Identifying Bid Leakage in Procurement Auctions

Abstract

We propose a novel machine-learning-based approach to detect bid leakage in first-price sealed-bid auctions. We extract and analyze the data on more than 1.4 million Russian procurement auctions between 2014 and 2018. As bid leakage in each particular auction is tacit, the direct classification is impossible. Instead, we reduce the problem of bid leakage detection to Positive-Unlabeled Classification. The key idea is to regard the losing participants as fair and the winners as possibly corrupted. This allows us to estimate the prior probability of bid leakage in the sample, as well as the posterior probability of bid leakage for each specific auction. We find that at least 16% of auctions are exposed to bid leakage. Bid leakage is more likely in auctions with a higher reserve price, lower number of bidders and lower price fall, and where the winning bid is received in the last hour before the deadline.

Detecting corruption in single-bidder auctions via positive-unlabelled learning

“…The idea of this work originated in a series of works on bid leakage in Russian procurement auctions [2,13,14]. Bid leakage is a form of corruption in a first-price, sealed-bid auction where the procurer reveals the contents of other firms' bids to a favoured firm, allowing the favoured firm to submit a marginally lower bid at the end of the auction and take the contract for the highest price possible.…”

“…The work of [2] is based on the assumption that in a fair auction the order of the bids should be independent of the winner; if it turns out that the last bidder is more likely to win than the others, there is reason to suspect bid-leakage has taken place. This was followed up by [13,14] who relaxed the independence assumption, as they demonstrated that in a game theoretic model there are legitimate reasons for a serious competitor to delay bidding -in particular, if a firm believes the procurer is corrupt and bid leakage could take place, by bidding near the deadline there would be no time to leak the bid to the favoured firm. Their approach was based on the DEDPUL positive-unlabelled classifier [12]: auctions where the last bidder did not win were labelled as "fair" (even if bid leakage did take place, it was not successful), and the classifier separated the remaining auctions into "fair" or "suspicious".…”

Detecting Indicators of Horizontal Collusion in Public Procurement with Machine Learning Methods

“…машинное обучение также вошло в основу подхода, представленного в работе (Ivanov, Nesterov, 2019), где анализируются данные о более чем 1,4 млн запросов котировок на российском рынке в период с 2014 по 2018 гг. авторы пытались выявить «утечку заявок» (bid leakage) -коррупционную схему, при которой поставщик незаконно предоставляет предпочитаемому участнику информацию о других заявках (Andreyanov et al, 2016).…”

“…Цель данной работы состояла в разработке метода для обнаружения признаков горизонтального сговора на торгах. с помощью методов машинного обучения Молчанова Глафира Олеговна, младший научный сотрудник, институт отраслевых рынков и инфраструктуры ранхигс при президенте рФ, москва, россия; ORCID 0000-0001-8130-1259; molchanova-go@ranepa.ru Рей Алексей Игоревич, к. э. н., заведующий лабораторией, институт отраслевых рынков и инфраструктуры ранхигс при президенте рФ, москва, россия; ORCID 0000-0001-8207-1790; rey-ai@ranepa.ru Шагаров Дмитрий Юрьевич, младший научный сотрудник, институт отраслевых рынков и инфраструктуры ранхигс при президенте рФ, москва, россия; ORCID 0000-0001-6544-5359; shagarov-dy@ranepa.ru 110 ЭНСР № 1 (88) 2020 Молчанова Г. О., Рей А. И., Шагаров Д. Ю. моделей, выявляющих картели (Huber, Imhof, 2019;Ivanov, Nesterov, 2019).…”

Auctions With Leaks About Early Bids: Analysis and Experimental Behavior

“…The data support the existence of widely prevalent leaks. Ivanov and Nesterov (2019) elaborated the analysis using machine learning methods to estimate that leaks occurred in 16% of over 600,000 auctions studied.…”