One of the key issues at class certification is whether plaintiffs have met their burden to establish commonality and predominance: that “questions of law or fact common to class members predominate over any questions affecting only individual members,” as required by Fed. R. Civ. P. 23(b)(3). Plaintiffs often rely on an expert model purporting to show that injury and damages can be determined classwide, so those issues do not defeat predominance.

A recent series of cases, most involving the insurance value of cars totaled in accidents, provide a useful reminder that, when a trial court reviews a plaintiffs’ model offered to show whether class members were injured and how much their damages are:

• a higher standard applies to whether the model shows any harm at all than it does to how the model measures the damages (following Tyson); and
• the harm and damages measured by the model must match the theory of liability (under Comcast).

In Sampson v. United Servs. Auto. Ass’n, 83 F.4th 414 (5th Cir. 2023), for instance, the Fifth Circuit vacated the certification of a class of plaintiffs whose cars had been totaled. The plaintiffs claimed that their insurer, USAA, underpaid them in violation of a Louisiana law that requires a totaled car be valued using certain methods, one of which is “a generally recognized motor vehicle industry source.” Louisiana Revised Statutes § 22:1892B(5)(b). USAA used CCC One Market Valuation Report (CCC), which plaintiffs argued was not such a “generally recognized” source. At class certification, plaintiffs presented a damages model based on the valuation provided by the National Automobile Dealers Association guidebook (NADA), seeking to recover the difference between the NADA and CCC values of the plaintiffs’ cars (prior to the accidents in which they were totaled). USAA argued that, although NADA is a generally recognized source, the law permitted the use of other generally recognized motor vehicle industry sources as well, such as Edmunds or Kelley Blue Book (KBB). Some class members could have had cars that were worth more in the CCC than in KBB, if less than in the NADA, and they would not have been harmed. 83 F.4th at 420.

The Fifth Circuit first considered the plaintiffs’ model in the context of damages. Recognizing that a plaintiffs’ damages model “‘need not be exact at the class certification stage’” but need only “‘reasonably account for the defendant’s liability,’” the court granted “arguendo that the district court did not reversibly err by choosing an imperfect damages model at the class certification stage.” Id. at 421 (quoting 4 NEWBERG AND RUBENSTEIN ON CLASS ACTIONS § 12:4 (6th ed.)). Sampson acknowledged Tyson Foods’ holding that “common questions may predominate under Rule 23(b)(3) ‘even though other important matters will have to be tried separately, such as damages.’” 577 U.S. 442, 453 (2016) (emphasis added). However, Sampson went on to state that “while damages are specifically described among these ‘other important matters,’ liability and injury are not. ” 83 F.4th at 421. The district court may have had discretion in determining that plaintiffs’ model was sufficient as a measure of damages at class certification, but that discretion did not, in the court’s view, “carry over from the context of damages to the context of liability.” Id. at 423. Because plaintiffs failed to show that NADA was the only true measure of a car’s value, or “put forward a coherent theory on which NADA, but not KBB or Edmunds, etc., can serve as a determinant of injury and liability as a matter of law,” the panel vacated the grant of class certification. Id. at 423 (emphasis original). Sampson cited Comcast, which makes sense. Comcast holds that a “model supporting a ‘plaintiff’s damages case must be consistent with its liability case . . . .’” 569 U.S. 27, 35 (2013). In Sampson, there was a mismatch between plaintiffs’ theory of harm (alleged violation of the statutory duty to use a “generally recognized” source) and the measure of damages (based on the failure to use NADA specifically).

Sampson found “particularly instructive the Ninth Circuit’s opinion in Lara v. First Nat’l Ins. Co. of Am., 25 F.4th 1134 (9th Cir. 2022), an “almost identical” case. Id. at 422. Lara also involved a claim based on an insurer’s use of the CCC to arrive at a totaled car’s value. In that case, plaintiffs argued the insurer violated Washington State insurance regulations by not “itemizing or explaining” the downward adjustment made to a car’s value based on its (pre-totaled) condition. Affirming the denial of class certification, Lara noted that “to show an injury will require an individualized determination for each plaintiff.” 25 F.4th at 1138. Sampson quoted Lara’s observation that “if there’s no injury, then the breach of contract and unfair trade practices claims must fail. That’s not a damages issue; that’s a merits issue.” Id. at 1139.

Two more opinions were issued after Sampson that distinguish Lara and certify classes of aggrieved insured motorists. The first, Coleman v. United Servs. Auto. Ass’n, certified a class of plaintiffs who alleged that USAA violated California law in providing insurance to members of military by shunting enlisted “good drivers” to a separate subsidiary from officers and charging them a higher rate. 2023 WL 9110926 (S.D. Cal. Dec. 22, 2023). In granting certification on the good driver claim, Coleman relied on a model presented by plaintiffs’ experts that compared rates from the two subsidiaries at various points in time to determine whether insured enlisted personnel were overcharged and by how much. Id. at *6. The district court noted that “Defendants have not identified any individual that Plaintiffs’ experts identify as a class member, who was not in fact injured,” and that they “focus their argument on inaccuracies in Plaintiffs’ premium calculations that do not materially affect Plaintiffs’ showing of classwide injury.” Id. at 24. Coleman distinguished Lara on the grounds that “[n]ot only did the plaintiffs in Lara not have any expert or model by which to determine injury classwide, but the nature of the plaintiffs’ claims also required discerning the actual value of an already totaled vehicle for each potential class member to determine if the respective class member was injured.” Id. 4. Coleman underscores the need for defendants seeking to defeat predominance to explain how a plaintiffs’ model could assign damages to class members who were not in fact injured.

The second, Costello v. Mountain Laurel Assurance Co., was another class action involving totaled cars where the claim was that an insurance company arrived at a value by improperly applying a projected sold adjustment (PSA) to reduce the advertised price of comparable vehicles. 2024 WL 239849 (E.D. Tenn. Jan. 22, 2024). Plaintiffs’ theory was that modern car buying has changed as a result of internet shopping and pricing software, such that “car dealerships now ‘price vehicles to market,’ meaning the listed price reflects the actual cash value of the vehicle.” Id. at *2. Costello distinguished Lara on the grounds that in that case the plaintiffs did not have a model at all, and so “did not offer any common proof of damages,” while in the case before it, plaintiffs “have provided a common method for calculating damages, removal of the PSA from the WCTL valuation reports.” Id. at 19. Costello emphasized the Comcast holding, noting explicitly that the plaintiffs’ model “matches the theory of liability,” which the model in Sampson did not.