The deterrence value of competition policy can and should be measured

(by Steve Davies) In a post last year, I argued that it was time for competition economists, both academics and practitioners, to start seriously tackling one of the big unknowns: how much harm is deterred by Competition Law and the Competition Authorities (CAs)? In this blog I pull together some results from three recently completed papers on cartel deterrence. I believe that these importantly move forward our understanding of this great unknown, and merit exposure to a non-academic audience in a non-technical way.

We have three ‘headline’ results. First, our estimates confirm that the deterrence effect is considerable, and is likely to be quantitatively far more important than detection in the fight against cartels. On the most conservative of our estimates, more than half of all potential cartel harm never occurs, because it is deterred. This is very much a lower bound, and the proportion could be as high as 90%. Second, it is unlikely that those cartels which are detected and therefore observed are representative of those which are deterred. Deterrence appears to be greater for potential cartels which would otherwise set either relatively low overcharges or very high prices in the absence of the cartel offence. Third, it is dangerous to judge the success of a CA purely on the basis of the number of cartels it detects – the strong CA might detect only few, because its strength deters more cartels from forming in the first place.

Given that deterrence can never be directly observed because it refers to events that never occur, how do we come to these findings? Putting aside the niceties of our economic theory and econometrics, these papers are based on two statistical regularities that we have uncovered from close scrutiny of large databases already in the public domain.

The first comes from an historical comparison of the overcharges set by 500 legal and illegal cartels[1]. This reveals a significantly lower incidence of illegal cartels in the two tails of the distribution – when it is illegal to cartelise, relatively fewer cartels charge either very low or very high prices. We explain this theoretically in terms of deterrence. On the one hand, the threat of detection and penalty will likely outweigh any potential benefit for those ‘weak’ cartels which can at best only moderately raise prices. On the other hand, those cartels with the ability and market conditions favourable to high prices may be deterred from setting very high overcharges, for fear that it would increase the temptation for cartel members to run to the CA under a leniency programme, and/or the likelihood of consumer complaints and/or detection by the CA.

This intuition and result are developed in the second paper[2], and placed alongside theoretical insights and empirical results (e.g. on cartel detection) from the existing literature. We construct a framework capable of estimating the proportion of total potential cartel harm which is deterred, as opposed to undeterred and detected, or undeterred and undetected. This is operationalised using a simulation model, and making the most conservative of assumptions, we find that deterrence accounts for at least half of all potential harm. We interpret this very much as a lower bound estimate – in our simulations, the upper bound is higher than 90%.

One intriguing result from these simulations concerns the relative magnitudes of deterrence and detection. We find that a relatively “weak” competition authority, which deters few cartels and then detects a smaller proportion of those that are undeterred, may nevertheless detect a larger absolute number than will a “strong” CA which has much higher deterrence and detection rates. The reason for this superficially paradoxical result is that, although the strong CA is far better at detecting cartels where they exist, far fewer cartels exist in the first place because most have been deterred.

The dangers of judging CAs simply by the number of cartels they detect is pursued in a third paper.

This uses a second database, which is an international comparison of the number of cartels detected by over 30 CAs over a number of years, recognising that different CAs lie at different parts of their life cycles[3]. We find evidence that, as a CA builds expertise and experience and its enforcement activities feedback to the business community, this deters future cartel formation, which results in an inverted U shape time pattern in the number of detected cartels. Thus, over the first decades of its existence, a CA typically increases its detection rates, sometimes very rapidly, but gradually as successful enforcement begins to feedback to the business community and begins to deter, we observe a tailing off, and even a decline, in the number of cartels detected.

Of course, these results are preliminary, and inevitably somewhat speculative. But the subject is so important, and I believe our results are sufficiently plausible to deserve further testing and development by others as well as ourselves in the future.

It is now nearly two decades since the DoJ[4] acknowledged that “we firmly believe that deterrence is perhaps the single most important ultimate outcome of the [Antitrust] Division’s work. We are just as sure that it presents the most significant measurement challenges…”. Since then, there have been few serious attempts to quantify the deterrence effect. This silence is surprising given that practitioners are always under pressure from politicians to justify their existence by quantifying their beneficial impact on consumer welfare. For academic economists too, whether researching the nature of collusion, or assessing the motives and success of mergers, or the efficiency versus abuse trade-off in vertical practices, etc., it is surely unsatisfactory that so many of our stylised empirical facts are drawn from those cases reported by the CA, i.e. undeterred but detected, when we know little or nothing about those cases which are deterred and therefore never observed. We hope that we have made a useful first step in addressing these gaps.

[1] “The Deterrent Effect of Anti-Cartel Enforcement: A Tale of Two Tails”, with Bos, Davies, Harrington and Ormosi, 2017

[2] “Quantifying the deterrent effect of Anti-Cartel Enforcement”, Davies, Ormosi & Mariuzzo, 2017

[3] “Cartel enforcement and deterrence over the life of a Competition Authority”, with Armoogum, Davies & Mariuzzo, (2017).

[4] US Department of Justice (2000) “Antitrust Division Congressional Submission for Fiscal Year 2001”

One Response to The deterrence value of competition policy can and should be measured

  1. […] Mike Walker^) In his post last month, Steve Davies bemoaned the lack of evidence on the magnitude of harm deterred by the activities of […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: