Ottorino Chillemi, LAUREA, Dep.of Economics, UNIVERSITY OF PADUA, VIA DEL SANTO 33, Padova, 35123, Italy and Claudio Mezzetti, Ph.D, Dep.of Economics, UNIVERSITY OF WARWICK, Coventry CV4 7AL, Coventry, United Kingdom.
Abstract
The paper studies optimal procurement auctions for a principal who aims at minimizing the probability of non performance. In the first model, the probability of default and non-performance by a bidder is negatively correlated with the value of the bidder assets. In this model non-performance arises endogenously in a standard auction when completion costs turn out to be high, because the winning bidder has bid too low in the auction. In the second model non-performance occurs for reasons that are exogenous to the procurement procedure, but the probability of non-performance is a decreasing function of the bidder's cost (safer firms have higher costs). The main result of our analysis is that standard auctions perform poorly in our settings; the optimal procurement procedure is a lottery that randomly assigns the contract. This finding sheds some light on the regulations designed to prevent winners' default by cutting abnormally low tenders. These procedures have Nash equilibria in which all bids are equal, just as it occurs in our optimal procedure. Hence, they may be justified when the procurer overriding objective is project completion. In the case of an endogenous probability of default, the procurer can succeed in lowering the risk of default to zero. Contrary to what asserted by Engel et al. (2006), we show that the price of guaranteeing completion is not necessarily high. If the procurer can charge all bidders participation fees, then it can use bidders' competition to reduce its outlays from the mechanism. For a sufficiently large number of bidders, the procurer's outlays are no more than the project's expected cost of completion. If participation fees cannot be used, then the procurer's outlays are independent of the number of bidders. Whether they are large depends on the variance of completion costs; the higher the difference between maximum possible cost and expected cost, the higher the outlays. If the probability of default is exogenous, the optimal procurement procedure never allows assigning the contract to the bidder with the lowest probability of default (or to guarantee completion with probability one). In this case, auction formats that assign the contract to the lowest bidder after a test of reliability may perform well, but they do not necessarily select the most reliable contractor.