23 May 2011

(by Morten Hviid) Who should pay for the enforcement of competition law? The current UK consultation suggests that in the main it should be [some of] the users. When considering how to finance enforcement, however, it is important to remember that this delivers positive externalities.

Firstly, enforcement leads to deterrence effects and so implies that those involved in the case are not the only “beneficiaries” of the decision. Secondly, enforcement clarifies the law and how it will be applied, which benefits all firms which is also a general benefit. Where there are positive externalities, there is reason to subsidise users out of general taxation. This may include a levy on firms subject to competition law but there is no reason for cost recovery on any specific case. This general point applies to merger fees and appeals but the unintended consequences of cost recovery seem most problematic for antitrust investigations. The consultation proposes to make those who have violated competition law pay the costs of investigating the case. On the face of it, this sounds reasonable, especially if one were to add that when the competition authority is not able to prove their case, they pay the costs of firms against whom a case has been initiated. I detected no desire for such symmetry in the consultation document, but it may be hard to argue against it. If cost allocations are symmetric, there is a real risk that this is going to make the competition authority more risk averse, resulting in [even] fewer cases being pursued. This is a Pandora’s box which might better be left unopened.

The fundamental problem is that while the courts may consider the cost allocated and the fine as two very separate things with two very separate motivations, costs to fund the enforcement activity and fines to deter future behaviour, for the firm they are one and the same thing – money which has to be handed over. Hence an allocation of costs is to the firm not distinguishable from an increase of the fine. Two implications follow from this:

  • The fine is currently assessed to be able to deter. If fines and costs are treated as completely separate, adding costs will lead to over-deterrence in all but the simplest open-and-shut price-fixing cases. However, if costs are allowed for in the computation of fines, then since both fines and costs are supposed to go to the Consolidated Fund, nothing is gained. A question which should then be asked is whether the distortion through allocating costs is greater than the distortion from simply raising taxes. A second problem arises when it comes to the effect of the fine on the future viability of the firm. We have lately seen several cases where a bankruptcy discount has reduced fines even if the violation was price fixing. Will costs and fined be dealt with in the same way?

There is an [admittedly arbitrary] 10% of turnover cap on fines. There is a danger that appeals courts will see the cost allocation as a ‘too clever’ means of circumnavigating this cap. This is particularly so since there are invariably difficult issues with cost allocations, particularly as regards fixed costs. Other issues relate to incentives:

  • It is unclear how allocating costs will affect incentives to settle and whether this can lead to unfair settlements. The risk of racking up very large costs if the case is not terminated now may push a firm to settle even when it is convinced that it did not violate the .
  • A benefit of cost allocation is that it may serve as a punishment to firms who “drag out” the case by being obstructive.
  • It is unclear how the incentives of the competition authority to manage its costs in a reasonable manner would be affected. Equally unclear is the solution to the practical problem about how to allocate costs when there are multiple defendants. There may also be more appeals specifically over costs.

Let me finish on an irritation about an oft cited and rarely questioned statistic. The consultation claims that cartels have a 15% detection rate. This claim does not have strong empirical foundation. It basically arises because decisions in the US and EU find the average length of a cartel is 7 years. If detection rates are the same every year, this suggests the probability of being detected in any given year is one divided by seven, i.e. approximately 15%. The fallacy in this approach is that in many cartel decisions we do not have a precise estimate of the duration of the cartel, partly because in order to minimise appeals, authorities err on the side of caution and underestimate duration in order to secure a conviction. Furthermore, we only know of the duration of the cartels which are detected (i.e. the sample is biased).