Conrad v. Jimmy John's Franchise, LLC

Full title: DONALD CONRAD, On Behalf of Himself & All Others Similarly Situated…

Court: UNITED STATES DISTRICT COURT FOR THE SOUTHERN DISTRICT OF ILLINOIS

Date published: Feb 16, 2021

Case No. 18-CV-00133-NJR (S.D. Ill. Feb. 16, 2021)

Opinion

Case No. 18-CV-00133-NJR

02-16-2021

DONALD CONRAD, On Behalf of Himself & All Others Similarly Situated, Plaintiff, v. JIMMY JOHN'S FRANCHISE, LLC, JIMMY JOHN'S ENTERPRISES, LLC, and JIMMY JOHN'S LLC, Defendants.

ROSENSTENGEL, Chief Judge

MEMORANDUM AND ORDER ROSENSTENGEL, Chief Judge :

This is an antitrust case brought by a putative class of current and former Jimmy John's employees. The plaintiffs allege that a "No-Poach Provision" in Jimmy John's Franchise Agreement effectively prohibited employees from switching between locations with different owners, stifling competition in the labor market through an unlawful exercise of monopsony power under Section 1 of the Sherman Act.¹

1 Quotations referring to the "No-Poach Agreement" or the "challenged conduct" have been altered to say "No-Poach Provision."

In support of their Motion for Class Certification, the plaintiffs rely on an expert report prepared by Dr. Hal Singer, who concludes that "common methods and evidence demonstrate Anticompetitive Effects and Common Impact to the Class, and that common methods and evidence demonstrate Aggregate Damages attributable to the No-Poach Provision totaling [Redacted] ." (Singer Report 2, Doc. 115-3; Pls.' Mem. in Support of Class Cert. 3-4, Doc. 115).

Naturally, Jimmy John's argues that Dr. Singer's report is unreliable. It submitted a rebuttal report prepared by its own expert, Dr. Janusz Ordover, who suggests that Dr. Singer's economic models suffer from several conceptual and statistical flaws. (Ordover Report 5, Doc. 133-56). It also submitted a rebuttal report prepared by another expert, Dr. Justin McCrary, who suggests that Dr. Singer's conclusions conflict with the plaintiffs' monopsony theory. (McCrary Report 9, Doc. 133-57). The litigants moved to exclude each other's expert reports.

BACKGROUND

A. The Allegations

From at least 2014 to 2018, Jimmy John's included a No-Poach Provision in its Franchise Agreement. (Am. Compl. 1, Doc. 75). By signing the Franchise Agreement, franchisees agreed not to "recruit as a partner or investor/owner, or hire as an employee, any person then employed, or who was employed within the preceding twelve (12) months, by [Jimmy John's], any of [Jimmy John's] affiliates, or a franchisee . . . without obtaining the existing or former employer's prior written permission . . . ." (Franchise Agreement 24, Doc. 115-17) (emphasis added).

The plaintiffs—a putative class of all Jimmy John's employees who worked there "at any time between January 24, 2014 to July 12, 2018"—allege that the No-Poach Provision "reflects a naked restraint of competition" under Section 1 of the Sherman Act, 15 U.S.C. § 1. (Am. Compl. at 1). More specifically, they allege that the No-Poach Provision had the effect of suppressing wages and stifling worker mobility, leading to class-wide injury and damages. (Pls.' Mem. in Support of Class Cert. at 1).

B. The Plaintiffs' Expert: Dr. Singer

Dr. Singer is "a managing director at Econ One, a senior fellow at the George Washington Institute of Public Policy, and an adjunct professor at the McDonough School of Business at Georgetown University, where [he] teach[es] advances pricing to M.B.A. candidates." (Singer Report at 2). He is also "an applied microeconomist with an emphasis on industrial organization and regulation" who has "testified before Congress on the interplay between antitrust and sector-specific regulation." (Id. at 3).

Dr. Singer offers an opinion purporting to show that common proof that can be used to establish that the No-Poach Provision suppressed compensation (i.e., antitrust impact) for all Jimmy John's employees. (See id. at 41-42). He begins by describing the basic monopsony theory underlying the plaintiffs' suit. (Id. at 5). "[A] firm possesses monopsony power if it wields market power over factors of production such as labor. The exercise of monopsony power in the labor market harms competition by suppressing wages and employment below competitive levels." (Id. at 5). "'A labor monopsony exists when lack of competition in the labor market enables employers to suppress the wages of their workers.'" (Id. at 6) (citing Ioana Marinescu & Eric Posner, A Proposal to Enhance Antitrust Protection Against Labor Market Monopsony 2, Roosevelt Inst. (2018) ("While [monopoly power] has been the core focus of antitrust practitioners since the inception of the Sherman Act, [monopsony power] has generally attracted scant attention from public or private antitrust enforcement, despite the fact that a worker who receives a subcompetitive wage is likely harmed just as much, if not more, than a consumer paying a supracompetitive price . . . .")).

Dr. Singer contends that the No-Poach Provision was evidence of Jimmy John's monopsony power: "[B]y restraining labor market mobility, the No-Poach Provision gave individual Jimmy John[']s stores more labor market power than they otherwise would have enjoyed." (Id. at 27) (emphasis in original). In other words, "in the absence of the No-Poach Provision, the wages of Jimmy John's workers could be bid up by competition among independently-owned stores . . . for the workers' specialized skills. The No-Poach Provision prohibits this form of labor market competition by suppressing labor mobility." (Id. at 15).

To test this hypothesis, Dr. Singer "developed an econometric model to compare (1) the wages paid to Class Members when the No-Poach Provision was in effect with (2) the wages paid to Class Members after the No-Poach Provision was officially halted in mid-2018, . . . while controlling for other factors that might explain movements in wages." (Id. at 28-29). This is known as a wage regression. (See id.).

Dr. Singer obtained the "hourly compensation data from Weekly Sales Reports ('WSRs'), which are regularly submitted by Jimmy John's Franchisees to corporate headquarters. The WSRs span all 44 states with Jimmy John's stores[] and include both Franchisees and corporate-owned stores. The WSRs include daily hours worked and compensation for over 615,000 employees working in over 2,800 stores. The WSRs include data for three categories of Class Members between January 2014 and June 2019: (1) drivers, (2) in-shop labor, and (3) managers." (Id. at 30).

Dr. Singer acknowledges that "the effect of removing the No-Poach Provision may not have manifested itself immediately," possibly because "the economic ramifications . . . would have taken time to work through the thousands of independently-owned Jimmy John's restaurants, as newfound labor mobility would have been increasingly discovered and leveraged by Class Members." (Id. at 36-37). He therefore conducted another test for each month after the removal of the No-Poach Provision in June 2018 to test whether the putative class members' compensation increased as a result. (See id.).

Several control variables were in the regressions. For example, "'fixed effect' variables . . . control for three types of labor (drivers, in-shop labor, and managers). The regression also includes fixed effects for individual stores. This controls for all store-specific factors that are fixed over time" and "provide[s] flexible controls for local economic conditions at the state level, county level, and below the county level . . . ." (Id. at 31-32). There is also a fixed-effect variable for each of the 2,800 stores in the WSRs. (Id. at 32). Similarly, Dr. Singer "include[s] control variables at the level of the individual employee, in the form of more than 600,000 individual employee fixed effects," which "control[] for any employee-specific characteristics that are fixed over time." (Id.).

Dr. Singer then uses two methods to test for common impact. (See id. at 45). The first employs the wage regression described above to "show that at least between 87 and 91 percent of Class Members suffered antitrust injury." (Id.). Again, the regression "compares the compensation that each Class Member actually received to the compensation they would have received in the absence of the No-Poach Provision, as determined by [the] regression model." (Id. at 46). "The 87 percent estimate is derived from Column (3) of Table 2 [below], and therefore ignores the likelihood that Class Member compensation took time to recover after [June 2019]. The 91 percent estimate is derived from Column (3) of Table 3 [below], and it uses the most recently available estimate of Class Member wage recovery (June of 2019) to estimate Common Impact." (Id. at 46-47).

TABLE 2: REGRESSIONS (WITHOUT EQUILIBRIUM WAGE ADJUSTMENT)

Dependent Variable: ln(Hourly Wage)
Explanatory Variable	(1)	(2)	(3)
PreAG	-0.038***(0.000)	-0.038***(0.000)	-0.037***(0.000)
In-Shop Labor	0.147***(0.000)	0.149***(0.000)	0.192***(0.000)
Managers	1.113***(0.000)	1.117***(0.000)	0.539***(0.000)
ln(Effective Minimum Wage)	0.701***(0.000)	0.502***(0.000)	0.304***(0.000)
Local Unemployment Rate	-0.012***(0.000)	0.006***(0.000)	-0.001***(0.000)
ln(Local Income Per Capita)	0.052***(0.000)	0.344***(0.000)	0.220***(0.000)
Linear Time Trend	-0.073***(0.000)	-0.023***(0.000)	0.169***(0.000)
Nonlinear Time Trend	0.001***(0.000)	0.000***(0.000)	-0.006***(0.000)
Constant	0.794***(0.000)	-2.353***(0.000)	-1.994***(0.000)
Store Fixed Effects	No	Yes	No
Worker Fixed Effects	No	No	Yes
Observations	14,973,000	14,973,000	14,973,000
R-Squared	40.7%	43.5%	84.8%

(Id. at 36).

