Essays on Procurement Auctions

This thesis empirically investigates important phenomena surrounding procurement auctions. The first chapter examines the effect of uncertain design evaluations on firms' behaviour in public procurement auctions in which firms compete on both price and design proposal. The second chapter takes a closer look at the source of variation in design evaluations. The last chapter then examines the costs and benefits of two widely used auction formats in procurement of construction projects. In chapter 1, we investigate firms' competition over price and product design in the context of Design-Build (DB) auctions. In a DB auction, firms' design proposals are independently evaluated by reviewers of the government. Reviewers' evaluations contain uncertainty from a bidder's point of view, leading luck to curtail differences in firms' chances of winning. The estimated model predicts that uncertain evaluations exacerbate auctioneer's uncertainty in auction outcomes. An alternative mechanism that shuts down the impact of uncertain evaluations on bidders' behaviour would mitigate the auctioneer's uncertainty in the price and the design quality by 22% and 33% , respectively.In chapter 2, we study strategic interactions among expert reviewers in evaluating a set of proposals. As expert reviewers may be concerned about their own career, they may not truthfully report the signal of proposal quality. We model and estimate the effects of such an incentive on reviewers' behaviour using a sample of public procurement auctions, so called Design-Build (DB) auctions, where potential contractors compete on price and design quality. The DB auctioneer assigns a set of expert reviewers to independently evaluate each design proposal submitted by the potential contractors. The challenge in identifying the strategic effects comes from unobserved heterogeneity in design quality. We circumvent the confounding issue by exploiting the exclusion restriction that peer's observed characteristics are independent of one's quality signals. We find that (i) reviewers have incentive to conform to the experienced reviewer's score, and also that (ii) reviewers bias his/her evaluation of favourite design upward. The empirical results suggest a need for careful mechanism design that takes into account of experts' incentives in various competitive environments where the outcome relies on experts' opinions.In chapter 3, we empirically investigate the costs and benefits of two widely used procurement auction formats: Lump-Sum Auction (LSA) and Unit-Price Auction (UPA). In an LSA, each bidder submits a single price for an entire project, and receives its price bid upon completion of the project if there is no contractual change. In a UPA, each bidder submits a price for each construction component of a project, and the final pay to the contractor depends on quantity change on contracted components during the construction phase. The existing literature only provides partial predictions about the performance differences between the two auction formats. In order to deal with the procurer's selection of auction formats, we exploit exogenous variation in (i) procurer's capacity constraints, and (ii) expected weather disturbances conditional on its realization. We estimate that LSA would reduce the final pay to the contractor by more than 19% relative to UPA for the projects with medium level of project risk. We find no evidence of differences in the degree of cost overruns/underruns between LSA and UPA projects. The empirical results indicate a potential for saving in infrastructure expenditure, and suggests a need for a model that takes into account for both ex-ante and ex-post bidders' behaviour in both types of auction formats.Ph.D