'I Demand a Refund'

Numerous class action complaints have been filed recently, challenging labels' claims that the products are “all natural.” Similarly, cosmetics products may be challenged for “organic” or “natural” labeling claims. The food-labeling cases allege that the “all-natural” label claims are deceptive because the products contain allegedly synthetic ingredients such as ascorbic acid (Vitamin C), citric acid, potassium citrate or calcium chloride. One of the many roadblocks for plaintiffs in establishing class certification in these cases has been finding a reliable damages methodology that can withstand scrutiny. This article summarizes several proposed models, and how the courts have dealt with them.

Damages Models

In Comcast Corp. v. Behrend, 133 S. Ct. 1426, 1433 (U.S. 2013), the Supreme Court held that at the class certification stage, the plaintiffs' damages model must be consistent with its liability case and “susceptible of measurement across the entire class.” Id. at 1433-34. The challenge in finding a methodology is to isolate a proper measure of what amount is necessary to compensate the purchaser for the difference between the product as labeled and the product actually received by the consumer. Three models have been discussed for trying to quantify this alleged damage in “All Natural” label cases.

1. The Price Premium Model

The Price Premium model allows as restitution the entire difference between the price of the product in question and that of allegedly comparable products without the contested label language. This model was rejected in Northern District of California Brazil v. Dole Packaged Foods, LLC, 12-CV-01831-LHK ( Dole ), 2014 U.S. Dist. LEXIS 74234 (N.D. Cal. May 30, 2014) and Werdebaugh v. Blue Diamond Growers (Blue Diamond , 2014 U.S. Dist. LEXIS 71575 (N.D. Cal. May 23, 2014). The model fails to control for advertising, brand name and other factors, and can conflate the alleged mislabeling with the value of the brand.

In Blue Diamond, the plaintiff challenged two statements in almond milk products' labels as allegedly false and misleading: 1) listing the sweetener used in the almond milk products as “evaporated cane juice” instead of “sugar”; and 2) including the statement “All Natural” when the products contained potassium citrate. The court in Blue Diamond explained that the expert had “no way of linking the price difference, if any, to the allegedly unlawful or deceptive label statements or controlling for other reasons why allegedly comparable products may have different prices.” The Price Premium model failed the Comcast test by simply calculating the price difference “without tying it to a legal theory.”

Moreover, the model did not account for factors such as quality or brand loyalty, which might lead consumers to prefer the challenged brand name product over other identical products. See also In re POM Wonderful, LLC, 2014 U.S. Dist. LEXIS 40415, 1 (C.D. Cal. Mar. 25, 2014) ( POM ) (challenging certain pomegranate juice products for claims of certain health benefits); Astiana v. Ben & Jerry's Homemade, Inc., 2014 U.S. Dist. LEXIS 1640 (N.D. Cal. Jan. 7, 2014) (Ben & Jerry's) (alleging that product labels of “All Natural” were misleading as the cocoa powder in ice cream is made with synthetic processing agents).

In the POM case, the court concluded that “where, as here, consumers buy a product for myriad reasons, damages from the alleged misrepresentation will not possibly be uniform or amenable to class proof.” In Ben & Jerry' , the court rejected the Price Premium model, equating it to a “fraud on the market” model that could not work without first establishing an efficient market that “prices a good on the basis of all available material information.” The court further found the model unreliable for measuring class-wide damages, finding speculative plaintiffs' assertions that an “All Natural” ice cream label requires a $0.50 to $0.75 premium over other ice creams. Ben & Jerry's successfully argued that the retail market price varies based on a number of factors that had nothing to do with a product's “All Natural” status, such as the nature and location of the retailer. But see Brown v. Hain Celestial Group, Inc. , 2014 U.S. Dist. LEXIS 161036, 54 (N.D. Cal. Nov. 14, 2014) (Price Premium model alloed and class certification granted to plaintiffs challenging “organic” labeling cosmetics).

