15 Sep 2017

(by Mike Walker^) In his , Steve Davies bemoaned the lack of evidence on the magnitude of harm deterred by the activities of the Competition Agencies. He presented some estimates from research in CCP on cartel deterrence, concluding most strikingly: “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%.”

By happy coincidence, Steve’s blog anticipated the publication of an extensive new on the measurement of deterrence. Our survey, which also covers merger control and monopoly abuse as well as cartels, was motivated by the CMA’s recognition of the need for some hard evidence on this important dimension to our work. Since 2007 the UK competition authorities have worked within the context of a “positive impact” target imposed on them. Originally this target was that the direct benefits to consumers of competition policy enforcement should be at least five times greater than the cost of the authorities, measured over a rolling three-year period. This target was raised when the CMA was created and now stands at ten times costs. Over the first three years of its existence, the CMA achieved a direct impact figure of more than 18 times costs.

Putting to one side the interesting debate on whether such targets might be counterproductive (see of last year), the problem with measuring only the direct impact on consumers of our work is that it omits the benefits to consumers due to our work deterring anti-competitive behaviour. We should expect that the existence of a competent competition authority should deter some companies from engaging in anti-competitive behaviour that they would otherwise have engaged in. The result is that our estimates of our direct benefits to consumers will be an underestimate of our total benefits.

This failure to incorporate the deterrent effect has two implications. First, it is hard for government to know what is the return to funding competition policy and therefore how much to spend on it. Second, it is hard for the competition authority to make prioritisation decisions on which cases to investigate based on the expected impact on consumer welfare. If we thought that the omitted indirect benefits of competition policy were the same for all our tools, there would be no concern here. But we do not think that.

For instance, the direct benefits of our market investigations are typically large (£887m per year 2014-17) as they cover whole markets (e.g. retail energy, personal and SME banking). This is unlike other investigations which typically only deal directly with one or a few firms in a market. It is plausible that the deterrent effect across the rest of the sector is larger in the latter case than the former. So although the direct benefits of our merger regime (£143m per year) and competition enforcement work (£138m per year) are much smaller than for market investigations, we expect that the regime has a substantial impact of deterring anti-competitive mergers and behaviour. At the limit, a perfect regime that blocked all anti-competitive mergers should deter all such mergers being proposed in the first place. The same is true of the perfect cartel enforcement regime: it should deter all cartels. In these cases, the deterrent impact of the competition authority would be very large compared to its direct impact. If the regime really was perfect, the direct impact would actually be zero.

If we just use the direct benefits figures for prioritisation, we are likely to spend too much time on market investigations and potentially too little on other types of interventions, such as merger control and cartel enforcement. For these reasons we are in need of evidence on the size of these indirect effects Our literature survey does come up with some evidence, for example from two surveys commissioned previously by the OFT. These suggest that: for each cartel uncovered by the OFT, between 5 and 28 were deterred; between 4% and 18% of mergers are abandoned on the grounds that the authorities will find them anti-competitive, whilst a further 2-15% are restructured to avoid being found anti-competitive; and for each abuse of dominance decision, a further 4 to 10 abuses are deterred. Although these are reassuringly large numbers, they are not precise estimates, are relatively old and are based on impressionistic responses to survey questions rather than on actual observed behaviour.

Beyond this, there is reasonable evidence that cartel enforcement seems to lead to fewer cartels being formed, with those that are formed being less stable, shorter lived and less able to raise prices. Whilst the literature does not provide many estimates of these effects, as Steve Davies reported from the CCP’s own research, there are reasonable grounds for believing that good cartel enforcement might deter more than 50% of the potential harm from cartels. That is a much larger number than the direct impact of cartel policy.

There is also good evidence that a merger regime improves consumer welfare by deterring anti-competitive mergers, but little evidence on the magnitude of deterrence from incremental increases in spending on merger control. So, while we know that it is worth spending money on a merger control regime, we do not know what the optimal amount is.

Unfortunately, there is almost no literature on indirect effects outside of cartel enforcement and merger control. Given the amount of competition authority resources that are devoted to enforcement against the abuse of dominance and against various other types of anti-competitive agreements, this is a significant gap in the literature. Moreover, since much of the existing literature focuses on the US, it would be reassuring to have more specifically European evidence.

So, where does this leave us? We think three main conclusions flow from our literature survey.

(a) There are significant deterrent effects arising from competition policy enforcement. This is no surprise, but it is reassuring to have some evidence. It implies that our direct impact assessments are an under-estimate, and quite possibly a very significant under-estimate, of our overall effectiveness.

(b) However, in general, we do not have good estimates of the size of these deterrent effects, either in absolute or relative terms. This makes it hard to know where a competition authority should focus in order to have the most impact on consumer welfare.

(c) It would be great if we had more evidence on the deterrent effects of competition policy interventions. By its nature, it is hard to get such evidence on an unobservable variable (i.e. how much behaviour that would otherwise have happened was deterred). However, it is an important area for further research, as is now being conducted in CCP.

^ Mike Walker is Chief Economic Adviser to the CMA and a regular visitor to CCP

It also omits the benefits from competition policy incentivising increases in productivity. For further details, see “” (2015).

The figures in this paragraph are from Table 1 of the “”.

These figures come from “” (2007) and “” (2011).