TABLE 3: REGRESSIONS WITH EQUILIBRIUM WAGE ADJUSTMENT

Dependent Variable: ln(Hourly Wage)
Explanatory Variable	(1)	(2)	(3)
PostAG Jul 2018	0.027***(0.000)	0.023***(0.000)	0.024***(0.000)
PostAG Aug 2018	0.029***(0.000)	0.028***(0.000)	0.029***(0.000)
PostAG Sept 2018	0.025***(0.000)	0.031***(0.000)	0.034***(0.000)
PostAG Oct 2018	0.027***(0.000)	0.033***(0.000)	0.038***(0.000)
PostAG Nov 2018	0.034***(0.000)	0.043***(0.000)	0.048***(0.000)
PostAG Dec 2018	0.043***(0.000)	0.049***(0.000)	0.053***(0.000)
PostAG Jan 2019	0.061***(0.000)	0.047***(0.000)	0.062***(0.000)
PostAG Feb 2019	0.068***(0.000)	0.06***(0.000)	0.074***(0.000)
PostAG Mar 2019	0.063***(0.000)	0.056***(0.000)	0.069***(0.000)
PostAG Apr 2019	0.060***(0.000)	0.065***(0.000)	0.073***(0.000)
PostAG May 2019	0.067***(0.000)	0.070***(0.000)	0.080***(0.000)
PostAG Jun 2019	0.075***(0.000)	0.068***(0.000)	0.084***(0.000)
Labor In Shop	0.147***(0.000)	0.149***(0.000)	0.192***(0.000)
Labor Manager	1.114***(0.000)	1.117***(0.000)	0.539***(0.000)
ln(Effective Minimum Wage)	0.702***(0.000)	0.502***(0.000)	0.306***(0.000)
Local Unemployment Rate	-0.012***(0.000)	0.006***(0.000)	-0.001***(0.000)
ln(Local Income Per Capita)	0.051***(0.000)	0.330***(0.000)	0.186***(0.000)
Linear Time Trend	-0.039***(0.000)	0.010***(0.000)	0.226***(0.000)
Nonlinear Time Trend	0(0.799)	-0.002***(0.000)	-0.009***(0.000)
Constant	0.562***(0.000)	-2.44***(0.000)	-2.008***(0.000)
Store Fixed Effects	No	Yes	No
Worker Fixed Effects	No	No	Yes
Observations	14,973,000	14,973,000	14,973,000
R-Squared	40.7%	43.6%	84.8%

(Id. at 38).

The second method builds on the first but asks one further question: "whether there is class-wide evidence of a compensation structure that would transmit the artificially reduced compensation . . . broadly across the class." (Id. at 47-48) (emphasis added). Dr. Singer based this theory on record evidence purporting to show that Jimmy John's "maintained an explicit, formulaic compensation structure for its managers . . . and recommended that the independently owned franchise stores follow" suit. (Id. at 54). He then "performed regressions to . . . measure the extent to which an increase in hourly compensation for Class Members generally is statistically associated with an increased in compensation for an individual Class Member." (Id. at 48-49).

Specifically, I estimated regressions in which the dependent variable was set equal to an individual Class Member's weekly pay rate, and the independent variable was set equal to either: (A) the average hourly wage paid to all other workers of that Class Member's same type (Labor In Shop, Driver, or Manager) in that year; or (B) the average hourly wage paid to all other workers of the Class Member's same type in the prior year.

(Id. at 49) (emphasis in original). He also "perform[ed] a similar analysis" that "include[ed] each of the control variables used in [his] impact regression." (Id. at 50). Finally, he tested whether "employee wages move together at the local (county) level" by repeating the regressions and "using the wages of other Class Members within a given county to predict the wages of individual employees in that county." (Id. at 51).

The results: "changes in compensation are broadly shared across [all] Class Members, both within a given time period and across different time periods." (Id. at 53) (emphasis added). "For example, . . . a ten percent increase in the compensation of all other employees is statistically associated with a 6.9 percent increase in the compensation of an individual Class Member in that same job category and county." (Id. at 52). This, Dr. Singer asserts, "confirm[s] that gains (or losses) in Class Members' hourly compensation are shared broadly across Class Members both within and across years." (Id. at 51). In other words, Dr. Singer contends that his second method shows that all putative class members (rather than 87 to 91 percent) were injured by the No-Poach Provision because "there is class-wide evidence of a compensation structure that would transmit the artificially reduced compensation (found by the first [method]) broadly across the Class." (Id. at 47).

Based on these tests, Dr. Singer "conclude[s] that all or almost all Class Members can be shown to have suffered antitrust injury across the two methods of proving common impact." (Id. at 46). He estimates aggregated damages of about [Redacted] : "the product of (1) the generalized wage effect from the regression model . . . and (2) the aggregate compensation paid to Class Members while the No-Poach Provision was in effect." (Id. at 56).

Finally, Dr. Singer refutes two potential "procompetitive" justifications that he anticipates Jimmy John's will advance. (Id. at 57). He calls the first justification the "Franchise Investment Efficiency": The No-Poach Provision was used to "encourage investment in Jimmy John's brand." (Id.). Aside from a lack of evidence to support such a justification, Dr. Singer contends that, even if the No-Poach Provision "made Jimmy John's more attractive to investors by cutting labor costs below what they otherwise would have been, this does not diminish the anticompetitive effects of the No-Poach Provision for workers" through suppressed wages. (Id. at 60) (emphasis in original).

Dr. Singer calls the second justification the "Free-Rider Efficiency": Without the No-Poach Provision, "investments in training would invite poaching of a newly-trained worker by another Franchisee." (Id. at 60). "This defense," he asserts, is "an admission of monopsony power: It presupposes that Jimmy John's exploited the No-Poach Provision to suppress Class Member compensation below what it would be otherwise." (Id. at 61) (emphasis added). At any rate, he again points to a lack of evidence suggesting that (a) "the No-Poach Provision created Efficiencies by eliminating Free Riding"; (b) "the cessation of the No-Poach Provision has any negative effect on Jimmy John's training of its employees"; or (c) "Jimmy John's ceased or substantially cut back on employee training as a result of the cessation of the No-Poach Provision." (Id.).

C. Jimmy John's Experts

(i) Dr. Ordover

Dr. Ordover is "Emeritus Professor of Economics and former Director of the Masters in Economics Program at New York University, where [he] taught from 1973 to 2015. From 1991-1992, [he] served as Deputy Assistant Attorney General for Economics at the Antitrust Division of the United States Department of Justice . . . , co-drafting the 1992 U.S. Department of Justice and Federal Trade Commission Horizontal Merger Guidelines." (Ordover Report at 1). Who's Who Legal named him "Competition Economist Individual Expert of the Year" in both 2015 and 2016, and he was named "Economist of Year" in 2011 by the Global Competition Review. (Id. at A-3).

Unlike Dr. Singer, Dr. Ordover concludes that "economic injury (in this case, i.e., whether wages were suppressed due to the No-Poach Provision) cannot be shown for all or nearly all members of the proposed class using common evidence and that highly individualized investigations would have to be undertaken to determine whether any putative class member suffered antitrust injury." (Id. at 5). Rather, "Dr. Singer has ignored many sound economic reasons why most putative class members would not have been harmed by the No-Poach Provision, and, as such, his highly aggregated statistical analyses—which suffer from many other conceptual, statistical, and data errors—are of no value in gauging class-wide impact." (Id.) (emphasis in original).

Dr. Ordover begins by addressing "a basic data error" in Dr. Singer's wage regressions "that results in inflated estimates of the alleged average wage suppression and the share of putative class members that suffered injury." (Id. at 5). As discussed, Dr. Singer compared (1) employee wages while the No-Poach Provision was in effect with (2) wages after the No-Poach Provision was removed. He accomplished this using wage data gathered from the WSRs that were submitted by franchisees to Jimmy John's, the franchisor. For most employees, the WSRs consistently reflect that they were paid per-hour. Others consistently reflect that they were paid per-shift. And for a few employees, like the named plaintiff, the WSRs are inconsistent either way. Dr. Ordover asserts that Dr. Singer glossed over that last difference. For example, "he wrongly treats Mr. Conrad as being paid on an hourly basis throughout the time period—so Mr. Conrad's supposed 'hourly' wage rate increased from [Redacted] to [Redacted] in Dr. Singer's analysis over a period of a few days."² (Id. at 13). He made "the same error for other managers that were paid a salary, which was recorded on a per-shift basis in the WSR." (Id. at 14). As a result, Dr. Singer's wage regressions led "to significantly inflate[d] estimates of the alleged anticompetitive harm and common impact he asserts were caused by the No-Poach Provision." (Id.)

2 This wage increase reflects the fact that Conrad was promoted from in-shopper to manager. (Ordover Report at 12). The [Redacted] is per-hour, and the [Redacted] is per-shift. (Id.).

Suppose a researcher wished to predict the height of a set of women by a common set of independent variables, including the height of their mothers. One would generally expect that shorter mothers would, on average, have shorter daughters, while taller mothers would have taller daughters. Assume that the researcher had collected data on mothers' heights that were measured in inches, but, for some of the daughters, their heights were measured in inches and for others they were measured in feet. The result of this mistake would be that predicted heights would be too low for those daughters whose heights are measured in inches and too high for those whose heights were measured in feet. In other words, the model will not have accurately captured the effect of mothers' heights on daughters' heights due to this data error. Dr. Singer's regression models suffer from this most basic mistake. Here, the result of Dr. Singer's mistake is that his wage regression models predict wage rates for salaried managers whose pay is recorded on a per-shift basis (like those daughters whose height is measured in inches) as too low and predicts wage rates of managers whose pay is recorded on a per-hour basis (like those daughters whose height is measured in feet) as too high—meaning that his model does not accurately explain the effect of the No-Poach Provision on wage rates.

(Id. at 15).

To remedy this, Dr. Ordover separated (or "unpooled") "the WSR data for the manager job category into two separate manager job categories: (i) those who appear to be paid on an hourly basis; and (ii) those who appear to be paid a salary." (Id.). He then repeated the same wage regression conducted by Dr. Singer. (Id.). "Correcting for [Dr. Singer's] data error reduced [the] estimate of aggregate damages [from [Redacted] ] to [Redacted] , a reduction of roughly 66 percent." (Id. at 23). This figure, in Dr. Ordover's opinion, goes against anticompetitive behavior:

While the real wages calculated by Dr. Singer increased over time, the increase in real wage rates after the end of the putative class period is comparable to the increases in those wage rates during the putative class period. Indeed, one would expect, if there was some persistent anticompetitive suppression in wage rate of the putative class members, that there would be some noticeable discontinuity (or jump) in the wage rates around the time that the alleged anticompetitive conduct ceased.

(Id. at 17). Simply "separating Dr. Singer's regression by manager pay type results in a finding that managers paid on an hourly basis had an average wage suppression of approximately two percent, while salaried managers suffered no suppression at all." (Id. at 21).³

3 At any rate, Dr. Ordover asserts that the WSRs fail to track "important aspects of compensation that are allegedly affected by the No-Poach Provision," such as tips for drivers and bonuses for managers, "mak[ing] it difficult (if not impossible) to determine whether the No-Poach Provision reduced total compensation for many putative class members." (Ordover Report at 43-45).

[Redacted] (Id. at 18).

[Redacted] (Id. at 20).

What's more, Dr. Ordover asserts that Dr. Singer ignored several "sound economic reasons why putative class members could not have been injured from the No-Poach Agreement." (Id.). For one, Dr. Ordover contends that for the nearly 37 percent of Jimmy John's locations without a competing franchise within 10 miles, employees at those locations "are unlikely to have been impacted by the No-Poach Provision." (Id. at 27-28). Put differently, the lack of a rival franchise with a reasonable distance means that there is no one to "bid up [employee] wages." (Id. at 25).

Dr. Ordover also argues that the No-Poach Provision could not have impacted the "roughly 36 percent of employees [that] leave within four weeks . . . and [the] 56 percent [that] leave in twelve weeks or less" because "they would have been unlikely to receive any increases in compensation in the short time of their employment. (Id. at 30-32). Rather, "[a]s a matter of basic economics,. . . a smaller pool of potential hires would tend to decrease the supply of potential employees for any particular Jimmy John's branded store, which would tend to increase their wages." (Id. at 29) (emphasis added). For prospective employees who have never worked for a Jimmy John's restaurant, two franchisees would have to compete for that prospective employee by offering higher wages. (Id.).

Similarly, Dr. Ordover asserts that Dr. Singer's opinion "inappropriately ignored diversity and variation in any effect of the alleged conduct of the wages of putative class members over time and in different regions." (Id. at 34).

To demonstrate this, I estimate separate regressions for each year and for each state in which Jimmy John's franchise or corporate-owned stores operate and find that for some states and years, Dr. Singer's regressions, when unpooled, show that wages were higher in the putative class period than in his chosen benchmark, not lower.

(Id.) (emphasis in original). In nine States, for example, "(Arizona, Colorado, Iowa, Mississippi North Carolina, New York, Oregon, South Carolina and Washington) wages were higher in the putative class period than in the benchmark period[;] and for another four states (the District of Columbia, Indiana, Kansas, and New Jersey) wages were not statistically significantly lower in the putative class period than they were in the benchmark period." (Id. at 35-36). These 13 States "account for nearly 143,000 putative class members—or 25 percent of the total putative class." (Id. at 36). Moreover, in Delaware, where there is "only a single franchise operator in the entire state, putative class members most likely could not have benefited from increased competition for their labor due to the removal of the No-Poach Provision." (Id.) (emphasis in original). "This indicates that, according to Dr. Singer's model, large portions of the putative class suffered no injury as a result of the No-Poach Provision and that Dr. Singer's average wage suppression regression model is a flawed and unreliable means of measuring anticompetitive effects and common impact." (Id. at 34) (emphasis in original).

In Dr. Ordover's opinion, "after accounting for the prior three flaws, over 85 percent of putative class members were not injured based on Dr. Singer's methodology for gauging common impact." (Id. at 5, 92).

Next, Dr. Ordover objects to Dr. Singer's "'predictive' approach to gauging common impact [that] is prone to 'false positives'—the finding of injury where none could logically exist." (Id. at 6). As a conceptual matter, because "Dr. Singer assumes that a putative class member was impacted if her wages were suppressed in a[t] least one week," Dr. Ordover contends that "Dr. Singer's model allows an employee paid on a bi-weekly basis to have suffered impact during one week of his two-week pay period, but not the other. Such an approach to gauging antitrust harm is particularly flawed in this matter, since economists typically think of wages as 'sticky,' and less volatile than the price of goods." (Id. at 51).

Recall also that Dr. Singer's second method for proving antitrust impact was to determine whether there is "class-wide evidence of a compensation structure that would transmit the artificially reduced compensation . . . broadly across the class." (Singer Report at 47-48) (emphasis added). Dr. Ordover contends that this approach is also "seriously flawed and mis-specified and contradicted by the very wage data on which [Dr. Singer] relies." (Ordover Report at 6). For instance, Dr. Ordover argues that "if the wage rates of Jimmy John's employees were all connected, as claimed by Dr. Singer, one should not observe some employees' wage rates declining when other employees' wage rates are increasing." (Id. at 57). Yet the data suggests that "in any given year, many counties experience an increase in wages and many experience a decrease in wages." (Id.). For that reason, "Dr. Singer's conjecture that the wages at Jimmy John's franchise and corporate-owned store[s] all move in lockstep throughout the U.S. is demonstrably false." (Id. at 56).⁴

4 Dr. McCrary similarly concludes that "[f]ranchisee testimony corroborates that starting wages are set competitively" with other QSRs. (See McCrary Report at 52).

Figure 9

Annual Changes in County-Level Wage Rates Used by Dr. Singer Do Not Support the

Alleged National Wage Structure

Jimmy John's County Average Wage - In-shopper

Image materials not available for display. (Id. at 58).

Moreover, Dr. Ordover suggests that Dr. Singer's compensation-structure method "cannot be used to test for common impact because [it] assume[s] that all Jimmy John's are subject to the same impact." (Id. at 58) (emphasis in original). "In other words, Dr. Singer's test . . . is under the assumption that all employees . . . are equally affected . . . by the wages of others in their job category. It is not a test of whether this equal-effect assumption is correct." (Id. at 59 n.105) (emphasis in original).

Dr. Singer's models do not allow for a more fine-grained (and likely more realistic) compensation structure where different individuals within a group would be affected differently by a policy (e.g., the No-Poach Provision), depending on where they are within the distribution of wages, their seniority, skill level, their job performance relative to others, and other relevant factors for a wage structure.

(Id. at 59). His "'test' therefore assumes the very answer it seeks to test and is irrelevant for demonstrating common impact." (Id. at 60).

What's more, Dr. Ordover contends that "Dr. Singer has not demonstrated that the No-Poach Provision caused any anticompetitive effects." (Id. at 6) (emphasis in original). In other words, "Dr. Singer confuses the fact that his . . . models find some correlation between lower wages and the No-Poach Provision with a finding that the No-Poach Provision caused the lower wages." (See id. at 66-69) (emphasis in original).

Similarly, Dr. Ordover asserts that "Dr. Singer's analysis is completely untethered from the economic theory of monopsony" because he "fails to define a relevant labor market and has not demonstrated that Jimmy John's . . . possess[es] monopsony power in any relevant market." (Id. at 6). "Dr. Singer assumes the No-Poach Provision increased monopsony power because the No-Poach Provision would have no effect on compensation of putative class members if there was no monopsony power." (Id. at 70). Again, Dr. Ordover asserts that this contention conflates correlation with causation, this time because Dr. Singer failed to define the relevant market. (See id. at 70-71).

[I]n order to demonstrate that the No-Poach Provision increased Jimmy John's . . . monopsony power and reduced the wages of putative class members, Dr. Singer would have to establish: (i) the relevant labor market in which putative class members sold their services, (ii) that the No-Poach Provision allowed Jimmy John's . . . to increase their monopsony power in that relevant labor market; and (iii) that Jimmy John's . . . suppressed the wages of all or almost all putative class members by reducing its purchases of their services.

(Id. at 72). Whereas Dr. Singer assumes that the appropriate market definition is "employment at Jimmy John's franchise and corporate-owned stores nationwide," Dr. Ordover argues that the appropriate market definition also includes "at least other Quick Service Restaurants ('QSRs')." (Id. at 76). Given that "98 percent of Jimmy John's branded stores have at least ten other QSR brands within ten miles, with an average number of nearby brands of 53" and "an average number of QSR locations of nearly 257," Dr. Ordover contends that Jimmy John's alleged monopsony power is likely "quite small, and well below the levels that antitrust regulators consider likely to lead to adverse competitive effects." (Id. at 84).

Lastly, because the "available evidence does not indicate that the number of Jimmy John's . . . stores or the revenues of those stores decreased over time," Dr. Ordover contends that the No-Poach Provision did not cause "a reduction in output" consistent with the plaintiffs' monopsony theory. (Id. at 87).⁵

5 Dr. McCrary agrees. (See McCrary Report at 54) ("The monopsonist makes higher profit per unit of labor but they hire fewer workers (reduce labor demand) and restrict output in the process.").

(ii) Dr. McCrary

Dr. McCrary is "an economist with expertise in labor economics, antitrust, corporations, law and economics, economic modeling, and statistical methods, among other subjects." (McCrary Report at 3). He is currently the Paul J. Evanson Professor of Law at Columbia University, and he "previously held academic positions at the University of Michigan (2003-2007) and Berkeley (2008-2019)," where he has "taught courses on labor economics, antitrust, corporations, law and economics, and statistics to undergraduates, M.B.A., J.D., L.L.M., and Ph.D. students." (Id.).

Like Dr. Ordover, Dr. McCrary concludes that the "plaintiffs' claims of monopsony do not make economic sense within Jimmy John's franchise system." (Id. at 46). Dr. McCrary focuses on the day-to-day enforcement of the No-Poach Provision to determine (1) whether it was "enforced in a common manner across class members"; (2) whether it was indicative "of a vertical economic relationship between" the franchisees and Jimmy John's, the franchisor; (3) whether there are "potential competitive benefits of a vertical, intrabrand restraint as compared to an interbrand restraint"; (4) whether Jimmy John's, the franchisor, had "an economic incentive to orchestrate a monopsony conspiracy to suppress wages and labor demand"; and (5) whether the No-Poach Provision had "the potential to enhance competition." (Id. at 7) (emphasis added).