In Jones v. Conagra Foods, Inc., 2014 U.S. Dist. LEXIS 81292, 71 (N.D. Cal. June 13, 2014) ( ConAgra ), the plaintiff challenged “100 percent natural” or “free of artificial ingredients & preservatives” label claims on Hunt's canned tomato products that contained citric acid or calcium chloride, as well as similar claims with respect to PAM cooking spray, and claims regarding “natural antioxidants” in Swiss Miss hot cocoa products. The expert proposed a “benefit of the bargain approach,” taking a “comparable product” and comparing the price between it and the allegedly mislabeled Hunt's product. The court held that this approach was “deeply flawed.” Id. at *73-75. The comparators either contained a “100% natural” label statement as well, or were generic brands being compared with brand-name products.

2. The Regression Model

Although the Regression model has enjoyed somewhat more success, it is still strenuously challenged at class certification and has proven difficult for plaintiffs to attribute certain damages to the allegations in the complaint. Plaintiffs claim that this model provides a more precise measure of damages because it controls for factors commonly recognized as associated with sales such as “price of the product, prices of competing and complementary products, income, advertising, seasonality, and regional differences.” See Dole, 2014 U.S. Dist. LEXIS 74234, at *58. The model further considers sales of the products before and after the challenged labeling language. Id .

In ConAgra, the court rejected the Regression model, stating that “[expert] does not provide a clearly defined list of variables, he has not determined whether the data related to any or all of his proposed control variables exists, and he has not determined, or shown how he would determine, which competing and complementary products he would use.” 2014 U.S. Dist. LEXIS 81292, at *76-78. The ConAgra plaintiffs have appealed the denial of class certification to the Ninth Circuit Court of Appeals; thus, there may soon be circuit court guidance on this issue. In their brief to the Ninth Circuit, the plaintiffs argue that the court erred in requiring precise damages calculations at the class certification stage and erred in conflating damages with restitution. Appellants' Brief, Case No. 14-16327, Dkt. Entry 21 (9th Cir. Nov. 21, 2014). They also argue that the Regression model satisfies Comcast .

In Dole, the expert proposed to determine the damages from the alleged misrepresentations “by examining sales of the identified products before and after Dole placed the alleged misrepresentations on its product labels, using regression analysis to control for other variables that could otherwise explain changes in Dole's sales.” 2014 U.S. Dist. LEXIS 74234 at *58. The court held that the Regression model “provide[d] a means of showing damages on a class-wide basis through common proof,” thus “satisf[ying] the Rule 23(b)(3) requirement that common issues predominate over individual ones.” Id. The court then certified the proposed Damages Class, although the expert had not yet conducted the regression analysis, citing lack of discovery from Dole. Despite this early plaintiff's victory, in November 2014, the district court later decertified the damages class. Dole, 2014 U.S. Dist. LEXIS 157575, 1 (N.D. Cal. Nov. 6, 2014).

In Dole, the expert was unable to isolate the damages attributable to Dole's alleged misbranding. Most of the product labels had not changed during the class period, making the before-and-after analysis impossible. The methodology employed by the expert was ultimately unable to control for other variables such as advertising, and the expert failed to verify whether the labels of non-Dole products contained “all natural” claims. The analysis thus failed the Comcast test that the damage be related to the liability claim ' specifically, it failed to measure only those damages attributable to Dole's use of the “All Natural Fruit” label statements. Id.

The same occurred in Blue Diamond, where class certification was originally granted under the Regression model, but later denied when the Regression model proved impossible to conduct, because before-and-after data was unavailable. Blue Diamond, 2014 U.S. Dist. LEXIS 173789, *29-44 (N.D. Cal. Dec. 15, 2014). The expert further proposed a “hedonic regression analysis” considering the prices of the defendant's products (with the labeling claim) as well as prices of comparable products (with or without the labeling claim) to isolate the impact of challenged label claims on the defendant's products. The model was rejected because it failed to control for advertising and for value of the defendant's brand name. Id. A few months after decertification, the court granted the plaintiffs' motion for voluntary dismissal with prejudice. Blue Diamon , 2015 U.S. Dist. LEXIS 16955 (N.D. Cal. Feb. 11, 2015).

