Detecting Auctioneer Corruption: Evidence from Russian Procurement Auctions
Abstract: This paper develops a novel method for detecting auctioneer corruption in first-price sealed-bid auctions. We study the leakage of bid information by the auctioneer to a preferred bidder. We construct a formal test for the presence of bid-leakage corruption and apply it to a novel data set of 4.3 million procurement auctions in Russia that occurred between 2011 and 2016. With bid leakage, the preferred bidder gathers information on other bids and waits until the end of the auction to place a bid. Such behavior creates an abnormal correlation between winning and being (chronologically) the last bidder. Informed by this fact, we build several measures of corruption. We document that more than 10% of the auctions were affected by bid leakage. Our results imply that the value of the contracts assigned through these auctions was $1.2 billion over the six-year study period. We build a model of bidding behavior to show that corruption exerts two effects on the expected prices of the contracts. The direct effect inflates the price of the contract. The indirect effect reduces the expected price, since honest bidders are trying to undercut corrupt bidders. We find both effects in the data, with the direct effect being more pronounced.