The first half of Dr. McCrary's report is "a summary of key concepts . . . on the economics of franchising and vertical restraints." (Id. at 10). He begins by discussing the difference between intrabrand restraints (good) with interbrand restraints (bad). The crux of the plaintiffs' theory is that "franchisees are independent horizontal competitors." (Id. at 14) (emphasis added). This, according to Dr. McCrary, "is inconsistent with . . . the economics of franchising." (Id.). Rather, franchisors, like Jimmy John's, have an interest in imposing "vertical restraints on individual franchisees to help ensure that the business decisions of each individual franchisee do not undermine the goals of the franchise or [the] brand as a whole." (Id.) (emphasis added). These intrabrand restraints, according to Dr. McCrary, facilitate interbrand competition because they ensure "higher-quality products" and "a higher-quality brand." (Id.).

With that in mind, Dr. McCrary contends that "[w]hen a firm with a popular brand (like Jimmy John's) decides to establish a franchising system, it turns over control of store operations to many individual franchisees whose goal—absent any restrictions from the franchisor—is to maximize the profit of their own stores." (Id. at 17) (emphasis in original). In other words, "brand and store interests can diverge." (Id.).

Consider a stylized example in which, for a given franchise organization, there are repeat customers that visit many franchises of the organization, but never (or rarely) visit the same store twice. In such an organization, no franchisee experiences the full benefits of investments in brand quality. Thus, it would be expected that in the absence of [intra]brand restrictions, franchisees would cut corners on investments in brand quality, for example by reducing food quality or customer service to reduce costs. For any individual franchisee, the short-run cost to the overall brand reputation of such deviations is small because the store is just one of many stores, while the benefit in profit can be significant. However, if all stores behave this way, then the overall quality of the brand will deteriorate in the long-run, which will harm all stores.

(Id. at 17-18). Put differently, "without intrabrand restraints requiring that individual franchisees protect the value and quality of the brand, each franchisee will underinvest in the brand[,] and the brand's value will deteriorate over time[,] harming the franchise, franchisees, consumers, and workers alike." (Id. at 18). This is known as freeriding. (Id.).

Similarly, another "incentive problem that franchises face is a problem sometimes referred to as encroachment: "[I]f a franchise clusters too many stores close together, then new locations may cannibalize sales from existing locations." (Id. at 18). Franchisors must therefore strike a balance "between expanding their own network with new locations, while ensuring that each franchisee remains committed to investing in the brand and does not try to undermine other local franchisees." (Id. at 19).

Dr. McCrary asserts that these two incentive problems—freeriding and encroachment—"are particularly acute for QSR franchises because consistency and uniformity of quality across all stores [are] essential to their business. . . . [I]f even a small number of stores free-ride on the brand, or do not cooperate in promoting the produces, those stores can undermine the brand as a whole." (Id. at 19) (emphasis in original).

To counter these incentive problems, franchisors use intrabrand restraints to "align[] the incentives of incentives of individual franchisees with those of the franchisor," such as "explicit contractual arrangements designed to incentivize franchisees to engage in behaviors that benefit the brand." (Id. at 20).

For example, given the critical value of standardization and consistency of quality across stores in the QSR industry, franchisors commonly employ franchise agreements with numerous restraints that explicitly require franchisees to follow the specific, standardized processes that promote the success of the brand. Those restraints predictably involve some potential consequences for non-compliance, in order to incentivize compliance with those restrictions, such as the threat of contract termination or other remedies.

(Id.). These intrabrand restraints could include requirements that franchisees maintain similar layouts, menu items, operation manuals, and promotional materials, among other things. (Id. at 20, 26-29) (describing similar brand standards in the Franchise Agreement).

Again, Dr. McCrary asserts that intrabrand restraints are "procompetitive precisely because, by strengthening the quality of the brand, they strengthen interbrand competition." (Id. at 21) (emphasis altered).

Ultimately, consumers benefit most when multiple competing brands provide high-quality products with strong customer service across many locations at good prices—i.e., strong "interbrand" competition—and increasing consumer welfare is a primary goal of competition policy. Thus, if franchisors cannot align the incentives of individual franchisees to invest in the quality of the brand through the use of intrabrand restraints, consumers will miss out on features they value—like strong customer service, and consistent product quality across every location. Moreover, a strong brand is also critical to recruiting potential new franchisees, and to incentivizing existing franchisees to purchase and open new locations.

(Id. at 22). Dr. McCrary also asserts that intrabrand restraints "ultimately benefit[] workers within any franchise system by increasing demand for the brand's products and expanding the size of the franchise system, which, in turn increases the total labor demand of the franchise system." (Id.) (emphasis in original).

The final economic principle that Dr. McCrary addresses is the difference between interbrand restraints and intrabrand restraints within the franchise context. The plaintiffs suggest that because Jimmy John's "franchisees are independent business owners, the [No-Poach Provision was] a restraint between independent, horizontal competitors." (See id. at 23) (citing Am. Compl. at 13). This characterization, according to Dr. McCrary, ignores the fact that franchisees are "not fully independent: They are tied together by a common interest in the success of the brand, and the franchise agreements reflect the interconnectedness of their economic fates." (Id.).

For example, if a set of horizontal competitors that are not part of the same brand agreed to offer the exact same menu in their stores (rather than compete on product selection), one might be concerned they were coordinating in a way that was anticompetitive. However, the exact same behavior within the context of a QSR franchise system clearly facilitates that system's ability to compete with both other franchise systems and other businesses in the same space. Consequently, a simple and widely observed form of conduct would be misinterpreted as anticompetitive if one were unwilling to acknowledge the economic reality of the franchise model and its implications for franchisees' conduct.

(Id.) (emphasis in original).

Dr. McCrary then "turns to Jimmy John's business model specifically[] and provide[s] a summary of Jimmy John's operations and strategy." (Id. at 11). As alluded, Jimmy John's, like other franchisors, expects its franchisees to uphold certain brand standards "ranging from the general appearance of the restaurant, to the restaurant's physical location, to employee dress and grooming, among others." (Id. at 26). Jimmy John's sends franchise consultants to each Jimmy John's location "to verify that franchisees [are] upholding those Brand Standards." (Id. at 30). " [Redacted] " (Id. at 31).

Franchisees are also required to attend "orientation in Champaign, Illinois," as well as a three-week training program for first-time franchisees. (Id. at 35). "In these trainings, individuals learn . . . operation procedures, Brand Standards, and training standards . . . ." (Id. at 35-36). After completing [Redacted] (Id. at 37). When a franchisee opens a new restaurant, "shift coverage and training requirements change. For example, prior to opening a third Jimmy John's store[,] a franchisee must hire and train an Area Manager." (Id. at 39). To qualify for the Area Manager training program, "a trainee must have completed the Certified Manager training program and a four-week apprenticeship, unless the trainee has at least six-months of in-restaurant training experience . . . ." (Id.). Jimmy John's, however, only "covers the [Redacted] . . . . Current Certified Manager training fees per person are [Redacted] . Including travel and other costs, Certified Manager training is estimated to cost [Redacted] per manager." (Id.).

All that is to say, according to Dr. McCrary, that the No-Poach Provision "encourage[d] franchisees to invest in more training for their employees because it helped ensure that they would be able to recover the costs incurred in training employees. Without [it], there would be an increased risk of losing investments in training to other franchisees." (Id. at 40).

Moreover, the Franchise Agreement gave "each individual franchisee discretion as to when to invoke the No-Poach Provision." (Id. at 44). Dr. McCrary contends that "documents describing [Redacted] requests for release of individual employees" from the No-Poach Provision reveal that the requests "were granted 88 percent of the time." (Id.) (emphasis in original).

It is also notable that the patterns of release approvals differ between managers and other employees. Of the [Redacted] release requests, [Redacted] were for non-managers, and in those cases, over [Redacted] percent were released. Further, if we focus on the releases that were either approved conditional on compensation or refused, such releases were most often associated with managers. Specifically, in the [Redacted] instances where a release was refused for reasons other than poor performance or was approved conditional on compensation, 74 percent ( [Redacted] ) involved managers—and training investments were explicitly mentioned in the negotiations for approximately half of those managers.

(Id. at 48-49) (emphasis in original). "These facts alone," according to Dr. McCrary, "undermine the Plaintiff's claim that the [No-Poach Provision was] used to orchestrate a monopsony." (Id.).

Only a small fraction resulted in an outright refusal to release an employee. Thus, Dr. Singer's mechanism of harm (limited mobility) is not present for a great majority of class members. It is a basic tenet of cartel theory that all or nearly all members of an alleged cartel must enforce the alleged agreement for it to have a marketwide effect. The proposition that an alleged cartel mechanism would have a marketwide effect despite almost always being circumvented strains credulity.

(Id. at 47-48).

Like Dr. Ordover, Dr. McCrary contends that the No-Poach Provision "did not restrict wage competition (or mobility) between Jimmy John's and other brands." (Id. at 46) (emphasis in original). Thus, given that at least "[s]ome franchisees indicate that they compete with all other [QSRs] in the local area hiring minimum wage employees . . . , the only way an individual worker's wage could be suppressed by the No-Poach Provision is if the worker had developed specific skills at Jimmy John's that raised their productivity at Jimmy John's more than at other competing brands. If not, then competition from other brands would push the worker's wage at Jimmy John's up to the competitive level associated with that worker's skills." (Id. at 46-47).