In another more recent case against ConAgra, however, this time challenging the “100 Percent Natural” label in Wesson cooking oil containing genetically modified organisms (GMOs), the Central District of California did grant class certification to the plaintiffs. In re ConAgra Foods, Inc., 2015 U.S. Dist. LEXIS 24971 (C.D. Cal. Feb. 23, 2015). The court had previously denied class certification, but then allowed the plaintiffs to amend. In reconsidering class certification, the court held that the hedonic Regression model alone still failed to satisfy Comcast. Id. at 238-39. The Regression model coupled with a conjoint analysis, however, did satisfy the Comcast requirement of tying damages to the liability theory.

The conjoint analysis proposed the use of customer surveys to segregate the price premium specifically attributable to the customer's belief that “100% Natural” meant that the product contained no GMOs. Id. at 239. The price premium, calculated by one expert using regression analysis, could then be multiplied by the percentage from the conjoint analysis to reach the damages corresponding to the plaintiffs' theory of liability, that by placing a “100% Natural” label, the defendant misled customers into believing the product contained no GMOs. Id.

3. Full Refund Model

The Full Refund model proposes simply refunding the entire purchase price of the challenged product. However, this model, also discussed in Dole , does not satisfy the purpose of restitution, and amounts to a windfall to the consumer. As the courts in POM and Brazil explained, the model erroneously presumes that the consumers received no benefit from the product purchased. POM , 2014 U.S. Dist. LEXIS 40415; Brazil, 2014 U.S. Dist. LEXIS 74234, at *53. See alsoCaldera v. J.M. Smucker Co., 2014 U.S. Dist. LEXIS 53912, 11 (C.D. Cal. Apr. 15, 2014) (“Awarding class members a full refund would not account for these benefits conferred upon class members”); ConAgra, 2014 U.S. Dist. LEXIS 81292, 72 (“Return of the full retail or wholesale prices is not a proper measure of restitution, as it fails to take into account the value class members received by purchasing the product”).

If the plaintiffs' liability theory is that they received no benefit from the product, however, a full refund model may be appropriate. In re Scotts EZ Seed Litig., 304 F.R.D. 397; 2015 U.S. Dist. LEXIS 9116, 30 (S.D.N.Y. 2015) (“This model matches plaintiffs' first theory of liability ' that EZ Seed does not grow grass, and is thus valueless.”); Allen v. Hyland's Inc., 300 F.R.D. 643, 671 (C.D. Cal. 2014) (“Plaintiffs' theory is that the [homeopathic] products are entirely ineffective and thus any purported “benefit” customers experience can be attributed to the placebo effect”). With regard to cosmetic labeling, a disgorgement of profits model has been accepted as a plausible model at the class certification stage. Brown, 2014 U.S. Dist. LEXIS 161036 at *54 (the first model proposed by the expert was “a restitutionary measure that calculates the amount of revenue and profit Hain Celestial has wrongfully gained from sales of its [allegedly] mislabeled products, meaning, the Jason and Avalon products in this case.”).

Conclusion

Although certain class-wide damages models have been proposed in food and cosmetic labeling class actions, thus far a methodology capable of withstanding scrutiny has been elusive. The Regression model has been, at least in preliminary phases, the most successful at establishing class-wide predominance. The challenge for plaintiffs has been quantifying the effect of the alleged unlawful conduct on the market price of the product purchased by the class; that is, the dollar value that consumers give to the “All Natural” label. This undoubtedly requires expert testimony and control of all variables that may affect the consumer's choice of one brand over another. And, even where a Regression model in theory may satisfy the Comcast standard of tying damages to the plaintiffs' liability theory, it has in practice proved difficult to execute.

Thus, the Full Refund model in the end, while not as attractive to plaintiffs' counsel, may be the plaintiffs' last resort for avoiding dismissal of their claims. We anticipate that plaintiffs will continue to hone their damages' theories and attempt to provide expert testimony to further reduce the likelihood that their labeling cases will fail at certification time.

Vivian M. Quinn , a member of this newsletter's Board of Editors, is a partner and Lynette Nogueras-Trummer is a senior associate at Nixon Peabody LLP. They concentrate their practice on commercial and consumer product litigation, including work for the food and beverage industry.

