In the first part of this article, we described how French public transport company Keolis was awarded a 2-year emergency contract by the Dutch government, allowing them to keep providing services despite their fraudulent behaviour in the tender procedure. We showed that without proper punishment, this could incentivise fraud in future public tenders. In this second part, we will explain how the dynamics of a next tender process might also have some undesirable outcomes.
The government awarded Keolis the emergency concession partly because it did not want to engage in capital destruction. The original order Keolis had placed with BYD and Ginaf was therefore not cancelled, and Keolis currently owns the required electrical bus fleet. If the new tender is issued without any additional regulation on the existing bus fleet, Keolis would have an enormous competitive advantage over their competitors.
A new-generation electric bus from a European manufacturer costs approximately 450 thousand Euros, according to industry insiders. Assuming that Chinese products are roughly 30% cheaper (per industry insider assessment), the value of the Keolis fleet is roughly 80 million Euros ((259 x 450,000) x 70%), whereas a European fleet would be valued at approximately 115 million Euros. On top of that, the value of the Keolis fleet is already depreciated for two years. If we assume that the bus fleet will be depreciated over the course of the 10-year contracting period, this then indicates roughly 16 million Euros depreciation over two years. The other competitors would still have to make the entire capital investment to acquire a fleet.
Given the fact that Keolis deemed the initial side letters necessary to win the previous tender, we can assume that bus production prices are a decisive factor in this awarding. Therefore, it seems the only way other competitors would be able to beat Keolis in the upcoming tender is to negotiate prices with their subcontractors on the bus fleet that are equal to or lower than the fraudulent prices of Keolis minus two years depreciation. If we use the numbers portrayed above, Keolis currently owns a fleet valued at 64 million Euros where competitors at best can acquire a new fleet for 80 million Euros, but more likely in the range of 100-110 million Euros. It will be near impossible for competitors to compensate for this 15-45 million Euros difference in other parts of the cost function, which is why Keolis is all but guaranteed to win the business if they are allowed to compete. Other competitors will know this, and therefore have no incentive to place an aggressive bid. In the worst-case scenario, they will even refrain from participation if the cost of effort of preparation outweighs the extremely minimal chance of winning the business. Due to this lack of competition, Keolis will win the tender and probably even at a premium compared to the previous tender, as they will come to realise their favourable position. Eventually this premium could total, or even exceed, the value of two years’ depreciation on the fleet investment.
Silver lining in the process
Luckily, the initial tender documentation included a pre-condition that the winner of the tender is obliged to sell the existing fleet against the current book value to the next concession holder when the contract ends. Originally this requirement was adopted to allow concession holders to depreciate the costs of zero-emission vehicles over a longer period, which is supposed to make the investment in eco-friendly material more attractive. However, it turns out to be a useful mechanism to eliminate the competitive advantage of Keolis, as all tender participants can compete with the same capital costs. It also allows for a continuation of the current contract at a similar price level in case Keolis is excluded from the next tender. Unfortunately, this also means that regardless of their participation in, and the outcome of, the next tender, Keolis stands to see their full investment and depreciation costs of the fleet reimbursed. The province stated that Keolis will not be able to make a profit on the emergency concession as they are not able to depreciate their registration and implementation costs. However, they are obviously still allowed to cover their fixed costs and retain their current business, which gives them a significantly better payoff than losing the business.
In conclusion, because of their fraudulent behaviour, Keolis is able to make a two-year profit at the expense of society without any punishment. Worse still, the process creates a dangerous incentive for future public tender participants to commit fraud.
This time, the buy-out clause included for another reason ‘accidentally’ limited the damage. Just imagine the perverse incentives if this fallback scenario was not agreed upon beforehand.
How can we safeguard fair competition without rewarding fraud?
The difficulty of the current situation is two-fold. The government wants to set up a tender process that delivers maximum benefit to the community out of the available funds. Simultaneously, they do not want Keolis to profit from their fraudulent activities. Whether Keolis is invited for the upcoming tender or not, the latter outcome will materialise in the absence of a proper solution.
One way to potentially still achieve a fair and optimal process outcome is by using a game theory method known as ‘mechanism design’. Mechanism design is all about designing processes that incentivise desired behaviour. As imposing a fine is not an option, the government would have to find a more creative solution to let Keolis compensate for the welfare loss caused by their actions. One example could be to utilise Keolis’ eagerness to participate in the upcoming tender to incentivise paying an ‘entry fee’. This entry fee is simulated by giving Keolis a pre-condition for participation: if they do not win, they are obligated to sell their fleet to the winner of the tender at an additional discount (below book value). Because their competitors now have lower capital costs, Keolis can only win if they compensate the difference through a lower bid.
Regardless of whether Keolis wins or not, the government spend on the contract is decreased by the discount on the bus fleet. This can be considered as an additional benefit to society to compensate for the initial welfare loss.
As the concession makes up a quarter of their annual revenue, Keolis will likely prefer to be included in the tender process rather than refrain from participating. However, this depends on Keolis’ assessment of their competitiveness, including the stipulated discount on the fleet. Ideally, the risk that Keolis goes for the second option is hedged by decreasing the incentive for not participating. One way to do so is to exclude Keolis for all upcoming tenders in a certain time frame. As the government would give up competition, this initially does not seem to be in their own interest. In game theory, this is referred to as a non-credible threat. To make the threat credible, the government would need to formally commit to these repercussions prior to Keolis’ decision. This would show Keolis that the government is committed to whatever decision Keolis takes and allows them to choose the option that is most to their benefit, which should then be to still participate and agree to the stipulated rules. While these rules are of course to comply with EU tender regulations, they are likely to do so providing they are communicated well in advance and govern a fair process. Competing companies are likely to recognise these rules to be fair, if not even favourable for them, as Keolis’ participation would allow them a fair chance to compete with dramatically reduced risks related to investment. In a situation where Keolis does not compete, the situation would be similar to the initial tender but with reduced competition. So, other ‘players’ are not likely to appeal this design of the tender process.
Redesign the process to secure the optimal outcome
In conclusion, as it stands today, it looks like committing fraud in a large tender pays off and could even be perceived as a ‘dominant strategy’ from a game-theoretical standpoint. This is as a result of the desired behaviour not being incentivised through the rules of the process. This, in turn, could have the very undesirable side effect of signalling this dominant strategy to participants of other public tenders, risking more fraud to occur in the future.
Therefore, the government should consider these implications and take the necessary steps to re-design the process in such a manner that they end up securing an outcome that best serves the public and which, at the same time, does not reward fraudulent behaviour.