Selection Bias in Evaluating Policy: The Case of Anti-Trust

Abstract

Evaluations of competition policy typically establish that consumer benefits from cases detected and rectified comfortably outweigh the costs of competition authorities (CA). However, such assessments are inevitably partial because they do not quantify the impact of deterrence, nor do they acknowledge that the CA does not root out all anti-competitive cases. This paper proposes a broader framework for policy evaluation which encompasses these unobserved impacts. Calibration is difficult because one cannot rely on estimates for cases which have been observed to make deductions about those that have not - an example of the sample selection problem which is endemic across much of the empirical Industrial Organisation literature. However, we show how economic theory can be used to at least narrow down the range to plausible bounds on the magnitudes of how much harm is deterred and how much remains undetected. It also establishes what are the key issues to be resolved in our future empirical research in order to narrow the ranges still further.

Keywords: anticompetitive harm, competition enforcement, impact assessment, selection bias, cartels, mergers

JEL Classification: K21, L44, H11