Dr. McCrary also asserts that the No-Poach Provision was not uniformly enforced. For example, "[n]umerous franchisees testified that they did not enforce [it], and others testified that they were not aware of [it]." (See id. at 49-50) (citing five franchisee-declarations). Similarly, he says, several Jimmy John's employees "indicated that the No-Poach Provision [was] not generally enforced." (See id. at 51) (citing three employee-declarations). Rather, "[i]n the limited number of instances where the request to transfer to a new franchisee was denied, it was often for clearly articulated reasons unrelated to the Plaintiff's claims of a conspiracy." (Id. at 51). Indeed, Dr. McCrary states that "39 percent of denied requests" were because of "poor employee performance," which, he argues, "is inconsistent with the Plaintiff's claims that the purpose of the No-Poach Provision was to orchestrate a monopsony." (Id.). He concludes, therefore, that "[i]f the purpose of the No-Poach Provision was to orchestrate a monopsony conspiracy against all employees, [Jimmy John's (the franchisor)] would not have designed [it] in such a flexible way that allowed each franchisee the ability to selectively enforce it." (Id. at 53).

With that in mind, Dr. McCrary contends not only that the No-Poach Provision was selectively enforced, but also that Jimmy John's lacked incentives "to impose a monopsony conspiracy." (Id. at 57).

Plaintiff's claim that the [No-Poach Provision was] used as a tool to orchestrate a conspiracy does not make sense because (1) the monopsony suggested by Dr. Singer and Plaintiff would reduce the revenue at each Jimmy John's location, which would reduce [Jimmy John's (the franchisor)] profits and undermine its goal of growing the brand; and (2) the clauses granted franchisees the discretion to allow mobility between stores, which on its own would contradict the claim of a franchise-wide conspiracy.

(Id. at 23).

For support, Dr. McCrary, like Dr. Ordover, goes back to a classic example of monopsony:

[W]hen a monopsonist is in a competitive market, the price of its products (in this case, sandwiches) is unchanged by a monopsonistic reduction in labor demand because there is still product market competition. However, the amount of the monopsonist's output (i.e., the number of sandwiches made) falls because it has less labor. Thus, the revenue of the monopsonist (output multiplied by price) falls. Thus, in this textbook case, revenue falls by virtue of monopsonistic behavior, while profits rise.

(Id. at 55).

Dr. McCrary contends that monopsonistic behavior might benefit the franchisees but not Jimmy John's. This is because Jimmy John's earns royalties based "on the revenue—not the profit—of each franchisee." (Id. at 55) (emphasis in original). In other words, although monopsony could help franchisees earn more profits because they will hire fewer workers, their revenue would decrease because they are making fewer sandwiches. (Id.). As a result, the franchisees would pay fewer royalties to Jimmy John's, the franchisor. (Id.). Put differently, Jimmy John's would prefer "franchisees to maximize sales instead of profits." (Id. at 56) (citing Roger D. Blair & Francine Lafontaine, Understanding the Economics of Franchising & the Laws That Regulate It, 26 Franchise L.J. 55, 58 (2006)). Yet "[u]nder the Plaintiff's theory, [Jimmy John's] orchestrated a conspiracy to suppress labor opportunities within its own franchise system—when in fact [its] clear financial incentive is to expand labor opportunities within the brand in order to expand its revenue, size, and, ultimately, the value of the brand." (Id.) (emphasis added).

At any rate, Dr. McCrary concludes that the No-Poach Provision "can be explained as [a] procompetitive restraint[] that help[ed] align incentives and ensure quality and consistency of Jimmy John's brand across all stores." (Id. at 12). For one, "consistent with the labor economics literature on human capital investment and internal labor markets, the No-Poach Provision can help to encourage each individual franchisee to make investments in training by protecting those investments from poaching." (Id. at 57). "Broadly, the literature explains that training increases a worker's productivity and that firms will choose to invest in training when they expect to recoup the costs of their training investment through the worker's increased productivity." (Id. at 58) (citing Gary S. Becker, Human Capital: A Theoretical & Empirical Analysis, with Special Reference to Education 13-44 (2d ed. 1975)).

Whereas Dr. Singer says that this supposed justification does not make up for the "anticompetitive effects for workers," Dr. McCrary asserts that, "by protecting training investments, the No-Poach Provision can also benefit workers because [Jimmy John's has] an incentive to share the returns of training investment with workers to strengthen their internal labor market." (Id. at 57) (emphasis added).

With the No-Poach Provision in place, there is more incentive to provide training that increases the value of the franchisee-employee relationship. However, if the worker leaves, the franchisee loses the value of the training, so the franchisee needs to set wages to trained workers to reduce the likelihood the workers will leave for another job.

(Id. at 93). Thus, according to Dr. McCrary, not only did the No-Poach Provision benefit franchisees by protecting their investments, but it also benefited workers by providing training that increased their skills and their advancement opportunities within the "internal labor market." (See id. at 62-67). In other words, "as the Jimmy John's brand grows, labor opportunities within the franchise system expands—the opposite of what one expects in a monopsony." (Id. at 67). "In particular, encouraging franchisees to invest in training and allowing them to protect their training investments through the No-Poach Provision can lead to improvements in Jimmy John's store quality and consistency, which increases competition between QSR brands and can expand the labor demand for Jimmy John's." (Id. at 76).

To show this, Dr. McCrary also used regression analysis to "quantify the effect of investments in Certified Managers on store performance using available metrics of store performance . . . ." (Id. at 67). These metrics "include several key measure of performance for each store: Peak Scores, average out-the-door-times per delivery order, and the number of complaints, in addition to financial metrics, such as annual profits and annual cost of goods sold." (Id. at 67-68). A store's Peak Score is "based on a one-hundred point scale that describes the extent to which a given Jimmy John's store is meeting Brand Standards . . . ." (Id. at 68). It is assessed periodically by "Jimmy John's Business Coaches," who also "suggest room for improvement." (Id.) The other metrics are largely self-explanatory and collectively capture how Jimmy John's measures store performance. In fact, Jimmy John's includes this data in its franchise newsletter, Jimm-enews. (Id. at 69). Dr. McCrary suggests that these metrics confirm a correlation between store performance and the number of certified managers on staff:

I start my analysis by summarizing the average Peak Scores across stores with a Certified Manager count of zero, one, two, three, or four or more. Stores with a Certified Manager count of zero score around [Redacted] , while those with a count of one score [Redacted] . . . . The overall trend is that more extensive use of Certified Managers is associated with higher Peak Score increased.

(Id. at 69). He then "control[s] for a variety of variables, including" store type, store location, franchisee experience, store age, and state and year fixed effects to establish that "more extensive use of Certified Managers in a store" correlates with order being processed faster, fewer complaints, and reduced costs (possibly because "part of the Certified Manager training includes teaching trainees how to minimize costs"). (Id. at 71-76). These results, he asserts, reveal the strong incentives that Jimmy John's franchisees have for investing in certified managers. (See id. at 76). The No-Poach Provision helped protect that investment, encourage more training, and ultimately improve "store quality and consistency, which increases competition between QSR brands and can expand the labor demand for Jimmy John's." (Id. at 76).

Similarly, "[t]o assess whether training investments in Certified Managers changed before and after the removal of the No-Poach Provision in 2018," Dr. McCrary compared their use from 2014 to 2018. (Id. at 86). The results: [Redacted] (Id.)

[Redacted] (Id. at 87). "This decline coincides broadly with three significant changes to the No-Poach Provision": (1) in 2016, the No-Poach Provision was modified so to "allow[] workers to move to other stores so long as the other store did not initiate the transfer with a solicitation; (2) in 2018, Jimmy John's " [Redacted] "; and (3) again in 2018, because of an investigation by the Attorney General of the State of Washington threatening legal action, Jimmy John's stopped enforcing the No-Poach Provision. (Id. at 87-88). Dr. McCrary therefore asserts that the erosion of investment protections once offered by the No-Poach Provision made franchisees wary to train new certified managers, as reflected by data showing a reduction in the number of scheduled shifts. (Id. at 87).

Finally, Dr. McCrary explains that the No-Poach Provision "helped minimize infighting and encourage cooperation among franchisees," negating Dr. Singer's "unsubstantiated claims that training and brand quality were not impacted following the removal of" the No-Poach Provision. (Id. at 57).

[T]here are a variety of cooperative behaviors between local franchisees that help strengthen the brand and that Jimmy John's wants to incentivize (e.g., sharing supplies, referring catering business, cooperative advertising, hosting training sessions for multiple stores at once). Such behaviors can be threatened by infighting. Thus, to the extent Jimmy John's can find mechanisms to avoid intrabrand infighting, and encourage cooperation, it can strengthen the brand overall. The [No-Poach Provision is] one such mechanism.

(Id. at 76-77). By maintaining cooperation between franchisees, Jimmy John's can "ensure quality of service" and thus "strengthen the brand." (Id. at 79). For example, "[t]o the extent that some franchisees might be better at recruiting Certified Managers in the absence of the No-Poach Provision, those franchisees might be more aggressive in poaching a larger share of the best managers in a given area, leaving other stores with weak performers. The No-Poach Provision can help manage this form of intrabrand conflict directly." (Id. at 82). Again, Dr. McCrary contends that reducing infighting—and thus strengthening the brand—benefits workers by "expand[ing] labor opportunities at Jimmy John's[,] which in turn would provide more opportunities for existing employees." (Id. at 83).

LEGAL STANDARD

"Antitrust cases often involve the collection, assimilation, and evaluation of vast amounts of evidence regarding numerous transactions and other economic data." Manual for Complex Litigation § 30.2 (4th ed. 2004). Litigants often "retain economists to study such topics as relevant markets, the concentration of economic power, pricing structures, elasticity of demand, barriers to entry, marginal costs, and the effect of the challenged practices on competition and the claimants." Id.

