Based on summary statistics generated from his database, Kwoka concludes that domestic antitrust agencies are excessively tolerant in their merger enforcement; that merger remedies are ineffective at mitigating market power; and that merger enforcement has become increasingly lax over time.
As Federal Trade Commission economists, we appreciate Kwoka’s effort to distill for a broader audience the wide range of published analyses of antitrust enforcement actions. The FTC has a longstanding culture of reflection and self-assessment; indeed, current or former FTC Bureau of Economics (BE) staff authored many of the studies in Kwoka’s survey. The FTC recently released its second major assessment of its merger remedies; the first was released in 1999. Retrospective studies of consummated mergers have increased the FTC’s antitrust knowledge base, and have helped it improve enforcement accuracy by providing valuable insights into when merger policy has worked, and when it has not.
For example, after a series of unsuccessful hospital merger challenges, FTC Chairman Timothy Muris launched the Hospital Merger Retrospective Project in 2002, which contributed to the FTC’s recent successes in challenging harmful hospital mergers. More generally, as Orley Ashenfelter et al. have noted, prior to the publication of merger retrospectives in the academic literature, there was virtually no empirical evidence on the actual competitive effects of horizontal mergers. Retrospective analyses also help the agencies assess and improve the empirical tools used for prospective merger enforcement. We therefore regard retrospective analyses of policy decisions as a vital input into the refinement and improvement of antitrust merger policy.
While acknowledging many of these valuable contributions of retrospective analysis to merger policy, Kwoka’s principal focus is to use this literature to draw conclusions about contemporary federal merger policy. These include:
- “[W]e can conclude that recent merger control has not been sufficiently aggressive in challenging mergers.”
- “[T]he evidence also indicates that many challenged mergers are subject to remedies that fail to prevent post-merger price increases. Neither conduct nor structural remedies on average succeed in that objective, but of the two types of remedies, the conduct approach has shown itself to be particularly ineffective.”
- “It can therefore reasonably be concluded that . . . the numerical shift toward more accommodating policy across these decades reflects a bona fide policy change.”
Are these conclusions justified? In what follows, we provide a detailed examination of Kwoka’s conclusions, and the studies and methods on which they rely. For a variety of reasons, we find that his evidence cannot support such broad conclusions. For example, as we will show, several of the studies that form the basis for his criticisms of negotiated remedies actually provide no information about the effectiveness of those remedies. When those studies are removed from his analysis, the remaining evidence is weak and equivocal.
In addition, there are substantial methodological issues with his analysis. Kwoka does not employ well established meta-analytic techniques for computing average price effects and their associated standard errors. For example, he does not weight his observations by their estimated variances. The absence of weighting means that imprecise estimates of price effects receive the same weight as precisely estimated effects in the computation of his averages, which is a substantial departure from standard meta-analytic methodology. More importantly, he reports no standard errors for his estimated average price effects, which means that neither he nor his readers can conduct the hypothesis tests (e.g., is the average merger price effect statistically different from zero?) that normally are the focus of this type of inquiry. The absence of standard errors not only is a substantial departure from basic meta-analytic methods, but also is a substantial departure from normal econometric practice.
Last, we show that Kwoka’s analysis provides little support for the conclusion that enforcement standards have become weaker over time.
In what follows, we explore and discuss these issues in greater detail.
Continue reading the full text of this article in PDF format.
