Dimas Fazio | Alminas Žaldokas

Kamikazes in Public Procurements: Bid-Rigging and Real Non-Market Outcomes

Mar 11, 2026

Key Takeaways

  • Research Question: This study examines whether collusive bidding in public procurement auctions translates into real harms beyond fiscal losses, focusing on public health and infrastructure safety outcomes.
  • Data and Method: Using 15 million Brazilian federal procurement auctions between 2005 and 2021, the authors identify a prevalent “kamikaze” strategy in which the lowest bidder intentionally withdraws after the auction, enabling a connected second-lowest bidder to win at an inflated price.
  • Findings:
    • Kamikaze bidding weakens competition and raises procurement prices by 10%–12%, primarily by deterring genuine bidders from competing aggressively. Observable coordination markers (shared ownership, common addresses, or repeated partnerships) are present in approximately two-thirds of kamikaze procurements.
    • These price distortions translate into severe non-market outcomes: hospital mortality rises by 19% following kamikaze auctions for essential medicines, and road-maintenance contracts awarded through kamikaze auctions require repairs nearly one year earlier than comparable roads, with road accidents due to poor maintenance increasing by 13.7% under kamikaze-awarded infrastructure contracts.
  • Implication: Procurement collusion imposes substantial social costs by degrading public service quality, whereas evidence from the 2014 transparency reform shows disclosure of bidder ownership can effectively curb coordinated bidding and overpricing.

Source Publication:

Dimas Fazio and Alminas Žaldokas, “Kamikazes in Public Procurements: Bid-Rigging and Real Non-Market Outcomes,” SSRN Working Paper.

Background and Research Focus

Public procurement accounts for roughly 12% of global GDP, making it a central channel through which governments deliver essential goods and services. Although a large literature documents cost inflation arising from bid-rigging, far less is known about whether such distortions extend to real, non-market outcomes that directly affect citizens’ well-being. This paper addresses that gap by asking whether strategic coordination in procurement auctions compromises the quality of public service provision, with a particular focus on healthcare outcomes and road safety.

Data and Empirical Approach

The analysis draws on the universe of federal procurement auctions conducted through Brazil’s ComprasNet platform between September 2005 and August 2021, covering approximately 15 million first-price electronic reverse auctions across 4,800 public institutions. In these auctions, firms compete in real time by sequentially lowering their bids, and the lowest bidder is expected to win conditional on post-auction compliance checks.

 

To identify collusive behavior, the authors focus on auctions in which the lowest bidder withdraws after the auction concludes, allowing the second-lowest bidder to be awarded the contract. The empirical strategy combines several complementary designs. A difference-in-differences framework exploits a 2014 reform that made shared ownership information visible to auctioneers, increasing transparency about bidder connections. A regression discontinuity design leverages the quasi-random assignment of second place caused by randomized auction closing times to test whether withdrawal decisions respond strategically to the identity of the runner-up.

 

To assess non-market outcomes, the procurement data are linked to administrative hospital mortality records from DataSUS and detailed road accident statistics from the Federal Highway Patrol, enabling a direct connection between procurement distortions and downstream public harms.

Findings and Interpretation

The study shows the “kamikaze” strategy is widespread, occurring in roughly 17% of all auctions. In these auctions, the lowest bid is, on average, about 15% lower than the final contract price, consistent with an intentional attempt to deter genuine competitors rather than reflecting an overly aggressive but sincere bidding strategy. Multiple indicators point to deliberate coordination: winning firms are significantly more likely to share owners or addresses with the kamikaze bidder, and repeated role-sharing across auctions is common. Observable markers of coordination are present in approximately two-thirds of kamikaze procurements.

 

The regression discontinuity analysis provides particularly compelling evidence of strategic intent. Kamikaze bidders are significantly more likely to withdraw precisely when a connected firm narrowly secures second place, but not when an unconnected firm does. This sharp behavioral response indicates information sharing and coordination rather than random post-auction failure.

 

These strategies materially weaken competition. Auctions involving kamikaze bidders attract fewer effective bids, exhibit lower bid dispersion, and result in procurement prices that are 9.7%–11.7% higher than comparable auctions for the same goods. Importantly, the price effects are substantially larger when the kamikaze and winning firms are observably connected, reinforcing the interpretation that collusion, rather than error or incompetence, drives the results.

 

The paper further finds these distortions generate severe real-world consequences. In healthcare, overpricing for standardized essential medicines does not directly affect drug quality, but it depletes hospital budgets, reducing subsequent purchases of both related and unrelated medical supplies. Hospitals exposed to kamikaze auctions experience a 19% increase in mortality rates for specific causes, with no evidence of differential pre-trends. In infrastructure, road-maintenance contracts awarded through kamikaze auctions require repairs nearly one year earlier and are associated with a 13.7% increase in accidents attributable to road conditions, with no corresponding rise in accidents attributable to driver behavior. These patterns point to the systematic selection of less efficient contractors and persistently lower service quality.

Figure 1: Excess Deaths in Public Hospitals and Number of Road Accidents

Note: This figure plots the dynamic effects of kamikaze versus non-kamikaze auctions, along with the corresponding 95% confidence intervals. Panel A shows the effects on hospital mortality rates for essential medicine purchases, and Panel B shows the effects on the number of road accidents due to road fault conditions (blue circles) and non-road fault accidents (red triangles) for road-repair contracts. In both panels, the x-axis denotes the year relative to event 0 (the 12 months after the occurrence of kamikaze). The y-axis indicates the changes in mortality rate (Panel A) and the number of accidents (Panel B)

Policy Implications

The findings demonstrate that bid-rigging in public procurement generates large and measurable social costs that extend far beyond government budgets. By directly linking collusive auction behavior to hospital mortality and traffic accidents, the study reframes procurement fraud as a public welfare issue rather than a narrow fiscal concern.

The evidence also points to feasible policy responses. The Brazilian transparency reform demonstrates that relatively low-cost disclosure of bidder ownership structures can significantly reduce collusion and overpricing. More broadly, the evidence suggests antitrust and procurement enforcement frameworks should explicitly account for non-market outcomes, such as public health and infrastructure safety, particularly in settings where contracting or verifying service quality is difficult.

Further Reading

Related working papers from SSRN

Shopping Basket