Numerous class action complaints have been filed recently, challenging labels' claims that the products are “all natural.” Similarly, cosmetics products may be challenged for “organic” or “natural” labeling claims. The food-labeling cases allege that the “all-natural” label claims are deceptive because the products contain allegedly synthetic ingredients such as ascorbic acid (Vitamin C), citric acid, potassium citrate or calcium chloride. One of the many roadblocks for plaintiffs in establishing class certification in these cases has been finding a reliable damages methodology that can withstand scrutiny. This article summarizes several proposed models, and how the courts have dealt with them.

Damages Models

In Comcast Corp. v. Behrend , 133 S. Ct. 1426, 1433 (U.S. 2013), the Supreme Court held that at the class certification stage, the plaintiffs' damages model must be consistent with its liability case and “susceptible of measurement across the entire class.” Id . at 1433-34. The challenge in finding a methodology is to isolate a proper measure of what amount is necessary to compensate the purchaser for the difference between the product as labeled and the product actually received by the consumer. Three models have been discussed for trying to quantify this alleged damage in “All Natural” label cases.

1. The Price Premium Model

The Price Premium model allows as restitution the entire difference between the price of the product in question and that of allegedly comparable products without the contested label language. This model was rejected in Northern District of California Brazil v. Dole Packaged Foods, LLC, 12-CV-01831-LHK ( Dole ), 2014 U.S. Dist. LEXIS 74234 (N.D. Cal. May 30, 2014) and Werdebaugh v. Blue Diamond Growers (Blue Diamond , 2014 U.S. Dist. LEXIS 71575 (N.D. Cal. May 23, 2014). The model fails to control for advertising, brand name and other factors, and can conflate the alleged mislabeling with the value of the brand.

In Blue Diamond, the plaintiff challenged two statements in almond milk products' labels as allegedly false and misleading: 1) listing the sweetener used in the almond milk products as “evaporated cane juice” instead of “sugar”; and 2) including the statement “All Natural” when the products contained potassium citrate. The court in Blue Diamond explained that the expert had “no way of linking the price difference, if any, to the allegedly unlawful or deceptive label statements or controlling for other reasons why allegedly comparable products may have different prices.” The Price Premium model failed the Comcast test by simply calculating the price difference “without tying it to a legal theory.”

Moreover, the model did not account for factors such as quality or brand loyalty, which might lead consumers to prefer the challenged brand name product over other identical products. See also In re POM Wonderful, LLC, 2014 U.S. Dist. LEXIS 40415, 1 (C.D. Cal. Mar. 25, 2014) ( POM ) (challenging certain pomegranate juice products for claims of certain health benefits); Astiana v. Ben & Jerry's Homemade, Inc., 2014 U.S. Dist. LEXIS 1640 (N.D. Cal. Jan. 7, 2014) (Ben & Jerry's) (alleging that product labels of “All Natural” were misleading as the cocoa powder in ice cream is made with synthetic processing agents).

In the POM case, the court concluded that “where, as here, consumers buy a product for myriad reasons, damages from the alleged misrepresentation will not possibly be uniform or amenable to class proof.” In Ben & Jerry' , the court rejected the Price Premium model, equating it to a “fraud on the market” model that could not work without first establishing an efficient market that “prices a good on the basis of all available material information.” The court further found the model unreliable for measuring class-wide damages, finding speculative plaintiffs' assertions that an “All Natural” ice cream label requires a $0.50 to $0.75 premium over other ice creams. Ben & Jerry's successfully argued that the retail market price varies based on a number of factors that had nothing to do with a product's “All Natural” status, such as the nature and location of the retailer. But see Brown v. Hain Celestial Group, Inc. , 2014 U.S. Dist. LEXIS 161036, 54 (N.D. Cal. Nov. 14, 2014) (Price Premium model alloed and class certification granted to plaintiffs challenging “organic” labeling cosmetics).

