DAMAGE METHODOLOGY TRENDS WITHIN FALSE ADVERTISING AND PRODUCT DEFECT CLASS ACTIONS

By Dan Werner, Ph.D., CPA and Garrett Glasgow, Ph.D.¹

I. INTRODUCTION

Over the past decade, courts have increasingly focused on issues related to the calculation of damages during the class certification stage.² At the class certification stage, Rule 23(b)(3) requires the court to find that common issues predominate, including with respect to the alleged injury.³ Cases such as Comcast highlight the need for a damages model to be tied to the theory of liability and for damages to be measurable on a classwide basis.⁴ The increased rigor with respect to issues of predominance and damages can also be seen through the evolution of damage approaches within false advertising and product defect litigation. As a result, engaging a damages expert at the class certification stage is now a de facto requirement in many cases.

This article reviews several different approaches to calculating damages in false advertising and product defect class action litigation in recent years. Specifically, we review the full refund approach, the promised discount approach, and several different applications of a price premium approach, including a simple price comparison, regression analysis, conjoint surveys, and economic market simulations. Although one or more damages methodologies may be available in theory, practitioners must be careful to design and implement their damages model in a manner that fits the facts of the case. Whether a damages model is sufficient for class certification often hinges on the details of implementation and the facts of the case, as we show in several examples.

[Page 125]

II. FULL REFUND APPROACH

The simplest damages model proposed by plaintiffs in false advertising and product defect class actions is the "full refund" approach. Under this approach, plaintiffs argue that consumers purchased the product solely due to some falsely advertised benefit, and the product is worthless without that benefit.⁵ Thus, under this scenario, class members are entitled to a full refund of the purchase price. Damage calculations under the full refund approach are appealingly simple. In fact, the full refund approach can be regarded as a special case of the "price premium" approach discussed below, in which the price premium resulting from the alleged misconduct is equal to the entire purchase price.⁶ However, demonstrating that this model is appropriate for a given case may pose significant challenges. For instance, defendants may argue that the product received by each class member had at least some value, and thus the full refund approach to damages does not apply.

Recent court decisions involving alleged misrepresentations on product labels have helped clarify when the full refund approach to damages is appropriate. For example, In re Tobacco Cases II, the California Court of Appeal wrote, "A full refund may be available in a UCL case when the plaintiffs prove the product had no value to them."⁷

When consumers received at least some value from their purchase, however, courts tend to reject the full refund approach for failure to satisfy Comcast’s requirement that classwide damages must be tied to the liability theory to satisfy Rule 23(b)(3)’s predominance requirement. For example, in Werdebaugh v. Blue Diamond Growers, a case involving allegations that various almond milk products were falsely advertised as "all natural," the court ruled against plaintiffs’ full refund model during the class certification stage, "because it is based on the assumption that consumers receive[d] no benefit whatsoever from purchasing the accused products [, which] cannot be the case, as consumers received benefits in the form of calories, nutrition, vitamins, and minerals."⁸ Another example is In Re: Pom Wonderful LLC Marketing and Sales Practices Litigation, a case involving allegations that certain Pom juice products falsely claimed to provide various health benefits. There, the court rejected the full refund methodology because it did not account for any value received by consumers independent of the contested labeling claims.⁹ Without accounting for the value received, the damages model did not isolate damages that resulted from the alleged misconduct and failed Comcast. More recently, the court denied class certification for the same reasons in Bruton v. Gerber Products, a case involving allegations that certain Gerber food products are deceptively labeled with various claims.¹⁰

[Page 126]

Courts apply similar logic outside the context of food labeling claims. For example, in Chowning v. Kohls, a case involving allegations of a deceptive reference price for apparel (e.g., comparing a selling price alongside a significantly higher fictitious "original" price), the plaintiff put forth a full refund approach to damages, among others.¹¹ However, the court rejected the full refund model because it "fail[ed] to account for the value Plaintiff received" and "neither party dispute[d] that the [contested products] each provided some value."¹² The court also suggested that a "viable measure of restitution" could include "a ‘price premium’ model in which an expert isolates the amount of the price attributable to the false representation."¹³ On appeal, the Ninth Circuit confirmed that the proper calculation of restitution in this case is the difference between the price paid and the value received.¹⁴

Even in situations where the full refund approach may apply, the implementation of the approach may come under scrutiny. For example, in Lambert v. Nutraceutical Corp., which involved a nutritional supplement that was alleged to be falsely advertised, the court initially certified a class that had proposed using the full refund approach, but later decertified the class because the plaintiff’s damages model was based on the suggested retail price, rather than the actual retail price paid by class members.¹⁵ As a result, common issues did not predominate since the plaintiff had not provided the evidence required to apply the full refund model for classwide damages.¹⁶ However, on appeal, the Ninth Circuit reversed the order decertifying the class, noting that "uncertainty regarding class members’ damages does not prevent certification of a class as long as a valid method has been proposed for calculating those damages."¹⁷ Since the plaintiff had presented evidence that the product was worthless, the court suggested that the full refund model was a workable method with only the suggested retail price and unit sales information, sufficient for the class to be certified.¹⁸

