What was happening in French public transport in the 1990s that makes it fertile ground for studying collusion?
France’s public transport sector is not a single unified market. Each urban area runs its own transport network, and contracts are awarded through independent auctions. In normal competitive conditions, this should encourage operators to increase their efficiency. In practice, it led to abuse of market power as the same handful of dominant firms kept winning contracts.
In 2005, the French competition authority (FCA) found that the three major players – Keolis, Connex (now Veolia), and Transdev – met multiple times to coordinate bidding strategies between 1994 and 1999. The cartel conspired not to compete against incumbent operators, threatening retaliation against any firm that might disrupt their scheme.
In 2005, the French competition authority imposed fines for collusion on three public transport operators: €3.9 million for Keolis, €5.05 million for Connex (now Veolia), and €3 million for Transdev
The regulator feared collusion would lead to higher costs for local authorities. What did your empirical analysis reveal?
Surprisingly, our results show that the price distortions caused by collusion were quite limited. While the FCA was right to be concerned about the potential for inflated prices, our counterfactual analysis – comparing observed cartel prices to the “but for” prices that would have prevailed under competition – found only modest differences.
This suggests that the cartel’s primary aim was not to extract higher prices. Instead, the dominant motive appears to have been to avoid the substantial costs and risks associated with preparing tenders, bidding wars, and operating in unfamiliar markets.
Cartel firms were able to benefit by preparing fewer bid proposals, the cost of which could vary from €46,000 to €760,000
How did you deal with the complexity of high-stakes bidding strategies?
Thanks to a rich dataset covering 205 urban transport networks, this case provided a unique opportunity to empirically study the effects of collusion in a real-world bidding market. We used a structural model to represent the underlying decision-making process for transport firms. We assume they act strategically, weighing their costs, chances of winning, and market conditions.
By matching the model to our real-world data, we can then infer things we don’t directly observe, like how much firms invest in R&D. This approach is powerful because it allows us to simulate “what if” scenarios. For example, what if firms had innovated more? We can then move beyond surface-level observations and gain policy-relevant insights into deeper mechanisms, such as how risk aversion and the cost of preparing tenders influences firms’ willingness to compete.
How does collusion affect innovation?
In a competitive tendering system, firms have strong incentives to innovate in order to reduce their costs and submit more attractive bids. Collusion dampens these incentives. If firms know they can retain contracts without fear of being undercut by a rival’s innovation, they have less reason to invest in new technologies or processes.
Collusion dampens incentives to innovate. If firms know they can retain contracts without fear of being undercut by a rival's innovation, they have less reason to invest in new technologies or processes
Our model allowed us to simulate what would happen if firms increased their innovation efforts. The results were clear: it would have led to significantly lower prices, as innovation reduces a firm’s operating costs across all markets. So, while the direct price impact of collusion was limited, the indirect effect of reduced innovation was potentially much more significant.
What lessons can policymakers and industry partners gain from your findings?
Our research suggests that the main harm of collusion in bidding markets may not always be higher prices, but rather a reduction in innovation and efficiency. Policymakers are right to worry about cartels inflating prices, but they should also use rigorous analysis to pay close attention to other effects. Even when price impacts appear modest, collusion can erode the long-term vitality of markets by discouraging new entrants and reducing investment in R&D.
For industry partners, our findings highlight the importance of balancing risk management with competitive behavior. While risk aversion is natural, excessive coordination can stifle innovation that benefits both operators and the public.
What’s next for research in this area?
We see great potential for applying our structural model to other “natural experiments” in bidding markets, such as mergers. The French public transport industry has experienced several major mergers, and analyzing their effects on competition, costs, and innovation is a logical next step. More broadly, we hope our approach will inspire further empirical work on the dynamics of bidding markets in other sectors.
KEY TAKEAWAYS
• Empirical analysis shows collusion by French public transport operators (1994–1999) had little direct impact on costs for local authorities.
• Rather than jacking up prices, the cartels’ main motivation appears to have been to avoid risk and the cost of preparing bids.
• Collusion reduced incentives to innovate, which may have had a much larger long-term impact on prices and efficiency.
FURTHER READING
The research agenda of TSE Infrastructure & Network Center examines how market design and regulation affect network industries like transport, utilities, and digital services. ‘Collusion in Bidding Markets: The Case of the French Public Transport Industry’ is coauthored by Philippe Gagnepain. Other research by David is available to read on the TSE website.
Article published in TSE Reflect, July 2025