In Jones v. Conagra Foods, Inc., 2014 U.S. Dist. LEXIS 81292, 71 (N.D. Cal. June 13, 2014) ( ConAgra ), the plaintiff challenged “100 percent natural” or “free of artificial ingredients & preservatives” label claims on Hunt's canned tomato products that contained citric acid or calcium chloride, as well as similar claims with respect to PAM cooking spray, and claims regarding “natural antioxidants” in Swiss Miss hot cocoa products. The expert proposed a “benefit of the bargain approach,” taking a “comparable product” and comparing the price between it and the allegedly mislabeled Hunt's product. The court held that this approach was “deeply flawed.” Id. at *73-75. The comparators either contained a “100% natural” label statement as well, or were generic brands being compared with brand-name products.

2. The Regression Model

Although the Regression model has enjoyed somewhat more success, it is still strenuously challenged at class certification and has proven difficult for plaintiffs to attribute certain damages to the allegations in the complaint. Plaintiffs claim that this model provides a more precise measure of damages because it controls for factors commonly recognized as associated with sales such as “price of the product, prices of competing and complementary products, income, advertising, seasonality, and regional differences.” See Dole, 2014 U.S. Dist. LEXIS 74234, at *58. The model further considers sales of the products before and after the challenged labeling language. Id .

In ConAgra, the court rejected the Regression model, stating that “[expert] does not provide a clearly defined list of variables, he has not determined whether the data related to any or all of his proposed control variables exists, and he has not determined, or shown how he would determine, which competing and complementary products he would use.” 2014 U.S. Dist. LEXIS 81292, at *76-78. The ConAgra plaintiffs have appealed the denial of class certification to the Ninth Circuit Court of Appeals; thus, there may soon be circuit court guidance on this issue. In their brief to the Ninth Circuit, the plaintiffs argue that the court erred in requiring precise damages calculations at the class certification stage and erred in conflating damages with restitution. Appellants' Brief, Case No. 14-16327, Dkt. Entry 21 (9th Cir. Nov. 21, 2014). They also argue that the Regression model satisfies Comcast .

In Dole, the expert proposed to determine the damages from the alleged misrepresentations “by examining sales of the identified products before and after Dole placed the alleged misrepresentations on its product labels, using regression analysis to control for other variables that could otherwise explain changes in Dole's sales.” 2014 U.S. Dist. LEXIS 74234 at *58. The court held that the Regression model “provide[d] a means of showing damages on a class-wide basis through common proof,” thus “satisf[ying] the Rule 23(b)(3) requirement that common issues predominate over individual ones.” Id. The court then certified the proposed Damages Class, although the expert had not yet conducted the regression analysis, citing lack of discovery from Dole. Despite this early plaintiff's victory, in November 2014, the district court later decertified the damages class. Dole, 2014 U.S. Dist. LEXIS 157575, 1 (N.D. Cal. Nov. 6, 2014).

In Dole, the expert was unable to isolate the damages attributable to Dole's alleged misbranding. Most of the product labels had not changed during the class period, making the before-and-after analysis impossible. The methodology employed by the expert was ultimately unable to control for other variables such as advertising, and the expert failed to verify whether the labels of non-Dole products contained “all natural” claims. The analysis thus failed the Comcast test that the damage be related to the liability claim ' specifically, it failed to measure only those damages attributable to Dole's use of the “All Natural Fruit” label statements. Id.

The same occurred in Blue Diamond, where class certification was originally granted under the Regression model, but later denied when the Regression model proved impossible to conduct, because before-and-after data was unavailable. Blue Diamond, 2014 U.S. Dist. LEXIS 173789, *29-44 (N.D. Cal. Dec. 15, 2014). The expert further proposed a “hedonic regression analysis” considering the prices of the defendant's products (with the labeling claim) as well as prices of comparable products (with or without the labeling claim) to isolate the impact of challenged label claims on the defendant's products. The model was rejected because it failed to control for advertising and for value of the defendant's brand name. Id. A few months after decertification, the court granted the plaintiffs' motion for voluntary dismissal with prejudice. Blue Diamon , 2015 U.S. Dist. LEXIS 16955 (N.D. Cal. Feb. 11, 2015).

