| Popis: |
I. INTRODUCTION Procurement auctions are one of the most widely used mechanisms in modern economies to disburse funds for public works projects. A substantial portion of these funds are spent on the construction and maintenance of roads and highways, an activity on which the U.S. government, for instance, spends $ 160 billion annually. In this article, we examine the bidding behavior of experimental participants in order to shed light on the strategic incentives generated by these auctions. Our investigation is centered on two observations made in the existing literature. In a typical state-level procurement auction held in the United States, firms submit sealed bids for a particular project that has been advertised in advance by the state's Department of Transportation. Subject to some qualifications, the lowest bidder wins the project and is paid the amount of its bid. Empirical work analyzing these auctions has found that construction firms suffer from capacity constraints, whereby a firm with a larger backlog of preexisting work commitments finds it more costly to add new work to its roster (De Silva, Dunne, and Kosmopoulou 2003; Jofre-Bonet and Pesendorfer 2003). (1) Bidders with larger backlogs therefore tend to submit higher bids and win fewer auctions. Because a typical department of transportation auctions off hundreds of such projects a year, capacity constraints create an intertemporal linkage between adjacent auctions. A bidder that wins a given auction incurs an opportunity cost in terms of forgoing expected profits in future auctions on account of its anticipated higher costs. Our first research question therefore is: how does the presence of capacity constraints affect the bidding behavior of experiment participants? Our second research question is inspired by the fact that procurement markets have been found to be vulnerable to collusion by bidders (Bajari and Ye 2003; Porter and Zona 1993). In most states, contracts are typically won by a small set of firms that interact repeatedly in procurement auctions over the course of the year. This can provide an ideal setting for a bid-rigging scheme. In procurement auctions, where firms compete by setting prices, observing the prices set by one's rivals can be crucial to the feasibility of any collusive scheme (Porter 2005). If bidders are not colluding--and if past bids are not correlated with current costs--then observing a rival's past bids will not lead to observable variation in a firm's bids. If however firms are attempting to collude, a change in the bid revelation policy of the procurer might lead to a measurable change in bids. Thus, revealing the past bids of players can influence their strategic behavior as it alters the ways in which they can provide information to each other about their strategies. Even in the absence of explicit collusion, when all bids are revealed, a firm may be able to signal its desire to tacitly collude by submitting noncompetitive bids. (2) This leads us to the question: what is the effect of varying the amount of information revealed about past bids on the current bids of experimental participants competing in dynamic auctions? The answer to this question may be useful for policy makers given that currently most state departments of transportation in the United States publicize all the bids from past auctions. The central contribution of our article is to answer the two questions posed with evidence collected from laboratory economic experiments. In the experimental sessions, bidders play one-shot, two-round, and infinite-horizon procurement auction games, where winning an auction can increase a bidder's cost in the next auction. In half of the experimental sessions only the winning bid is revealed after each auction. This is done to restrict the amount of information that is shared between bidders. Explicit communication is not allowed. In the other half of the sessions, we reveal winning and losing bids at the conclusion of each auction. … |