[Page 127]

III. PROMISED DISCOUNT APPROACH

Within allegations of a falsely advertised discount or reference price, plaintiffs have also proposed a "promised discount" approach, sometimes called an "actual discount" or a "false transaction value" model. Damages under this approach are based on the advertised discount that consumers did not receive. For example, suppose a product was listed for sale at $20 and this price was advertised as being "50% off," even though $20 was the normal selling price. If the promised discount of 50% had been applied to the normal selling price of $20, consumers would have paid $10 (instead of $20). Under this hypothetical example, a promised discount approach would calculate damages of $10 based on the difference between the price paid ($20) and the price consumers would have paid if the advertised discount was applied to the normal selling price of the product ($10).

Courts have had mixed reactions to the promised discount approach. This approach was sufficient to certify the class in Spann v. J.C. Penney Corporation, a case involving allegations of a deceptive "original" price advertised next to a "sale" price.¹⁹ In that case the court concluded that "[t]he amount plaintiff thought she was saving was a factor in her purchase decisions" and that "[p]laintiff is entitled to argue that payment of the ‘transaction value’ [the difference between expected and received value], if measurable and supported by evidence, would restore sums acquired by defendant’s allegedly unfair pricing practices."²⁰ However, the same approach was rejected in Chowning v. Kohls, in which the court noted that, "[t]o determine Plaintiff’s loss for purposes of restitution, the focus should be on what Plaintiff actually received given the price she paid, not on the bargain Plaintiff thought she was receiving."²¹ The Ninth Circuit upheld this decision on appeal, noting that the correct approach was "price paid versus value received," and pointing out that the promised discount approach was not available as a method for calculating restitution in a UCL action, because it "would effectively seek damages sounding in contract, not equity."²²

IV. PRICE PREMIUM APPROACH

The economic logic for damages under a price premium theory is straightforward: plaintiffs allege that the misconduct (e.g., false advertising or concealing a product defect) leads to higher consumer demand for the product, which in turn leads to higher market prices for the product. Thus, under this theory of injury, all class members paid a higher price for the product relative to what they would have paid if class members had not been misled when purchasing the product. Damages are calculated as the difference between what class members actually paid in the real-world (the "as-is" world) and the estimated market price for the product under a scenario in which the defendant did not engage in the alleged misconduct (the "but-for" world). As suggested by the prior cases, the price premium approach is economically appropriate when consumers received a product with some value. The challenge for any damages expert is to isolate the harm caused by the alleged misconduct (i.e., properly identify the price premium tied to the alleged misconduct). Below we discuss four recent approaches to calculating a price premium: (i) a simple price comparison, (ii) regression analysis, (iii) conjoint analysis, and (iv) economic market simulations.

[Page 128]

A. Calculating Price Premiums Through a Simple Price Comparison

One simple approach to isolating the price premium, assuming sufficient data exist that are tied to the facts of the case and the relevant marketplace, may be to compare the price of the product with the alleged deception to a nearly identical product without the alleged deception. This approach is consistent with the well-accepted market approach to valuation, which "provides an indication of value by comparing the asset with identical or comparable (that is similar) assets for which price information is available."²³

Although this method is conceptually sound, price comparisons of insufficiently similar products will fail to accurately identify the value of the product received by the plaintiffs and, as a result, fail to isolate the price premium caused by the alleged misconduct. Several cases involving false advertising allegations have noted this issue. For example, In re Pom Wonderful LLC Marketing and Sales Practices Litigation, plaintiffs put forth a price premium model that compared the price of the contested pomegranate juice product to a benchmark of prices for orange, grape, apple, and grapefruit juices. However, the court rejected this approach, concluding that this simple price comparison merely, "observed that Pom’s juices were more expensive than certain other juices. Rather than answer the critical question why that price difference existed, or to what extent it was a result of Pom’s actions, [plaintiffs] instead assumed that 100% of that price difference was attributable to Pom’s alleged misrepresentations."²⁴ The court’s decision to decertify the class underscores the importance of an appropriate damages model in light of the Comcast decision.

Similarly, in Bruton v. Gerber Products, plaintiffs put forth a simple price premium model comparing the price of an allegedly mislabeled product to the prices of allegedly comparable products that did not include the contested labeling statements. The court rejected this model, finding that it "presume[d] that the entire price difference between Gerber baby food and a competitor such as Beech Nut baby food [was] attributable to Gerber’s misleading statements, when any number of factors—such as brand recognition or loyalty, ingredients, and product quality—might explain all or part of the difference."²⁵

These court decisions, among others, underscore the importance of isolating the alleged misconduct from other factors that may influence pricing, and thus tying the alleged misconduct to the price premium paid by class members. Although a simple price comparison model may be informative in certain cases where a sufficiently similar product is found without the contested labeling claims, a comparison that does not properly account for all relevant factors influencing pricing cannot sufficiently measure the harm sustained by class members due to the alleged misconduct.

[Page 129]

B. Calculating Price Premiums Through Regression Analysis

A "regression analysis" is a statistical method that measures the influence of one or more independent variables (e.g., supply and demand factors) on a dependent variable (e.g., price). By including a number of independent variables in a properly specified regression model, the practitioner is effectively "controlling" for those factors (i.e., measuring the effect of each independent variable on the dependent variable, while holding all of the other independent variables constant). As a result, with proper data and implementation, in some cases a regression analysis can measure the incremental impact of alleged misconduct on prices paid by class members, while controlling for other important factors that also influence price.

Courts have long accepted a properly implemented regression analysis with appropriate data as a reliable method for analyzing the impact of alleged misconduct on prices paid by class members (e.g., under price fixing allegations).²⁶ When applying regression analysis to analyze the price of consumer products, economists often view such products as a "bundle of attributes," in which each component of the product is responsible for some portion of the price. These types of regressions are often called "hedonic regressions."²⁷ Under certain market conditions and with sufficiently rich data on comparable products, product attributes, and prices paid, a regression analysis can account for each of the contested product’s attributes while measuring the incremental effect of the alleged misconduct (e.g., the falsely advertised feature or undisclosed product defect) on the product’s price at a specific time. The acceptance of hedonic regression analysis by courts for the purposes of class certification has been mixed, depending on the facts of the case, available data, and implementation of the method. Courts have certified classes under this approach in some false advertising cases such as Brown v. The Hain Celestial Group, a case involving allegations that certain cosmetic products were falsely advertised as organic, and in Pettit v. Procter & Gamble Company, a case involving allegations that certain cleaning wipes were falsely advertised as "flushable."²⁸ In other cases, courts reject a regression-based approach to estimating a price premium for a variety of reasons, some of which are discussed below.

[Page 130]

Although regression analysis is theoretically able to measure a price premium caused by false advertising or the failure to disclose a product defect, care must be taken to ensure proper application. For instance, when two or more independent variables are highly correlated (an issue known as "multicollinearity") a regression model can have difficulty measuring the extent to which each variable is independently influencing the dependent variable. In the extreme case, independent variables can be "perfectly collinear" in a way that prevents a regression model from measuring the independent effect of each variable.

An example of the perfect collinearity issue arose in Werdebaugh v. Blue Diamond Growers. Although the court initially found that plaintiffs’ regression model was sufficient for class certification,²⁹ upon implementation of the model the expert omitted the Blue Diamond brand variable in an attempt to address the problem of perfect collinearity, and defendants successfully argued that plaintiffs’ application of the regression model "conflate[d] the effect of the alleged mislabeling with the value of Blue Diamond’s brand."³⁰ As a result, the court decertified the class, "conclud[ing] that the perfect collinearity problem here renders the damages model insufficient under Comcast because [plaintiff’s] model is incapable of isolating the damages attributable to Defendant’s alleged wrongdoing."³¹

The concept of "omitted variable bias" is another well-known issue that could result in a flawed application of the regression approach in false advertising and product defect cases. Omitted variable bias can occur when a regression model excludes one or more important variables that influence price. As a result, the regression model may produce a biased result that does not accurately measure the influence of the alleged misconduct on pricing, meaning that the price premium has not been reliably estimated. For example, in Bruton v. Gerber Products, plaintiffs put forth a regression analysis to measure the price premium. Although the court granted that "absolute precision is not required at the class certification stage," and it is not necessary at the class certification stage to show that the method will work with certainty,³² the court found the proposed regression did not satisfy Comcast in part because plaintiffs "d[id] not explain how the proposed Regression Model w[ould] account for independent variables that might affect the products’ price or sales, such as advertising, brand recognition or loyalty, the prices of competing products, regional differences, consumers’ income, and seasonality."³³ From an econometric perspective, the absence of a variable within a regression analysis, in theory, does not necessarily prove that the regression results are (or will be) biased.³⁴ The standards of the economic discipline do not require that every theoretically possible variable be accounted for in a regression analysis, only that the primary determinants are included.³⁵ However, the recent decision in Bruton suggests that plaintiffs should perform analyses at the class certification stage to establish that the regression analysis is workable.

[Page 131]

The time frame covered by a regression analysis is another important consideration. While some analyses compare the price of the contested product to a benchmark of prices for products sold without the alleged misconduct during the proposed class period, others take the form of a "before-and-after" approach, which compares the price of the contested products before and after the alleged misconduct took place. The strength of the before-and-after approach is that the price comparison is for the same product with and without the contested labeling, eliminating the concerns related to comparing the contested product to other products. However, a regression analysis using the before-and-after approach should also consider additional variables related to supply and demand factors that might have caused prices to change over the relevant time period.

Practitioners also should ensure that sufficient data are available to implement the before-and-after approach in a manner that will accurately capture the alleged misconduct. In Werdebaugh v. Blue Diamond Growers, plaintiffs first put forth a before-and-after regression analysis in support of their motion for class certification, but later abandoned the approach following the close of fact discovery because plaintiffs’ expert concluded that there was not sufficient evidence to precisely determine the before-and-after periods.³⁶

When there is uncertainty as to when the "before" period (i.e., without alleged misconduct) and "after" period (i.e., with alleged misconduct) actually occurred, "measurement error" may influence the variable identifying the accused conduct. For example, plaintiffs in Bruton v. Gerber Products put forth a before-and-after application of regression analysis, but the court acknowledged an "important flaw" that undermined their ability to "determine precisely when consumers were buying Gerber products with the challenged label statements, and when consumers were buying Gerber products without the challenged label statements."³⁷ For example, Gerber reported that there were timing differences in when a new label might appear on store shelves and submitted evidence that at certain times, there were two different labels—one with the contested statement and one without—on a product for sale within the same store. The court concluded, "Given the absence of a reliable means to compare products with and without the challenged labeling, the Court finds that the proposed Regression Model fails to satisfy Comcast."³⁸

[Page 132]

In sum, regression analysis is a generally accepted and widely used methodology in the economics discipline. However, the widespread acceptance of the method does not necessarily mean that it can be proven to be a workable method to reliably measure the economic harm caused by the alleged misconduct in every case. Practitioners should carefully consider whether a given regression analysis is appropriate for the facts of the case and the available data.

C. Calculating Price Premiums Through Survey-Based Approaches

Surveys of consumers have also been used to calculate damages in false advertising and product defect class actions. Rather than analyze historical data, survey-based approaches rely on survey responses from individuals whose preferences, purchasing habits, and/or responses to product advertising, labeling, and/or characteristics could be said to reasonably represent the putative class (or the desired target population).

Perhaps the most common type of survey experts use to calculate damages in false advertising and product defect class actions is a choice-based conjoint analysis, often referred to simply as "conjoint analysis." In a choice-based conjoint analysis, survey participants are presented with a series of questions, each of which asks them to select their most preferred product from among several available product options with different attributes (including price). A statistical model can use data from a conjoint survey to estimate the influence of each product attribute on the choices made by survey respondents. In particular, the model would measure how each product attribute affects the probability that consumers will select a given product configuration. The results from this statistical model therefore provide a measure of demand for each attribute of a product. Researchers can use these results to estimate consumers’ "willingness-to-pay" for a given product attribute by observing how consumers make trade-offs between higher or lower prices and the presence or absence of the attribute. For instance, a conjoint analysis might reveal that consumers are willing to pay an additional $1 for a product advertised with a misleading claim, or would want a $1 discount for the product when the presence of a defect is known. Thus, a properly implemented conjoint analysis may help to provide empirical conclusions that directly link consumer demand to the contested claims (e.g., mislabeling).

Courts are well-attuned to analyzing the reliability of a survey design,³⁹ and they have not hesitated to reject damages calculations based on data where the underlying survey

failed to replicate the real-world conditions in which the accused conduct took place. For example, in Townsend v. Monster Beverage Corp., plaintiffs put forth a survey designed to estimate the price premium associated with allegedly deceptive labeling on certain Monster Beverage products. However, the survey did not use the precise language in the accused labeling, but instead relied on paraphrased versions of the allegedly misleading statements. The court concluded that, in this case, "[i]n order to evaluate the price premium attributable to each of the alleged misstatements, [plaintiffs] would have had to conduct a survey that included the specific statements themselves."⁴⁰ The court excluded all testimony that relied on the paraphrases of the allegedly deceptive advertising, and denied class certification.⁴¹

[Page 133]

While a properly applied conjoint analysis can measure incremental changes in demand and willingness to pay due to the alleged misconduct, a central issue in recent cases is whether or not the measured willingness to pay can be used as the estimate of the price premium paid by class members.⁴² Basic economic theory notes how prices are the result of not only demand conditions (e.g., consumers’ willingness to pay for the product), but also supply conditions (e.g., companies’ willingness to supply a given amount of the product for a given price). Therefore, quantifying a change in demand for a product as a result of certain challenged conduct is not necessarily sufficient to quantify the effect of the conduct on the price of the product. Thus, depending on what is assumed about the economic market, a conjoint analysis that only considers the effect on demand through the consumer’s willingness to pay for a given product attribute may not accurately capture the price premium consumers paid.

