Damages Models Under Comcast

Fit, Not Precision: Damages Models at the Rule 23(b)(3) Gate

The anchor holding treats the damages model as a certification question, not a trial question: "a model purporting to serve as evidence of damages in this class action must measure only those damages attributable to that theory. If the model does not even attempt to do that, it cannot possibly establish that damages are susceptible of measurement across the entire class for purposes of Rule 23(b)(3)." Precision can wait — "[c]alculations need not be exact" — but "at the class-certification stage (as at trial), any model supporting a ‘plaintiff’s damages case must be consistent with its liability case, particularly with respect to the alleged anticompetitive effect of the violation’" Comcast Corp. v. Behrend, 569 U.S. 27 (2013). The Court got there on familiar machinery: certification "is proper only if ‘the trial court is satisfied, after a rigorous analysis, that the prerequisites of Rule 23(a) have been satisfied,’" that analysis "will frequently entail ‘overlap with the merits of the plaintiff’s underlying claim,’" and Rule 23(b)(3)’s "predominance criterion is even more demanding than Rule 23(a)," demanding a "close look" at whether common questions predominate. The model in Comcast itself aggregated damages across four theories of antitrust impact when only one had survived for class treatment — so it "cannot possibly" carry predominance.

The sharpest fights are over uninjured class members inside the model. In re Rail Freight Fuel Surcharge Antitrust Litigation, No. 18-7010 (D.C. Cir. Aug 30, 2019) affirmed denial of certification where the plaintiffs’ own regression showed that "2,037 members of the proposed class—or 12.7 percent—suffered ‘only negative overcharges’ and thus no injury from any conspiracy": "even assuming the model can reliably show injury and causation for 87.3 percent of the class, that still leaves the plaintiffs with no common proof of those essential elements of liability for the remaining 12.7 percent." The "few reported decisions" on uninjured members "suggest that 5% to 6% constitutes the outer limits of a de minimis number," Rule 23 as construed in Comcast requires a "hard look at the soundness of statistical models that purport to show predominance," and confronting the winnowing problem "is part-and-parcel of the ‘hard look’ required by Wal-Mart and Comcast." The Ninth Circuit en banc pushed back — Olean Wholesale Grocery Co-Op v. Bumble Bee Foods LLC, 31 F.4th 651 (9th Cir. 2022) held that plaintiffs "must prove the facts necessary to carry the burden of establishing that the prerequisites of Rule 23 are satisfied by a preponderance of the evidence," credited pooled regression as "a generally reliable econometric technique to control for the effects of the differences among class members and isolate the impact of the alleged antitrust violations on the prices paid by class members," and rejected "the dissent’s argument that Rule 23 does not permit the certification of a class that potentially includes more than a de minimis number of uninjured class members" — a per se ceiling "is inconsistent with Rule 23(b)(3), which requires only that the district court determine after rigorous analysis whether the common question predominates over any individual questions, including individualized questions about injury or entitlement to damages."

In consumer price-premium cases the model is usually a conjoint survey, and the accepted and rejected lines turn on the supply side. Hadley v. Kellogg Sales Co., 324 F. Supp. 3d 1084 (N.D. Cal. 2018) maps both: "courts have repeatedly rejected conjoint analyses that only measure demand-side willingness-to-pay" — surveys that look "only ‘to the demand side of the market equation,’ converting what is properly ‘an objective evaluation of relative fair market values into a seemingly subjective inquiry of what an average consumer wants’" — but conjoint analyses "can adequately account for supply-side factors" and "estimate price premia without running afoul of Comcast" "when (1) the prices used in the surveys underlying the analyses reflect the actual market prices that prevailed during the class period; and (2) the quantities used (or assumed) in the statistical calculations reflect the actual quantities of products sold during the class period." The Ninth Circuit then settled how finished the model must be — Lytle v. Nutramax Laboratories, Inc., 99 F.4th 557 (9th Cir. 2024) held that "there is no general requirement that an expert actually apply to the proposed class an otherwise reliable damages model," so "class action plaintiffs may rely on an unexecuted damages model to demonstrate that damages are susceptible to common proof so long as the district court finds, by a preponderance of the evidence, that the model will be able to reliably calculate damages in a manner common to the class at trial"; the certifying court makes "a predictive judgment about how likely it is the expert’s analysis will eventually bear fruit." Comcast still polices the boundary: "where an expert’s damages model is untethered from plaintiff’s theory of liability such that it has no possibility of demonstrating the amount of damages in a particular case, Comcast holds that a plaintiff may not rely upon it to show that damages are capable of measurement on a classwide basis."

Treat the damages model as part of the liability architecture, not an appendix to it. Plaintiffs should tie every damages figure to the precise theory being certified and no other; in a conjoint case, anchor the survey to actual market prices and hold quantities to historical sales so the output reads as a market premium rather than subjective willingness to pay; and interrogate the model for uninjured members before the defense does, because the winnowing question is now part of the certification record. Defendants should depose the expert on fit rather than polish, quantify the uninjured share the plaintiffs’ own model implies — double digits has defeated predominance where 5–6% has not — and calibrate the attack to the forum: headcount arguments carry the D.C. Circuit line, while in the Ninth Circuit the fight is over reliability, since an unexecuted but reliable model can carry certification.