"Expert witnesses play a limited role in class certification hearings; though some courts admit testimony on whether Rule 23 standards—such as predominance—have been met. . . . Courts have applied a high threshold for assessing the need for expert testimony at the certification stage. A judge should not be drawn prematurely into a battle of competing experts." Id. § 21.21. Even so, "[w]hen an expert's report or testimony is 'critical' to class certification, . . . a district court must make a conclusive ruling on any challenge to that expert's qualifications or submissions before it may rule on a motion for class certification." Messner v. Northshore Univ. HealthSystem, 669 F.3d 802, 812 (7th Cir. 2012). "That is, the district court must perform a full Daubert analysis before certifying the class if the situation warrants," including when there is a "challenge to the reliability of information provided by an expert if that information is relevant to establishing any of the Rule 23 requirements for class certification." Am. Honda Motor Co., Inc. v. Allen, 600 F.3d 813, 816 (7th Cir. 2010).

Here, the disputed expert testimony involves Rule 23(b)'s predominance requirement. Analyzing predominance begins with the elements of the cause of action. See Erica P. John Fund, Inc. v. Halliburton Co., 563 U.S. 804, 809 (2011). So the Court must ultimately examine whether the plaintiffs can "establish each of the required elements of an antitrust claim—(1) a violation of antitrust law; (2) injury and causation; and (3) damages—using common evidence." Messner, 669 F.3d at 815. "Importantly, individual injury" (known as antitrust impact) "is an element of the cause of action; to prevail on the merits, every class member must prove at least some antitrust impact resulting from the alleged violation." In re Hydrogen Peroxide Antitrust Litig., 552 F.3d 305, 311 (3d Cir. 2008). "[I]mpact often is critically important for the purpose of evaluating Rule 23(b)(3)'s predominance requirement because it is an element of the claim that may call for individual, as opposed to common, proof." Id.

The task for plaintiffs at class certification is to demonstrate that the element of antitrust impact is capable of proof at trial through evidence that is common to the class rather than individual to its members. Deciding this issue calls for the district court's rigorous assessment of the available evidence and the method or methods by which the plaintiffs propose to use the evidence to prove impact at trial.

Id.

When confronted with expert testimony, "the trial judge must determine at the outset . . . whether the reasoning or methodology underlying the [expert] testimony is . . . valid and . . . whether that reasoning or methodology can be applied to the facts in issue." Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 592-93 (1993). "Proposed testimony must be supported by appropriate validation," id. at 590, and not merely "connected to existing data only by the ipse dixit of the expert," Gen. Elec. Co. v. Joiner, 522 U.S. 136, 146 (1997). "The focus, of course, should be on principles and methodology, not on the conclusions that they generate." Daubert, 509 U.S. at 595.

The Court's gatekeeping duty "often must be exercised with special care." Joiner, 522 U.S. at 148 (Breyer, J., concurring). Expert testimony is admissible when:

(a) the expert's scientific, technical, or other specialized knowledge will help the trier of fact to understand the evidence or to determine a fact in issue;

(b) the testimony is based on sufficient facts or data;

(c) the testimony is the product of reliable principles and methods; and

(d) the expert has reliably applied the principles and methods to the facts of the case.

Fed. R. Evid. 702. The Daubert Court listed several considerations that may bear on the inquiry:

(1) whether the expert's technique or theory can be or has been tested—that is, whether the expert's theory can be challenged in some objective sense, or whether it is instead simply a subjective, conclusory approach that cannot reasonably be assessed for reliability;

(2) whether the technique or theory has been subject to peer review and publication;

(3) the known or potential rate of error of the technique or theory when applied;

(4) the existence and maintenance of standards and controls; and

(5) whether the technique or theory has been generally accepted in the scientific community.

Id. at advisory committee's note to 2000 amendment; Daubert, 509 U.S. at 593; see also Kumho Tire Co. v. Carmichael, 526 U.S. 137, 155 (1999) ("Daubert's general principles apply to the expert matters described in Rule 702.").

The Seventh Circuit gave further guidance in Manpower, Inc. v. Insurance Co. of Pennsylvania, 732 F.3d 796 (7th Cir. 2013). There, the appellate court vacated the exclusion of expert testimony because the trial judge erroneously "comment[ed] on the soundness of the factual underpinnings of [the expert's] calculation," id. at 810, rather than testing "the reliability of the methodology the expert employed," id. at 806. More specifically, the Manpower court objected to the district court's criticism of the expert's data selection:

Whether [the expert] selected the best data set to use . . . is a question for the jury, not the judge. Assuming a rational connection between the data and the opinion—as there was here—an expert's reliance on faulty information is a matter to be explored on cross-examination; it does not go to admissibility.

Id. at 809. The court also noted, however, "That is not to say that an expert may rely on data that has no quantitative or qualitative connected to the methodology employed": Federal Rule of Evidence 702's "requirement that expert opinions be supported by 'sufficient facts or data' means 'that the expert considered sufficient data to employ the methodology' . . . ." Id. at 808. There too, the focus must remain on "the validity of the methodology employed by an expert, not on the quality of the data used in applying the methodology or the conclusions produced. . . . Admittedly, this is not always an easy line to draw." Id. at 806.

For all that, there is no "definitive checklist or test" for evaluating expert testimony: The inquiry must be "a flexible one." Daubert, 509 U.S. at 593-94. But the Court "may not duck hard questions by observing that each side had some support. . . . Tough questions must be faced and squarely decided, if necessary by holding evidentiary hearings and choosing between competing perspectives." West v. Prudential Sec., Inc., 282 F.3d 935, 938 (7th Cir. 2002). The burden rests with the proponent of the expert testimony to prove its admissibility (for class certification) by a preponderance of the evidence. See Fed. R. Evid. 702 at advisory committee's note to 2000 amendment; Manual for Complex Litigation (4th ed. 2004) ("The judge need not decide at the certification stage whether such testimony satisfied standards for admissibility at trial.").

ANALYSIS

From the onset, the Court notes that "[i]f an expert witness uses regression analysis in estimating antitrust damages, one can hardly claim that the methodology is unreliable and therefore the expert's testimony should not be excluded on Daubert grounds." Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law: An Analysis of Antitrust Principles and Their Application ¶ 399 (5th ed. 2020). Still, the expert "may employ this reliable methodology in an unreliable way." Id. "As a practical matter, it makes little difference whether unreliable testimony is excluded on Daubert grounds or accorded no weight at the [class certification] stage. In either event, the unreliable testimony will not influence the outcome of the dispute, which is as it should be." Id.

With that in mind, the Court will not place undue emphasis on the accuracy of the WSR data. Instead, Dr. Singer's models are unreliable because of a methodological flaw— his "random measurement error" leads to inflated results. On the other hand, both Dr. Ordover's and Dr. McCrary's testimony withstand scrutiny—the remaining questions go to their weight, not admissibility.

A. The Experts

None of the experts' credentials is in serious doubt. Both Dr. Singer and Dr. McCrary are accomplished economics professors whose testimony is often admitted by courts. And Dr. Ordover is an internationally recognized antitrust expert and co-author of the U.S. Horizontal Merger Guidelines. To the extent that the litigants challenge the experts' qualifications, the Court is unpersuaded.

B. Dr. Singer

Although Jimmy John's raises several arguments about why the Court should exclude Dr. Singer's testimony, only two require discussion here. As discussed below, concerns about the accuracy of the WSRs go to their weight, not admissibility. In any event, Dr. Singer's models fail to adjust for those two percent of WSRs that do not consistently record employee wages as per-shift or per-hour, leading to inflated results. Because of this methodological flaw, the Court must exclude his testimony.

(i) The Data

First, Jimmy John's argues that the Court must exclude Dr. Singer's testimony because the results of his economic models derive from flawed data. "[The] WSR data is unverified and demonstrably unreliable, and Dr. Singer's analyses relying on it should therefore be excluded." (Jimmy John's Mot. to Exclude at 5). It relies on franchisee testimony to show that the wages reflected in the WSR data are seldom accurate (id. at 6), and are thus unreliable "indicator[s] of actual wages" (id. at 7) (emphasis in original). The Court disagrees.

Jimmy John's qualms about the WSR data were addressed by the Seventh Circuit in Manpower: "[A]n expert's reliance on faulty information is a matter to be explored on cross-examination; it does not go to admissibility." 732 F.3d 796 (7th Cir. 2013). Whether there is a better source for the wage data is beside the point—challenges to the factual underpinnings go to the weight of Dr. Singer's testimony.

Moreover, the nonbinding cases Jimmy John's relies on for support are inapt. True, Judge Lee of the Northern District of Indiana partly excluded an expert's opinion because it was "based on unverified data" in State Farm Fire & Casualty Co. v. Electrolux Home Products, Inc., 980 F. Supp. 2d 1031, 1040 (N.D. Ind. 2013). (Jimmy John's Mot. to Exclude at 7). There, the plaintiff-insurer sued the defendant-manufacturer after an insured's dryer caught fire and caused damage. 980 F. Supp. 2d at 1035. The defendant sought to introduce expert testimony at trial purporting to show that its dryer was not unreasonably dangerous. Id. at 1037. Unlike here, the court in State Farm did not face an "assert[ion] that the data was skewed or improperly provided . . . ." Id. at 1039. Instead, the court objected to how the plaintiff's expert combined "data from two entirely different sources" yet never questioned whether those sources were similarly reliable. Id. at 1039-40. At any rate, the expert's testimony apparently suffered from several other flaws, including the fact that the expert's analysis was too generalized and "not specific to the issue in [the] case." Id. at 1040. And insofar as the State Farm court excluded the expert's testimony because it rested on faulty information, that decision no longer follows precedent. Manpower, 732 F.3d at 809 ("[A]n expert's reliance on faulty information is a matter to be explored on cross-examination; it does not go to admissibility.").