Two recent decisions highlight these issues: Hadley v. Kellogg Sales, which involved allegations that certain products are mislabeled as healthy, and Schechner v. Whirlpool, a case involving allegations that certain ovens were falsely advertised as self-cleaning.⁴³ In Hadley, plaintiffs submitted a choice-based conjoint analysis to measure the monetary value that consumers attached to the alleged misrepresentations, which was used as the price premium for damages. Defendants argued that the approach could only measure demand-side factors, and thus failed to properly measure the price premium, while plaintiffs argued that the model properly considered supply-side factors by using market prices and by assuming that supply is fixed.⁴⁴ The court explained:

[Page 134]

[C]ourts have repeatedly rejected conjoint analyses that only measure demand-side willingness-to-pay. However, courts have also found that conjoint analyses can adequately account for supply-side factors—and can therefore be utilized to 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.⁴⁵

Thus, in this case, the court accepted plaintiffs’ willingness to pay measure of the price premium for the purposes of class certification.

In Schechner, the court initially ruled that plaintiffs’ conjoint analysis for estimating damages was admissible under Daubert, but ultimately decided that the conjoint analysis failed to satisfy Comcast, because it did not sufficiently consider prevailing market conditions under which the product would be sold.⁴⁶ The court rejected plaintiffs’ argument that the use of historical transaction prices within a conjoint analysis incorporates supply-side factors, noting that the use of historical transaction prices, at best, only accounts for historical supply-side factors, whereas a price premium approach should account for the difference between actual prices paid and the prices that would have been paid without the misconduct (prices in the but-for world).⁴⁷ In this case, but-for prices depended on the prevailing market conditions without the misconduct, including competitors’ reactions to the change in demand for the product in the absence of the misconduct, meaning the use of historical prices alone did not imply proper consideration of the supply-side conditions.⁴⁸ As a result, the court ruled that plaintiffs’ approach failed the Comcast requirements and denied plaintiffs’ motion to certify the class.⁴⁹

Finally, some plaintiffs have proposed a hybrid approach to damages that combines a conjoint analysis with a regression analysis. One example comes from In re Conagra Foods, which involved allegations that products contained genetically modified organisms ("GMOs") and were thus falsely advertised as "100% Natural."⁵⁰ In this case, plaintiffs put forth a two-step approach: First, they proposed a regression analysis to measure the market price premium associated with the labeling statement "100% Natural."⁵¹ Second, to tie the overcharge to the non-GMO aspect of their claim, plaintiffs proposed a conjoint analysis to apportion the price premium associated with "100% Natural" down to the value consumers attributed to the product being GMO-free.⁵² Rejecting defendants’ argument that the conjoint analysis failed to fully account for supply side issues, the court ruled that plaintiffs’ hybrid approach was sufficient under Comcast and certified the class.⁵³ While this hybrid approach was sufficient for class certification in 2015, it is possible that such a hybrid approach may be reevaluated by future courts in light of more recent scrutiny applied to price premiums calculated through conjoint analyses, as described above.

[Page 135]

D. Calculating Price Premiums Through Economic Market Simulations

More recently, economic market simulations have been applied in litigation as a way to estimate a price premium in a manner that accounts for the courts’ concerns about the ability of conjoint analysis to address supply-side factors (e.g., production costs, supply response, and competitor interactions). An extensive body of academic literature focuses on combining the consumer demand side of the market with the supply side of the market to determine market equilibrium prices.⁵⁴

In the presence of basic economic assumptions (e.g., multiple profit-maximizing sellers interacting with rational consumers), a properly applied market simulation incorporates information on the defendant’s (and its competitors’) prices, costs, and market share, along with demand-side information (e.g., consumer’s willingness-to-pay for product attributes and elasticity of demand) to estimate market equilibrium prices.⁵⁵ Economic market simulations are typically calibrated to real-world data using a representative price and product sold by each firm, to ensure the model will accurately capture real-world interactions to reasonably estimate actual market prices.

After proper calibration to real-world data, economic market simulations can estimate equilibrium prices in the but-for world by first calculating the change in demand that would have occurred in the absence of the alleged misconduct. This change of demand can be estimated using a variety of methods, including using the parameters from a statistical model based on conjoint survey data. The change in demand for the defendant’s product then leads to a supply response from each company, which in turn leads to a new set of equilibrium market prices. The new equilibrium price for the defendant’s product represents the price that consumers would have paid in the absence of the alleged misconduct. Finally, the economic simulation will calculate the price premium as the difference between the estimated equilibrium price and the actual price consumers paid.

Courts have allowed the economic market simulation approach under the Daubert standard. For example, in Johannessohn v. Polaris Industries, which involved an alleged product defect in all-terrain vehicles, plaintiffs’ expert first used a conjoint survey to test and measure the impact that an appropriate disclosure would have had on consumers’ willingness-to-pay for the contested products.⁵⁶ Next, a separate expert incorporated this information into an economic market simulation, estimated the market equilibrium price that would have resulted under proper disclosure of the alleged defect, and calculated price premium damages accordingly.⁵⁷ The court admitted the economic market simulation methodology under Daubert, noting that "[t]here is no question that a market simulation is a scientifically valid method to determine the market equilibrium price in a counterfactual world."⁵⁸

[Page 136]

However, as with all prior methods, the general acceptance of economic market simulations does not mean that courts will universally allow the method in all instances. The reliability of the results of an economic market simulation depends on its implementation. Indeed, courts have disallowed the application of this method under Daubert due to of the use of flawed inputs. For example, in Unwired Planet, LLC v. Apple, Inc., a non-class, patent infringement case, the court rejected plaintiffs’ proposed market simulation approach for failing to satisfy Daubert because the market simulation was based, in part, on unsupported assumptions.⁵⁹ The Unwired Planet decision underscores the importance of using reasonable assumptions and reliable inputs within an economic market simulation. Where there is uncertainty over a particular input, practitioners would be wise to test a range of reasonable input values to ensure their results are not being driven by a particular assumption about an input’s value.

V. CONCLUSION

In the wake of Comcast, courts have increasingly focused on damage models at the class certification stage. As a result, in false advertising and product defect class actions there has been a trend toward hiring experts to perform a thorough analysis of damages in advance of class certification, and the damages models themselves have become increasingly sophisticated and rigorous. Recent court rulings, Daubert motions, summary judgement motions, and class certification motions have helped to provide a roadmap of what constitutes an acceptable approach to the calculation of damages in false advertising and product defect class actions. However, while approaches to calculating damages in these cases continue to evolve, appropriate damages calculations will always depend both on the facts of the case and the proper application of the relevant damages model.

[Page 137]

——–

Notes:

^1. Dan Werner is a Senior Consultant at NERA Economic Consulting. Garrett Glasgow is an Associate Director at NERA Economic Consulting. The views expressed within this article are solely those of the authors and do not necessarily reflect the views of NERA Economic Consulting, its clients, or other NERA consultants.

^2. "The legal community can, however, agree on one central proposition: class certification has, in recent decades, become increasingly complex…This was not always the case, of course. Federal rule of Civil Procedure 23 was once a relatively simple procedural mechanism, and the class-certification determination typically occurred at the outset of any litigation, well in advance of significant discovery. With the advent of Wal-Mart and Comcast, however, and their expositions of the ‘rigorous analysis’ required of courts, litigants can expect that the recent trend of deferred class-certification decisions (in favor of substantial precertification discovery) will continue." Michael D. Hausfeld et al., Antitrust Class Proceedings — Then and Now, in 26 Research in L. & Econ. The L. & Econ. of Class Actions 77, 77—133 (James Langenfeld ed., 2014).

^3. "A court may find a lack of predominance if the plaintiffs cannot prove injury, causation, or an element of a substantive claim on a classwide basis. Predominance may also be lacking if the defendant can assert individualized defenses to class members’ claims or different state laws with material variations apply to different class members’ claims. Courts agree that the mere existence of individualized damages does not preclude class certification because to hold otherwise would eviscerate the purpose and use of Rule 23(b)(3). Nevertheless, the need to determine damages on an individualized basis may be considered as a factor when determining whether the predominance requirement has been met." Caroline H. Gentry, A Primer On Class Certification Under Federal Rule 23, ABA Section of Litigation Corporate Counsel CLE Seminar, https://www.classactiondeclassified.com/wp-content/uploads/sites/26/2017/08/a_primer_class_certification_under_federal_rule.pdf.

^4. In Comcast Corp. v. Behrend, 569 U.S. 27 (2013), plaintiffs alleged four different theories of antitrust impact, but only one was deemed suitable to classwide resolution. However, plaintiffs’ proposed damages model was unable to isolate damages from the one remaining theory, and the Supreme Court found class certification was improper because plaintiffs failed to establish that damages could be measured on a classwide basis.

^5. Similarly, in product defect cases, plaintiffs argue that the product was defective, and the alleged defect resulted in the product having no value.