In another more recent case against ConAgra, however, this time challenging the “100 Percent Natural” label in Wesson cooking oil containing genetically modified organisms (GMOs), the Central District of California did grant class certification to the plaintiffs. In re ConAgra Foods, Inc., 2015 U.S. Dist. LEXIS 24971 (C.D. Cal. Feb. 23, 2015). The court had previously denied class certification, but then allowed the plaintiffs to amend. In reconsidering class certification, the court held that the hedonic Regression model alone still failed to satisfy Comcast. Id. at 238-39. The Regression model coupled with a conjoint analysis, however, did satisfy the Comcast requirement of tying damages to the liability theory.

The conjoint analysis proposed the use of customer surveys to segregate the price premium specifically attributable to the customer's belief that “100% Natural” meant that the product contained no GMOs. Id. at 239. The price premium, calculated by one expert using regression analysis, could then be multiplied by the percentage from the conjoint analysis to reach the damages corresponding to the plaintiffs' theory of liability, that by placing a “100% Natural” label, the defendant misled customers into believing the product contained no GMOs. Id.

3. Full Refund Model

The Full Refund model proposes simply refunding the entire purchase price of the challenged product. However, this model, also discussed in Dole , does not satisfy the purpose of restitution, and amounts to a windfall to the consumer. As the courts in POM and Brazil explained, the model erroneously presumes that the consumers received no benefit from the product purchased. POM , 2014 U.S. Dist. LEXIS 40415; Brazil, 2014 U.S. Dist. LEXIS 74234, at *53. See alsoCaldera v. J.M. Smucker Co., 2014 U.S. Dist. LEXIS 53912, 11 (C.D. Cal. Apr. 15, 2014) (“Awarding class members a full refund would not account for these benefits conferred upon class members”); ConAgra, 2014 U.S. Dist. LEXIS 81292, 72 (“Return of the full retail or wholesale prices is not a proper measure of restitution, as it fails to take into account the value class members received by purchasing the product”).

If the plaintiffs' liability theory is that they received no benefit from the product, however, a full refund model may be appropriate. In re Scotts EZ Seed Litig., 304 F.R.D. 397; 2015 U.S. Dist. LEXIS 9116, 30 (S.D.N.Y. 2015) (“This model matches plaintiffs' first theory of liability ' that EZ Seed does not grow grass, and is thus valueless.”); Allen v. Hyland's Inc. , 300 F.R.D. 643, 671 (C.D. Cal. 2014) (“Plaintiffs' theory is that the [homeopathic] products are entirely ineffective and thus any purported “benefit” customers experience can be attributed to the placebo effect”). With regard to cosmetic labeling, a disgorgement of profits model has been accepted as a plausible model at the class certification stage. Brown, 2014 U.S. Dist. LEXIS 161036 at *54 (the first model proposed by the expert was “a restitutionary measure that calculates the amount of revenue and profit Hain Celestial has wrongfully gained from sales of its [allegedly] mislabeled products, meaning, the Jason and Avalon products in this case.”).

Conclusion

Although certain class-wide damages models have been proposed in food and cosmetic labeling class actions, thus far a methodology capable of withstanding scrutiny has been elusive. The Regression model has been, at least in preliminary phases, the most successful at establishing class-wide predominance. The challenge for plaintiffs has been quantifying the effect of the alleged unlawful conduct on the market price of the product purchased by the class; that is, the dollar value that consumers give to the “All Natural” label. This undoubtedly requires expert testimony and control of all variables that may affect the consumer's choice of one brand over another. And, even where a Regression model in theory may satisfy the Comcast standard of tying damages to the plaintiffs' liability theory, it has in practice proved difficult to execute.

Thus, the Full Refund model in the end, while not as attractive to plaintiffs' counsel, may be the plaintiffs' last resort for avoiding dismissal of their claims. We anticipate that plaintiffs will continue to hone their damages' theories and attempt to provide expert testimony to further reduce the likelihood that their labeling cases will fail at certification time.

Vivian M. Quinn , a member of this newsletter's Board of Editors, is a partner and Lynette Nogueras-Trummer is a senior associate at Nixon Peabody LLP. They concentrate their practice on commercial and consumer product litigation, including work for the food and beverage industry.