Mergers and acquisitions are extraordinarily prevalent in the United States, generating massive expenditures every year. However, a serious empirical puzzle lies at the heart of all that activity. That empirical phenomenon’s most remarkable feature by far is that even though it is well established in an extensive literature and implies far-reaching policy consequences, American law ignores it entirely.

Generations of researchers have failed to find evidence that merger and acquisition activity generates any lasting benefits for the combining firms’ owners or anyone else. No one seriously doubts that efficiencies of scale or technological integration are real or that acquisitions sometimes achieve them. Still, the evidence strongly implies that they are mostly available in small deals among small firms. While the results are no longer very seriously contested, no one has any conclusive explanation for the several puzzles they pose.

This paper comprehensively reviews the empirical literature and works through the policy implications for the major bodies of law it affects. The most important insights are two. First, the evidence should undermine the confidence among scholars of corporation law in the regulatory self-sufficiency of market institutions, on which corporation theory depends so heavily. In particular, a major hope for controlling managerial agency cost—one of the theory’s chief preoccupations—remains the hope that the so-called “market for corporate control” governing hostile takeovers will discipline underperforming managers. The evidence discussed here suggests that the market produces no meaningful benefits at all. The second lesson is simpler. The other major legal regime governing mergers and acquisitions—antitrust—should return to the simpler, pro-enforcement presumptions by which it once more effectively limited market concentration. The fear of jeopardizing pro-competitive gains, on which antitrust merger law has been rendered largely inert, is no longer very defensible. The balance between false positive and false negative should be reset.