^6. This view is consistent with the view of the California Court of Appeal in In re Tobacco Cases II, 240 Cal. App. 4th 779 (2015), which noted that the court in Ortega v. Natural Balance, Inc., 300 F.R.D. 422, 430 (C.D. Cal. 2014), "certified a class action where the plaintiffs sought a full refund for a dietary supplement on the ground it falsely advertised aphrodisiac qualities and had no value apart from that claim. In that type of case, it appears the Vioxx measure of restitution would still apply, since the price paid minus the value actually received equals the price paid. In that instance, a full refund is not based solely on the defendant’s conduct, but on the dual considerations of the plaintiffs’ losses and the defendant’s conduct." Id. at 795-96.

^7. Id. at 795.

^8. Werdebaugh v. Blue Diamond Growers, Case No. 12-CV-2724-LHK, 2014 WL 2191901, at *22 (N.D. Cal. May. 23, 2014) ("Werdebaugh I").

^9. In re Pom Wonderful LLC Mkt. & Sales Practices Litig., Case No. ML 10-02199 DDP (RZx), 2014 WL 1225184, at *3 (C.D. Cal. Mar. 25, 2014) ("Pom Wonderful").

^10. Bruton v. Gerber Products Co., Case No. 12-CV-02412-LHK, 2018 WL 1009257, at *9 (N.D. Cal. Feb. 13, 2018) ("Bruton").

^11. Chowning v. Kohl’s Dep’t Stores, Inc., Case No. 2:15-cv-08673-RGK-SP, 2016 WL 1072129, at *7 (C.D. Cal. Mar. 15, 2016) ("Chowning I").

^12. Id.

^13. Id. at *12.

^14. Chowning v. Kohl’s Dep’t Stores, Inc., 735 Fed. Appx. 924, 925 (9th Cir. 2018) ("Chowning II").

^15. Lambert v. Nutraceutical Corp., 870 F.3d 1170, 1174-75 (9th Cir. 2017), reversed on other ground sub. nom Nutraceutical Corp. v. Lambert, 139 S. Ct. 710 (2019).

^16. Id. at 1175.

^17. Id. at 1182.

^18. Id. at 1183-84. The Ninth Circuit concluded however, that "whether Lambert could prove damages to a reasonable certainty on the basis of his full refund model is a question of fact that should be decided at trial." Id.

^19. Spann v. J.C. Penney Corporation, 307 F.R.D. 508 (C.D. Cal., May 18, 2015).

^20. Id. at 520-521.

^21. Chowning I, 2016 WL 1072129, at *10 (emphasis original).

^22. Chowning II, 735 Fed. App. at 925-26.

^23. International Valuation Standards Council, International Valuation Standards 2017.

^24. Pom Wonderful, 2014 WL 1225184, at *5.

^25. Bruton, 2018 WL 1009257, at *9.

^26. "The most common statistical method employed in antitrust litigation involves the estimation of ‘reduced-form’ price equations. A typical reduced-form model might explain the variation in the price of a product as a function of a series of variables relating to cost, demand, and market structure. In some cases, the model would include additional ‘dummy variables’ (i.e., variables that take on the values zero or one) that represent geographic or time differences in prices that account for variables omitted from the model (the set of such variables are the ‘fixed effects’). The model is called ‘reduced form’ because the price equation is derived from other more basic economic relationships relating to demand and supply. As a result, the parameters (the variable coefficients in a multiple regression model) of a reduced-form equation are typically themselves functions of a number of the structural parameters (the parameters of the underlying economic relationships)" (Rubinfeld, Daniel L. "Quantitative methods in antitrust." Issues in Competition Policy 723 at 724-25, ABA Section of Antitrust Law (2008)).

^27. See, e.g., Sherwin Rosen, Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition, 82 The Journal of Political Economy 34 (1974); Michael Greenstone, The Continuing Impact of Sherwin Rosen’s ‘Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition’, 125 Journal of Political Economy 1891 (2017).

^28. Brown v. Hain Celestial Grp., Case No. 11-cv-03082-LB, 2014 WL 6483216, at *18-20 (N.D. Cal. Nov. 18, 2014); Pettit v. Procter & Gamble Co., Case No. 15-cv-02150-RS, 2017 WL 3310692, at *4 (N.D. Cal. Aug. 3, 2017).

^29. Werdebaugh I, 2014 WL 2191901, at *26.

^30. Werdebaugh v. Blue Diamond Growers, Case No. 12-cv-2724-LHK, 2014 WL 7148923, at *10 (N.D. Cal. Dec. 15, 2014) ("Werdebaugh II").

^31. Id. at * 11.

^32. Bruton, 2018 WL 1009257 at *12.

^33. Id. at *10.

^34. An omitted variable will cause bias when it is correlated with both the dependent variable and the variable of interest.

