(by Bruce Lyons) The CMA has recently published its annual report and associated impact assessment. Its performance management framework commits the CMA “to achieving direct financial benefit to consumers of at least ten times our cost to the taxpayer.” [Annual Report 2015-16, p.66]. Target setting and performance measurement are an important part of performance management. However, the precise way that the government requires the CMA to justify its funding is dangerously distortionary.
The government requires the CMA to achieve a benefit to taxpayer cost ratio of at least £10 for each £1 of taxpayer cost (i.e. budget). This ratio is completely arbitrary and without empirical foundation. I am actually quite confident that this ratio would be comfortably met if all the benefits could be credibly measured – but they cannot. My concern is that in order to tie tangible benefits to tangible case work, ‘benefit’ is essentially measured by expected direct price savings from the remedies on actual cases completed (averaged over three years). This hugely underestimates benefits, in particular by ignoring deterrence. Think how dangerous it would be if military leaders had to justify their budget by measuring the value of wars fought, ignoring wars deterred.
After providing a bit of background, I identify four specific ways that the 10:1 target distorts the CMA’s incentives. The following table is taken from the CMA’s recent impact assessment:
|Table 1: Estimated average annual CMA/OFT/CC consumer savings and costs for 2013–16|
Area of CMA work
|Consumer protection enforcement||74.1|
|Market studies and market investigations||522.7|
10.6 : 1
It is striking that the benefits (or ‘savings’) associated with market inquiries are crucial for the CMA to meet its benefit/costs target. 76% of the measured benefits come from market studies and market investigations (which incidentally are not available to any other leading competition authority, at least with statutory remedy powers). In their absence, the benefit/costs ratio would be only 2.5:1. There would be no way to meet the 10:1 target unless staffing was catastrophically slashed.
The role and implementation of the CMA’s performance target
The enforcement of competition law is fundamental to ensuring an attractive and successful economy. For a taxpayer cost that works out at just £1 per head of population per annum, the CMA deters cartels, prevents big powerful firms from excluding innovative rivals, stops firms merging to monopoly, creates a culture that competitive markets are important and generally ensures that firms are broadly responsive to consumer needs. In the absence of this firm but relatively light-handed market regulation, we could see the ugly extremes of rampant monopoly capitalism or sclerotic, politically-determined price and investment controls.
It is, of course, important for a non-ministerial government department to justify its budget. This might be achieved, for example, by a quinquennial review that seriously addresses impact including deterrence, and emphasises ex post evaluations of actual (as distinct from predicted) impact. Such studies are, in fact, regularly carried by the CMA as part of its reflective learning (e.g. BAA airports) and by academics including my colleagues Steve Davies and Peter Ormosi. A quinquennial review could also take greater account of non-consumer benefits, innovation and deterrence, or more generally of the impact on the wider process of competition. The retrospective nature and resource intensiveness of these alternative approaches mean that they cannot be used to appraise an annual target.
Annual review requires a quick quantification in £m. Given the statutory requirement to do this annually and given the need to use predicted benefits as set out in the CMA’s case decisions, the current impact assessment methodology is probably as good as is feasible. Unfortunately, that credibility may make it particularly dangerous because the measured benefits are appraised against an inappropriate target.
If the target means anything, there would presumably be substantial budget cuts, and the debilitating loss of skilled staff, if the 10:1 ratio could not be met. The CMA as an institution is motivated by professional excellence aimed at promoting competition. If its budget is cut, it will have to ‘lose’ skilled staff and its ability to achieve excellence will be compromised. It, therefore, has a powerful intermediate objective to meet its benefit target in order to preserve the budget that supports the CMA’s staff as well as its long term aims. This creates serious institutional problems.
Four distortions created by the CMA’s performance target
First, antitrust enforcement activities (e.g. cartels, abuse of a dominant position) require proof of illegal behaviour which is often hard work for a competition authority. The first distortion is that these cases often involve only small markets with a modest effect on price. They contribute little towards the CMA’s measurable benefit target even though they can be incredibly important for deterrence. The resulting temptation is to divert resources towards markets work, even though firms in these markets are not suspected of having done anything illegal and there is less deterrence benefit.
Second, large markets capable of generating large measured benefits need to be investigated to replace those rotating out of the three year benefits average (which is used for the target). These markets must be ‘correctable’ by proposed remedies which can be quantified to provide an estimated £500m of price reductions (‘savings’) every year. Small markets contribute little to measurable target, even if they have serious problems. The second distortion is that the pressure is on to find large markets to investigate, even though they may be working reasonably well.
Third, the CMA is subject to an insidious pressure to find and measure substantial detriment in every market that it investigates. The ethic of professional excellence is strong, but this presumably applies to meeting institutional objectives as well as in relation to specific case analysis. Why is this a problem? Benefit measures are not uncontroversial and often contain a strong element of judgement. The recent trend has been for the CMA to try to measure detriment using prices and/or profitability. Prices must be compared with other benchmark prices or relevant costs. Individual prices often need adjusting for quality or personalised customer requirements. Profitability typically requires cost allocation and the valuation of physical and intangible assets. The CMA sometimes argues that firms operate inefficiently (i.e. with over-inflated costs and under-reported profitability) and makes adjustments accordingly. All these measurements and benchmarks require a high level of judgement and no single number should be expected to represent the truth. Sensitivity analysis is essential and further judgement goes into selecting which parameters to flex and by how much. Judgement is also required to identify the scenario that is reported as the ‘base case’. This is then used to: a) measure harm; b) justify remedies; and, here’s the rub c) quantify benefits for the 10:1 target. There is much room for judgement, and it is a serious concern that the choice of base case in individual cases might be influenced by the CMA’s performance management target.
Fourth, a large number for the ‘base case detriment’ creates pressure for ‘strong’ intervention (‘remedy’), even though an appropriate sensitivity analysis may caution against this. The pressure may come from within the CMA or from outside, including from politicians and lobbyists in the media (“Look how much consumers are being ripped off! Make the remedy hurt!”). Thus, remedy selection may be biased by both public perceptions of toughness and the institutional need for a remedy that facilitates immediate quantifiable results. For example, a remedy of facilitating entry by removing barriers and promoting customer engagement (with benefits emerging over time through a more effective competitive process) may be eschewed in favour of an immediately quantifiable ‘strong’ intervention (e.g. price control or divestiture) even if this has a longer term adverse effect on competition.
In conclusion, the CMA’s 10:1 performance management target creates insidious distortions. The target requires benefit measurement and the only practical methodology crucially excludes the value of deterrence. It shifts focus away from cartels and the abuse of dominance, and may help to explain why the CMA conducts so few such enforcement cases (e.g. compared to France). The target creates undue pressure for large scale market investigations at the expense of more socially beneficial case choice. It also incentivises biased measurement, possibly influencing decisions and promoting unsubtle interventions in investigated markets. Overall, the target dangerously undermines good competition policy enforcement and should be scrapped.
 Thanks to Steve Davies, Amelia Fletcher and Peter Ormosi for very helpful discussions on this topic, but this blog does not necessarily reflect their views.
 As the annual report says: “The CMA undertakes this assessment itself, with subsequent review by an external academic. Its methodology is based on that developed by the OFT, validated by successive independent academic reviewers and consistent with approaches now regarded by the OECD as international good practice.”
 Damien Neven and Hans Zenger have argued that even ex post evaluations need to be very carefully designed if they are not to trigger such over-interventionist behavioural responses in competition authorities.