The Impact of Competition Policy: What are the Known Unknowns?

Evaluations of competition policy are increasingly common and typically establish that consumer benefits from detected cases easily outweigh the costs of competition authorities (CA). However, such assessments are often driven by data availability and only capture a small part of the total impact because they sidestep the difficult issue of how to evaluate deterrence. Similarly, they ignore the fact that policy does not root out all anti-competitive cases. This paper suggests a broader framework for evaluation which encompasses these unobserved impacts. Calibration is difficult precisely because we cannot rely on empirical observations on cases which have been observed to make deductions about cases which have not (because they are deterred or undetected). It thereby confronts the classic sample selection problem which is endemic in all studies based on data from CA decisions. Drawing on insights from economic theory, it argues that selection bias is likely to be substantial because the unobserved cases could well be those which are most harmful. If so, the deterrence of anti-competitive mergers may have a much greater positive impact, but the effects of non-detected cartels may be more serious than is usually supposed.