| This dissertation is comprised of three essays on auction theory and empirics.;Chapter 1 studies the issue of identification in first- and second-price sealed-bid (hence forth, FPSB and SPSB) auctions with random reserve prices. Specifically, I examine the identification of the reserve price distribution, without imposing parametric assumption, when data on reserve prices is not available. Using data on seller's award/reject decisions, together with data on maximum bids and/or winning prices, I show that the reserve price distribution can be partially identified using ideas similar to Haile and Tamer (2003)'s. I demonstrate how one can construct informative bounds on the reserve price distribution in FPSB and SPSB auctions, and how exogenous variations in the set of bidders---in particular bidders' asymmetries and variations in the number of bidders---can vastly tighten up these bounds.;Chapter 2 provides a framework to study information disclosure policies in FPSB auctions empirically. In U.S. highway procurement auctions, the state's cost estimate of each contract is critical information but the policy on its disclosure varies across states. Using data from Oklahoma, where this estimate is not published prior to bid opening, I estimate a FPSB auction model with asymmetric bidders and correlated costs. The equilibrium under the counterfactual policy of full disclosure is then simulated to evaluate its effects. I find that information disclosure reduces expected procurement cost and improves overall efficiency in the Oklahoma market.;Chapter 3 examines bidder-seller collusion in an English auction setting with weak-and-strong bidders. With asymmetric bidders, the English auction does not implement the optimal auction of Myerson (1981), which is inefficient and discriminative. When the seller can collude with all weak bidders, however, I show that the optimal auction can be implemented by collusion. When the seller can only collude with a subset of weak bidders, I show that there still exists a simple and intuitive collusion mechanism that increases the sellers expected revenue. All these mechanisms are inefficient; they discriminate against strong bidders by increasing the likelihood that the colluding weak bidders would win, and reducing their payments when they win. |
