(by Steve Davies) The time has come for us to stop ducking out of the big deterrence issue in competition policy – more precisely, the measurement thereof. This blog has been provoked by Bruce Lyons’s , in which he argues that the performance target placed on the CMA by government may have serious adverse consequences for the Authority’s incentives to undertake those investigations which generate relatively small measurable direct benefits, but potentially very large, unquantified, deterrent effects.

I agree with most of what Bruce says. In particular, the CMA’s target benefits/cost ratio of 10:1 is arbitrary. Where has it come from? What is the rationale for such a large number? Is the implication that Government should expect similar benefits/costs ratios in all area of public spending? If so, does this mean that the ratio of public expenditure to GDP should never exceed 1:10? In my opinion, ridiculous is more appropriate than arbitrary.

But where I disagree is in what to do about this target. Yes, of course, one might plead with the government to bring the target down, or, as Bruce advocates, use more ex-post evaluation and only quinquennial reviews. But my preference is to tackle the real problem. Much of the good work of any competition authority (CA) is in deterring anti-competitive behaviour in the first place, and if a CA is to be evaluated and set a target (a good thing), it is obvious that the benefits must include an allowance for the deterrent effect.

For so long competition economists have held back from attempting to measure the magnitude of deterrence, but we can’t carry on for ever wringing our hands, claiming that we can’t measure the unobserved. In fact, there are some indicative estimates emerging from my own work with Peter Ormosi in CCP, which are quite revealing. In on cartels we show that, even making the most conservative of assumptions, the magnitude of deterred harm is at least 5 times greater (and probably much more) than the harm observed in detected cartels.

Apply a 5:1 multiplier to the estimate of detected (i.e. observed) harm would of course require that the ratio of detected benefits to costs need only exceed 2:1. Market investigations might be exempt from such a multiplier if, as Bruce implies, they have no deterrent effect.

Either way, I am confident that, by explicitly introducing such a deterrence multiplier, the CMA would comfortably achieve its target – in my opinion, rightly so. But the annual impact evaluation should continue, give its important role, both internally and externally, as an aid in monitoring resource allocation.