Order statistics and the forensic detection of collusion in public infrastructure projects

Collusion in public procurement is a pervasive and costly form of white-collar crime, particularly in infrastructure investment. Characterised by covert coordination among bidders, such schemes typically involve multilayered transactions and are often intertwined with corruption, making detection and prosecution exceptionally challenging. This paper examines a real-world case from Brazil in which forensic experts used order-statistic methods to identify anomalously low winning-bid discounts in public infrastructure tenders. When combined with evidence obtained through police investigations, these statistical anomalies supported successful legal action and the recovery of misappropriated public funds. Statistical modelling also played a central role in quantifying the economic harm by reconstructing competitive counterfactuals to estimate losses attributable to bid rigging. The case demonstrates the value of integrating forensic statistical techniques into criminal investigations to detect, quantify, and prosecute collusion in public procurement, offering a replicable model for other jurisdictions. By synthesising statistics, probability, and auction theory with an understanding of the behavioural, organisational, and institutional mechanisms that enable bid-rigging practices – and by balancing theoretical rigour with practical applicability – our paper demonstrates how collusion can be detected in practice. Collusion in public procurement is a hidden but costly problem that diverts public money away from essential services and infrastructure. This paper shows how statistical tools, specifically order-statistic methods, can help government agencies, investigators, and oversight bodies identify suspicious bidding patterns in real time. Drawing on a real case from Brazil, the study demonstrates how unusually low winning-bid discounts, combined with a limited number of genuine competitors, can provide strong signals of bid rigging. For practitioners, the paper offers a transparent and practical method for detecting collusion using data that procurement authorities already collect. This approach does not depend on confidential company information or complex “black box” algorithms. Instead, it applies clear, reproducible statistical logic that can be readily communicated to auditors, investigators, and the courts, meeting established legal standards for forensic evidence. For policy makers, the study provides a replicable model for strengthening procurement integrity. It shows how statistical screening can support early detection, guide investigative priorities, and generate credible estimates of financial loss—enabling governments to recover misappropriated funds and deter future misconduct. More broadly, the paper demonstrates how combining statistical forensics with traditional investigative practices can improve transparency, accountability, and value for money in the delivery of public infrastructure.