^35. "Of course, the model is only a simplification of reality. It will include the salient features of the relationship of interest, but will leave unaccounted for influences that might well be present but are regarded as unimportant. No model could hope to encompass the myriad essentially random aspects of economic life. It is thus also necessary to incorporate stochastic elements. As a consequence, observations on a dependent variable will display variation attributable not only to differences in variables that are explicitly accounted for, but also to the randomness of human behavior and the interaction of countless minor influences that are not." Greene, William H. Econometric Analysis. Fifth Edition. Boston: Prentice Hall, 2003, p. 2.

^36. Werdebaugh II, 2014 WL 7148923, at *4, *9.

^37. Bruton, 2018 WL 1009257 at *11.

^38. Id.

^39. As the Manual for Complex Litigation explains, "The sampling methods used must conform to generally recognized statistical standards. Relevant factors include whether the population was properly chosen and defined; the sample chosen was representative of that population; the data gathered were accurately reported; and the data were analyzed in accordance with accepted statistical principles . . . In addition, in assessing the validity of a survey, the judge should take into account the following factors: whether the questions asked were clear and not leading; whether the survey was conducted by qualified persons following proper interview procedures; and whether the process was conducted so as to ensure objectivity (e.g., determine if the survey was conducted in anticipation of litigation and by persons connected with the parties or counsel or by persons aware of its purpose in the litigation)." Federal Judicial Center, Manual for Complex Litigation, § 11.493 (4th Ed. 2004).

^40. Townsend v. Monster Beverage Corp., 303 F. Supp. 3d 1010, 1024 (C.D. Cal. 2018).

^41. Id. at 1024, 1052.

^42. See, e.g., In re NJOY, Inc. Consumer Class Litig., 120 F. Supp. 3d 1050, 1070-73 (C.D. Cal. 2015); In re Dial Complete Mktg. & Sales Practices Litig., 320 F.R.D. 326, 334 (D.N.H. 2017) ("However, whether conjoint analysis can be used to ascertain a price premium attributable to a particular product feature is not fully resolved."); Broomfield v. Craft Brew Alliance, Inc., Case No. 17-CV-01027-BLF, 2018 WL 4952519, at *14-20 (N.D. Cal. Sept. 25, 2018).

^43. Hadley v. Kellogg Sales Co., 324 F. Supp. 3d 1084 (N.D. Cal. 2018); Schechner v. Whirlpool Corp., Case No. 2:16-cv-12409, 2019 WL 4891192 (E.D. Mich. Aug. 13, 2019).

^44. Hadley, 324 F. Supp. 3d at 1103-04. Under certain assumptions, a fixed supply may provide a reasonable but-for world for a short-term economic equilibrium. For example, if defendants had already manufactured a batch of products and subsequently discovered a product defect, a reasonable but-for world may be one in which the manufacturer has an incentive to sell all of them for any price the market would bear. However, for longer class periods and cases that do not adhere to this unique "sunk cost" circumstance, assuming a fixed supply is inconsistent with economic theory.

^45. Id. at 1105.

^46. Schechner, 2019 WL 4891192, at *7-8.

^47. Id. at *7.

^48. Id.

^49. Id. at *8-9.

^50. In Re Conagra Foods, Inc., 90 F. Supp. 3d 919 (C.D. Cal 2015).

^51. Id. at 944.

^52. Id. at 944, 954.

^53. Id. at 1027-31, 1035.

^54. For example, Steven Berry, et al., Automobile Prices in Market Equilibrium, 63 Econometrica 841 (1995); Aviv Nevo, A Practitioner’s Guide to Estimation of Random-Coefficients Logit Models of Demand, 9 Journal of Economic and Management Strategy 513-548 (2000); Steven Berry, et al., Differentiated Products Demand Systems from a Combination of Micro and Macro Data: The New Car Market, 112 Journal of Political Economy 68-105 (2004); Petrin, Amil, Quantifying the Benefits of New Products: The Case of the Minivan, 110 Journal of Political Economy 705—729 (2002); Greg Allenby, et al., Valuation of Patented Product Features, 57 The Journal of Law & Economics 629-663 (2014).

^55. Note that some damage calculations described as "market simulations" assume that the defendant and competitors do not adjust supply in response to demand, and thus measure willingness-to-pay rather than a price premium. See Greg Allenby, et al, Valuation of Patented Product Features, 57 The Journal of Law & Economics 629-663, 648 (2014).

^56. Johannessohn v. Polaris Indus., Inc., 450 F. Supp. 3d 931, 970 (D. Minn. 2020).

^57. Id. at 970-71.

^58. Id. at 972-73.

^59. Unwired Planet, LLC v. Apple Inc., Case No. 13-cv-04134-VC, 2017 WL 589195, at *2 (N.D. Cal. Feb. 14, 2017).