Take another case Jimmy John's cites, Fail-Safe, L.L.C. v. A.O. Smith Corp., 744 F. Supp. 2d 870, 888 (E.D. Wis. 2010). (Jimmy John's Mot. to Exclude at 7). There, Judge Stadtmueller of the Eastern District of Wisconsin cited a Fifth Circuit case for the proposition that the "[u]se of outdated or suspect data at the base of an expert's testimony are proper grounds to exclude that testimony." 744 F. Supp. 2d at 888 (citing Marcel v. Placid Oil Co., 11 F.3d 563, 567-68 (5th Cir. 1994)). Like State Farm, Fail-Safe also may be outdated given the Seventh Circuit's directive three years later to focus on "the reliability of the methodology the expert employed." Manpower, 732 F.3d at 806. To be sure, the Manpower court emphasized that "[t]he district court usurps the role of the jury, and therefore abuses its discretion, if it unduly scrutinizes the quality of the expert's data and conclusions rather than the reliability of the methodology the expert employed." Id. At any rate, Fail-Safe can still be differentiated as a challenge to the sufficiency of the data under Rule 702. After all, the issue with the expert's data in Fail-Safe was that it was "adopt[ed] wholesale from a single, undated . . . PowerPoint Slide, which stated that the company hoped to have '350k . . . Target Units.'" 744 F. Supp. 2d at 887-88 (emphasis in original). Jimmy John's pokes holes in the reliability of the WSRs through franchisee testimony about their inaccuracy, but the data is not so "incredibly shallow" as to warrant exclusion. Id. at 888. In sum, questions about the accuracy of the WSRs go to their weight, not admissibility.

(ii) The Methodology

Jimmy John's argues that Dr. Singer "grossly misinterpreted [the WSR] data by implausibly assuming all wages in the WSR data were hourly." (Jimmy John's Mot. to Exclude at 13). More specifically, Jimmy John's contends that, for some employees, the WSR data reflects per-hour wages, while for other employees, it reflects per-shift wages. (Id. at 8). According to Jimmy John's, Dr. Singer never accounted for that difference. Rather, he lumped them all together. For example, a manager making [Redacted] per-shift was recorded as making [Redacted] per-hour—or nearly [Redacted] -per-year. (Id.). Ultimately, Jimmy John's asserts that this supposed error "substantially inflates the results of Dr. Singer's models" and renders them unreliable. (Id.) (citing Ordover Report at 23 ("Correcting for [Dr. Singer's] data error reduced [the] estimate of aggregate damages [from [Redacted] ] to [Redacted] , a reduction of roughly [Redacted] .")).

For support, Jimmy John's relies on the Supreme Court case Tyson Foods, Inc. v. Bouaphakeo, 136 S. Ct. 1036, 1048-49 (2016), for the proposition that "[r]epresentative evidence that is statistically inadequate or based on implausible assumptions could not lead to a fair or accurate estimate of [damages]." (Jimmy John's Mot. to Exclude at 7-8). In Tyson, the Court held that a representative sample is sometimes appropriate to establish class-wide liability when "each class member could have relied on that sample . . . if he or she had brought an individual action." Id. at 1067. But the Court never passed on the reliability of the statistical evidence at issue—the defendant-petitioner did not raise a Daubert challenge, so the Tyson Court had "no basis in the record to conclude it was legal error to admit that evidence." Id. Still, the Court gave helpful guidance that "[t]he fairness and utility of statistical methods in contexts other than those presented here will depend on facts and circumstances particular to those cases." Id. at 1049.

The plaintiffs here, on the other hand, argue that Dr. Singer's control variables—the fixed effects—account for the wage discrepancy just described: The "Worker Fixed Effects control for all worker-specific characteristics . . . so long as the recording method is fixed over the course of the study." (Singer Rebuttal at 23) (emphasis in original). Because 98 percent of the WSRs consistently classify employee compensation as hourly or salaried, "any measurement is mathematically irrelevant because it is absorbed by the Worker Fixed Effects in [the] regressions." (Id. at 2). The Court disagrees.

The wage discrepancy in Dr. Singer's regressions makes them unreliable. True enough, even Dr. Ordover admits that "Dr. Singer . . . include[s] employee fixed effects in some of his regressions . . . ." (Ordover Report at E-7). Even so, the fixed effects "cannot control for this data error . . . because some managers can be . . . paid on both a per-shift and a salaried basis." (Id.) (emphasis added). Put differently, WSRs that do not consistently document wages as per-hour or per-shift are uncontrolled because they are not fixed effects.⁶ According to Dr. Singer, that only amounts to two percent of employees—a "random measurement error" that "is mathematically irrelevant . . . ." (Singer Rebuttal at 24, 28). He then provides an illustration purporting to show that the error "is effectively washed away (in a statistical sense) when the regression has a large number of data points . . . ." (Id. at 29). But that hardly explains his finding that, even with the fixed effects, managers earned on average [Redacted] (Singer Report at 34), "which would translate to almost [Redacted] assuming a standard 2,000 hour work year" (Ordover Report at E-3). Indeed, the highest wage observed by Dr. Singer was [Redacted] or nearly [Redacted] . (Singer Report at 34). When actual salaries are between [Redacted] and [Redacted] yet the regression yields an average of [Redacted] , the [Redacted] random measurement error may not, in fact, be mathematically irrelevant. While it generally may be true that outliers in a regression are absorbed as the number of inputs increase, that is not reflected by Dr. Singer's results. And given that 25 percent of all managers compose that two-percent error (Ordover Report at E-3), it makes sense that their average wage would be inflated. In his defense, Dr. Singer suggests that "[f]or Dr. Ordover's critique to have any merit, he would, at a minimum, need to establish that any random measurement errors that do crop up in the WSR data (and are not absorbed by Worker Fixed Effects) are somehow correlated with the No-Poach Provision," which he "makes no attempt to establish." (Singer Rebuttal at 34). But aside from that conclusory assertion, Dr. Singer says little else; and the plaintiffs' briefs do not either. Similarly, neither Dr. Singer nor the plaintiffs themselves try to explain how this "statistical noise" squares away the clearly inflated average wages. Worse, the plaintiffs bear the burden of establishing the reliability of their expert's testimony. This error is material. Comparing inflated estimates of average wages leads to inflated estimates of impact. Because Dr. Singer's models suffer from this methodological flaw, the Court must exclude his testimony as unreliable.

6 Dr. Ordover raises a similar argument about Dr. Singer's fixed effect for "the local area minimum wage": "[W]hile Dr. Singer uses a variety of fixed effects in different regression specifications, the fact that the relationship between the wage rates and the minimum wage varies in his job category-specific regressions indicates that use of such fixed effects does not solve this flaw." (Ordover Report at 43-44 n.84).

For these reasons, the Court GRANTS Jimmy John's Motion to Exclude.

C. Dr. Ordover

The plaintiffs also raise several objections to Dr. Ordover's testimony, but none justify its exclusion. The Court will address each in turn.

(i) Endogeneity Bias

First, the plaintiffs contend that Dr. Ordover's testimony must be excluded because his "approach . . . commits a well-known econometric error: endogeneity bias." (Pls.' Mot. to Exclude at 5-6) (emphasis added). They assert that "Dr. Ordover creates a purported 'independent' control variable that is not independent at all." (Id.). In other words, "[t]he data contains no formulaic relationship, but Dr. Ordover invests and imposes one, creating an additional manufactured 'independent' variable that is an explicit mathematical function of the dependent variable." (Pls.' Reply in Support of Mot. to Exclude 2, Doc. 206).

"Endogeneity bias is a more subtle flaw that could undermine the reliability of statistical evidence." Jennifer Gerarda Brown, Sweeping Reform from Small Rules? Anti-Bias Canons as a Substitute for Heightened Scrutiny, 85 Minn. L. Rev. 363, 408 (2000).

In any regression analysis, a researcher generates a series of variables or determinants (referred to as "independent" or "right-hand side" variables) and tries to determine which (if any) of them causes a particular phenomenon (the dependent or "left-hand side" variable). For example, when estimating whether hiring more police officers deters crime, a researcher might run a regression in which the left-hand side variable would be crime level and one of the right-hand side variables would be the number of police officers hired. But right-hand side variables should be exogenous: they should cause—not be caused by—the left-hand side variable. The trouble is that

with cops and crime, causation can go both ways. As the numbers of police increase, crime may decrease, but if crime levels rise, local governments could respond by hiring more police. Thus the number of police proves to be endogenous--it is not independent of the left-hand side. To remedy this endogeneity problem, researchers need a proxy (or "instrument") for the number of police—something truly exogenous—to identify the exogenous effect of the number of police on the crime rate.

Id.

The Court is not persuaded that Dr. Ordover's models suffer from endogeneity bias. The plaintiffs' argument on this point is conclusory. They provide scant proof that the variable Dr. Ordover created by unpooling the wage data is merely a function of the dependent variable. Instead, they assume that Dr. Ordover's models are "biased, inconsistent, and unreliable" without pointing to any demonstrable evidence that his results were skewed by any supposed endogeneity bias. (Singer Rebuttal at 32). Rather, Dr. Ordover's unpooling is conceptually like when Dr. Singer separated employees by job category. Without more, quoting economics textbooks cannot show that there is a mechanical relationship between Dr. Ordover's variables that biases his results.

(ii) State Regressions & Endogenous Sample Selection

Next, the plaintiffs argue that Dr. Ordover improperly conducted "separate regressions by state and by year, each time discarding from two-thirds to 90% of the relevant data." (Pls.' Mot. to Exclude at 8) (emphasis omitted). They claim that "[h]e does so without any underlying economic reason . . . ." (Id. at 2). Naturally, Dr. Ordover says that "[t]his is nonsense: I do not 'discard' any observations." (Ordover Rebuttal at 16). The truth, it seems, lands somewhere in between.

Although Dr. Ordover discarded some data points, that does not make his methodology unreliable. He admits that when it came down to employees whose wages were listed between $30 and $75 in the WSRs, "these workers cannot be reliably classified as being paid hourly or per-shift," so he excludes them. (Ordover Rebuttal at 12; Ordover Report at E-9). Still, that only represents "roughly 34,000 observations out of over 14.9 million . . . ." (Ordover Rebuttal at 12). Dr. Singer merely points to Dr. Ordover's Appendix F-4, but that does not support the proposition that Dr. Ordover discarded "millions of data points . . . ." (Singer Rebuttal at 34). Rather, Appendix F-4 reflects that Dr. Ordover uses the same 14 million observations as Dr. Singer and separates them according to state. (Ordover Report at F-5-F-7). Their sum confirms that Dr. Ordover only discarded about 34,000 observations—not millions. In brief, Dr. Ordover's testimony rests on sufficient data.

