Why Can’t We All Just Get Along: Structural Modeling and Natural Experiments in Merger Analysis

Abstract

Economists have two basic methodologies: structuralism, in which formal economic models control the analysis, and experimentalism, in which economic theory guides the analysis, but data from experiments determines the policy recommendation. The choice between the two approaches is often quite controversial, although each approach has its own strengths and limitations. Under either approach, the scientific method requires testing and, when choosing between models with comparable empirical support, verification. We describe how these approaches play out in merger analysis; where Cournot and Bertrand models of competition are the standard Structuralist tools. In contrast, Experimentalists generally search for some type of evidence of a relationship between structure and market performance. Which methodology is superior for merger analysis depends first on empirical testing of the relevant theories and then on how closely the facts of the market match the assumptions of the relevant economic theory and how much precision the analysis requires to generate a useful conclusion. We illustrate these tradeoffs with examples from three recently-litigated cases.

Keywords: Structural Models, Natural Experiments, Merger Policy, Falsification, Identification

JEL Classification: B41, K21, L40