Gordon C. Rausser1, Gareth J. Macartney2

(1. University of California, USA; 2. OnPoint Analytic, Inc., USA)

 

 

Abstract: Legal standards for class certification in class-action suits have risen steadily, as exemplified by the U.S. Supreme Court’s recent decisions to reject certification in Wal-Mart v. Dukes and Behrend et al. v. Comcast Corp. et al. The resulting higher standards present challenges for economic experts, who present complex econometric and statistical analysis to courts. Some experts have presented analyses that would not pass peer review (and thus have no place in court). Moreover, the new standards have exposed a tension between the economic and statistical methods that academic economists accept and the requirements of proof that arise in litigation. We document unacceptable methods of class-certification analysis and present methodologies consistent with modern economics. Sound hypotheses must be formulated, and rigorous empirical testing must be conducted. We validate the use of objective summary statistics, regression analyses, and reject the pervasive use of “data mining”. Further, analyses of liability and impact should be conducted prior to any implementation of a class-wide damages framework. Given the non-experimental nature of economic data, tests for commonality across the proposed class should be soundly motivated.

 

 

Key words: class certification; competition; econometric methods; hypothesis testing; statistical methods

 

JEL codes: K21, K41, C11, C12, C18