Market Definition as a Problem of Statistical Inference

Market definition is conducted under conditions of notable uncertainty, due to conceptual ambiguity and model uncertainty. Statistical decision theory can help to explain and improve the market definition decision. Specifically, a Bayesian decision rule can assist analysts in defining markets by considering (1) the weight of evidence in favor and against substitutability (implying a ranking of substitutes), (2) prior probabilities determined by previous cases and research (setting a benchmark for inclusion in the market), and (3) error costs of incorrect inclusion or exclusion from the market. The article studies how employing such a decision rule would have improved the market definition exercise in a landmark South African merger case