Is Collusion-Proof Procurement Expensive?

TL;DR

This paper evaluates the additional costs of implementing a collusion-proof procurement mechanism that guarantees efficiency, finding that the costs are relatively modest and potentially justifiable to prevent collusion.

Contribution

It provides an empirical estimate of the rents and costs associated with a collusion-proof auction mechanism using real-world data.

Findings

Estimated rents range from 1.6% to 5%.

Total costs, including taxes, range from 2.08% to 6.5%.

Results suggest the mechanism's costs are manageable.

Abstract

Collusion among bidders adversely affects procurement cost and in some cases efficiency, and it seems collusion is more prevalent that we would like. Statistical methods of detecting collusion just using bid data, in a hope to deter future collusion, is perilous, and access to additional data is rare and often always after the fact. In this paper, we estimate the extra cost of implementing a new procurement rule proposed by Chen and Micali [2012] that is robust to collusion and always guarantees the efficient outcome. The rule requires bidders to report their coalition and to ensure incentive-compatibility, the mechanism allows them to attain rents. We estimate this rent using data from California highway construction and find it to be anywhere between 1.6% to 5%. Even after we factor in the marginal excess burden of taxes needed to finance these rents, the cost ranges between 2.08% and 6.5%, suggesting that there is a room to think about running this new auction, suggesting we should consider this auction.