A Retrospective On Merger Retrospectives In The United States

Over the last decade, merger retrospectives have become increasingly popular in the economic literature. However, it is far from clear what implications can be drawn from these analyses, because the results suffer from a sample selection problem that undermines their implications. This article uses models of the Federal Trade Commission's enforcement activity to estimate challenge probabilities and address the selection issue. Although the small size of the sample limits the statistical interpretation of the results, general observations are possible. First, significant price effects generally appear in markets with very high challenge probabilities. When challenge probabilities are moderate, say 30 to 70 percent, retrospective results generate price effects in roughly half of the studies. For markets with low challenge probabilities, no price effects are observed in the retrospectives related to collusion theories. Positive price effects are observed for most of the retrospectives in differentiated goods markets, but theoretical analyses imply that most those results are highly problematic. Although retrospectives, interpreted in light of the likely competitive concern, can serve as a test for merger policy, an anomaly between the retrospective and the relevant policy prediction requires some type of case study to be undertaken to resolve the stark difference in implications prior to the analysis. More complete case studies are needed to fulfill the true promise of the retrospective research regime.