• A New Approach to Regression Analysis and Antitrust Litigation

    In an article to be published in the George Mason Law Review, Managing Principal Pierre Cremieux, Analysis Group affiliate Professor Edward Snyder, and attorney Ian Simmons examine the value of regression analysis in proving common impact in antitrust litigation.

    Pierre Y. Cremieux

    Managing Principal Pierre Cremieux

    Edward A. Snyder

    Analysis Group affiliate Edward Snyder

    In "Proof of Common Impact in Antitrust Litigation: The Value of Regression Analysis," the authors propose a systematic approach to test the appropriateness of regression analysis as a common method of proof. Such a test is particularly relevant because of the diversity of recent decisions, from the en banc decision in Dukes v. Wal-Mart to affirm class certification based in part on proposed regression analyses, to the Hydrogen Peroxide decision to overturn class certification because the mere promise that regression analysis could be done was considered insufficient.

    Regression methods may help isolate the potential effects of an alleged cartel from non-collusive factors. Of critical importance in antitrust litigation is the determination of the circumstances under which a regression analysis will yield a common proof, since this often determines whether a class will be allowed to proceed.

  • Economic theory and empirical evidence suggest that antitrust violations are likely to result in diverse ranges of impact, from none for some to significant impact for others, and thus underscore the importance of the distinction between proving impact and proving common impact.
  • The authors conduct two empirical inquiries using actual transaction data to illustrate the need for macro- and micro-commonality tests as necessary, although not sufficient, conditions for regression analysis to constitute a common method of proof in antitrust litigation. A macro-commonality test is an analysis of regression results to determine whether they are consistent (common) across broad groups within the class. Similarly, a micro-commonality test determines whether results are consistent (common) with the predominance of common issues.

    The authors conclude that passing the proposed macro- and micro-commonality tests is necessary, although not sufficient, for a proposed regression to constitute a common method of proof. ■

    Proof of Common Impact in Antitrust Litigation: The Value of Regression Analysis is available as a working paper on the website of the Stigler Center for the Study of the Economy and the State at the University of Chicago Booth School of Business.


  • This feature appeared in the Spring/Summer 2010 issue of Forum