Collusion in public procurement auctions: evidence from Russia

0584716 - NHU-C 2024 RIV CZ eng D - Thesis
Polishchuk, Valentina
Collusion in public procurement auctions: evidence from Russia.
Prague: Charles University, CERGE, 2023. 83 s.
R&D Projects: GA MŠMT(CZ) SVV260742
Institutional support: Cooperatio-COOP
Keywords : collusion * simultaneous bidding * procurement auctions
OECD category: Applied Economics, Econometrics
https://www.cerge-ei.cz/pdf/master_theses/Master_Thesis_Polishchuk.pdf

This paper documents collusion between firms using micro-level data on 4.4 million firstprice sealed-bid procurement auctions conducted in Russia in 2011 − 2017. The data contains unique information on the timestamps of all bids and the bidding data itself. This study is one of the first to use bid timing to design a method for detecting collusion between firms based on a simultaneous bidding pattern: bidders place bids simultaneously or within a small time interval. The method performs well and identifies at least 7−25% of winner — runner-up bid pairs as collusive in validation subsamples: the pharmaceutical industry, known for its propensity to collusion in Russia, and three cartels formed by pharmaceutical firms. In the main data, the share of collusive winner — runner-up bid pairs varies between 8% and 23%. For a more general case that considers the pairs of each bidder with four other auction participants closest in the rank price, the share of collusive bid pairs is around 13%. In both cases, the share of collusive bid pairs is the highest in two-bidder auctions and gradually declines as the number of bidders increases. Collusive firms tend to place bids simultaneously more frequently when a few bidders participate in auctions, because of higher chances of manipulating auction outcomes. I also document that higher contract prices and smaller differences between winner and runner-up bids characterize bid pairs submitted simultaneously. Controlling for industry, public body, and region fixed effects, collusion increases final contract prices by 10.9% on average and makes simultaneous bids up to 50% closer to each other. Collusion affected contract prices totaling 1.49 billion U.S. dollars over 6.5 years. If I interpret these estimates as causal, eliminating collusion would have saved around 162 million U.S. dollars.
Permanent Link: https://hdl.handle.net/11104/0352569