavior when compared to horizontal preferences or straightforward behavior. The second chapter is an attempt to reconcile the empirical literatures that estimate preferences from dating data and marriage data. Estimates from marital data rely upon the Becker's (1973) transferable utility framework whereas estimates from dating data use Gale and Shapley (1962) non-transferable utility framework. This chapter uses Becker's transferable utility framework to model Fisman et al's experimental speed-dating data. Two household production functions are tested. Our findings suggest that although the transferable utility model is a plausible explainer of the data, the specific nature of preferences may not be super-modular.;The final chapter introduces incomplete information into non-transferable utility, two-sided matching environments of the Gale and Shapley (1962) variety. It focuses upon ordinal matching mechanisms and considers restricted observability and repeated matching extensions. We find that some of the standard results (existence of stable allocations, one sided dominant strategy incentive compatibility) from the matching literature no longer hold.