Similarly, the plaintiffs' argument that Dr. Ordover's models suffer from endogenous sample selection also fails. In a mere sentence, the plaintiffs contend that Dr. Ordover's models "guarantee biased and unreliable results" because he "selects the data for each regression based upon a mathematical function of the dependent variable." (Pls.' Mot. to Exclude at 7). This claim too is unsupported, and Dr. Singer's elaboration that Dr. Ordover discarded "71 percent of all manager data" is likewise unpersuasive. As before, Dr. Ordover's results do not reflect that he "limit[s] the data to salaried managers," or 690,302 out of 2,360,933 observations. (Singer Rebuttal at 38 n.125). Rather, the figure referenced by Dr. Singer proves that Dr. Ordover accounted for 690,302 salaried managers and 1,636,275 hourly managers. (Ordover Report at F-2). Moreover, Dr. Singer's own regressions discard "2.5% of all observations" that he "consider[s] outliers" (Singer Report at 33), which is "10 times the number [Dr. Ordover] exclude[s] as outliers due to ambiguity in how managers' pay was recorded" (Ordover Rebuttal at 12) (emphasis omitted). For all that, neither expert considered insufficient data.

Finally, the plaintiffs' argument that Dr. Ordover erred by conducting separate regressions for each state and year is unavailing. The plaintiffs contend that Dr. Ordover had no reason to conduct these separate regressions given that Dr. Singer's fixed effects "control[] for . . . store-specific factors that are fixed over time[] and factors that are specific to each individual Class Member that are fixed over time." (Pls.' Mot. to Exclude at 8) (emphasis omitted). But again, Dr. Ordover's regressions are conceptually similar to the ones Dr. Singer ran—Dr. Singer unpooled wages according to job category, and Dr. Ordover took that same approach one step further by running additional regressions based on state and year. (See Ordover Report at 34 n.67). While the plaintiffs suggest that "[t]here is no reason to run separate regressions by state or by year, other than to engage in data mining without merit" (Pls.' Mot. to Exclude at 9), Dr. Ordover defends this methodology by noting that "the effect of the No-Poach Provision might vary across states due to variation in state-level economic and legal factors, including differences in labor regulations and employee protections, unionization rates, and the competitive significance of the quick service industry" (Ordover Rebuttal at 18). In other words, while Dr. Singer assumes that his fixed effects control for "differences in . . . local labor market conditions across time" (Singer Report at 31), Dr. Ordover puts that to the test and concludes that "for nine states . . . wages were higher in the putative class period than in the benchmark period, and for another four states . . . wages were not statistically significantly lower in the putative class period than they were in the benchmark period" (Ordover Report at 40-41). As discussed above, Dr. Singer's testimony is unreliable because of the wage discrepancy, so the Court need not decide whether "[t]he fact that [his] model predicts no wage 'suppression' in many states" is a separate ground for exclusion. (Ordover Report at 36). At any rate, Dr. Ordover's separate regressions are not methodologically flawed.⁷ If the plaintiffs contend that Dr. Singer's model "still finds that approximately 75 percent of Class Members sustained antitrust injury" (Singer Rebuttal at 36), then they are free to rely on it.

7 To the plaintiffs' argument that Dr. Ordover made "another basic mistake" by "not control[ing] for county-level economic conditions" (Pls.' Reply in Support of Mot. to Exclude at 4): "[T]he exclusion of major variables or in the inclusion of improper variables may diminish the probative value of a regression model. But such defects do not generally preclude admissibility, and courts allow use of a regression model as long as it includes the variables accounting for the major factors." In re Urethane Antitrust Litig., 768 F.3d 1245, 1260-61 (10th Cir. 2014) (citing Bazemore v. Friday, 478 U.S. 385, 400 (1986)).

(iii) False Positives

Along those lines, the plaintiffs also contend that Dr. Ordover's testimony must be excluded because his false-positives analysis was faulty. (Pls.' Mot. to Exclude at 9-10). Recall that Dr. Ordover asserts that Dr. Singer ignored several "sound economic reasons why putative class members could not have been injured from the No-Poach Agreement." (Ordover Report at 92) (emphasis in original). For one, Dr. Ordover contends that for the nearly 37 percent of Jimmy John's locations that do not have a competing franchise within 10 miles, employees at those locations "are unlikely to have been impacted by the No-Poach Provision." (Id. at 27-28). He also maintains that the lack of a rival franchise within a reasonable distance means there is no one to "bid up [employee] wages." (Id. at 25). The plaintiffs, on the other hand, say that "Dr. Ordover provides no support for this belief given the fact that "20% of his sample already worked at a Jimmy John's that is located more than 10 miles away from their home." (Pls.' Mot. to Exclude at 9).

Similarly, Dr. Ordover argues that the No-Poach Provision could not have impacted the "roughly 36 percent of employees [that] leave within four weeks . . . and [the] 56 percent [that] leave in twelve weeks or less" because "they would have been unlikely to receive any increases in compensation in the short time of their employment. (Id. at 30-32).⁸ Rather, "[a]s a matter of basic economics,. . . a smaller pool of potential hires would tend to decrease the supply of potential employees for any particular Jimmy John's branded store, which would tend to increase their wages." (Id. at 29) (emphasis added). For prospective employees who have never worked for a Jimmy John's restaurant, two franchisees would have to compete for that prospective employee by offering higher wages. (Id.). The plaintiffs say that this approach "reflect[s] long-disapproved notions of perfect competition that not even Dr. Ordover believes apply to labor markets." (Pls.' Mot. to Exclude at 13).

8 Dr. McCrary agrees, also finding that 60 percent of employees leave within 90 days. (McCrary Report at 52) ("Jimmy John's labor force has a very high turnover rate. High turnover means that even if . . . the [No-Poach Provision was] non-discretionary, [it was] unlikely to have had any effect on the many proposed class members who did not stay at Jimmy John's long enough to seek to move between stores.").

At the heart of this dispute is whether the plaintiffs must define a relevant market to assess monopsony power. For Dr. Singer, economists need not "define a relevant market to study anticompetitive effects of a horizontal wage-suppression scheme." (Singer Rebuttal at 11). Instead, he asserts that "the effect of no-poach clauses on wages is 'an empirical issue,' and should therefore be addressed through 'direct measurement of the effect of the no-poaching clause on wages' . . . . " (Id. at 11-12) (quoting Daniel Levy & Timothy Tardif, Measurement of Market Concentration Faced by Labor Pools: Theory and Evidence from Fast Food Chains in Rhode Island with No-Poaching Clauses, CPI Antitrust Chronicle 11 (May 2020)). Further, Dr. Singer contends that "[a] naïve attempt at labor market definition might erroneously conclude that an individual fast-food brand has no monopsony power, due to competition from other brands" even though "[d]ecades of economic scholarship has shown this to be false." (Id. at 15). At any rate, he suggests record evidence reflects that the No-Poach Provision was enforced: Proof of Jimmy John's monopsony power. (See Singer Report at 7-27). The Court disagrees.

Dr. Ordover did not err by conducting a preliminary analysis of the relevant product market. True, not every antitrust case requires market definition. Sometimes "antitrust law may condemn some conduct, such as naked price-fixing or market-sharing agreements among competitors, with little or no inquiry into market power of the participant," and other times "market structure and market power are . . . crucial in antitrust analysis." Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law: An Analysis of Antitrust Principles and Their Application ¶ 500 (5th ed. 2020). But in July 2018, the Court punted when asked whether the "per-se rule," the "rule of reason," or "quick-look" approach applies in this case. (Mem. Op. & Order, Doc. 46, at 19). The Court also stated, however, that "if the quick-look approach applies," then the plaintiffs "would not be required to go through the industry and market power analysis . . . ." (Id. at 18). See generally Ohio v. Am. Ex. Co., 138 S. Ct. 2274, 2285 n.7 (2018) ("Given that horizontal restraints involve agreements between competitors not to compete in some way, this Court concluded that it did not need to precisely define the relevant market to conclude that these agreements are competitive. . . . But vertical restraints are different. Vertical restraints often pose no risk to competition unless the entity imposing them has market power, which cannot be evaluated unless the Court first defines the relevant market."). Yet "at the class-certification stage (as at trial), any model . . . must be consistent with its liability case, particularly with respect to the alleged anticompetitive effect of the violation. And for the purposes of Rule 23, courts must conduct a rigorous analysis to determine whether that is so." Comcast Corp. v. Behrend, 569 U.S. 27, 35 (2013) (internal citations and quotation marks omitted). Indeed, the Court will ultimately examine whether the plaintiffs can "establish . . . a violation of antitrust law . . . using common evidence." See Messner v. Northshore Univ. HealthSystem, 669 F.3d 802, 812 (7th Cir. 2012). Until the Court decides the applicable legal standard, Dr. Ordover's approach to endeavor into the relevant market is the prudent one.

With that in mind, Dr. Ordover's false-positive analysis is reliable. To the plaintiffs' contention that Dr. Ordover erroneously selected the ten-mile radius, that goes to the weight of his testimony, not its admissibility. Whether Dr. Ordover chose the correct input amounts to asking whether he selected the best data set to use. See Manpower, 732 F.3d at 809. Still, the Court is persuaded with Dr. Ordover's critique that "Dr. Singer did not even attempt to test whether [employee] wages formed the 'starting rung' of some abstract compensation structure; he merely asserts that it is so." (Ordover Rebuttal at 21).⁹ On the other hand, Dr. Ordover considered the relevant market and built a logical bridge to his conclusion that new employees were not impacted by the No-Poach Provision.

9 Although the question of Dr. Singer's wage-structure model is moot, the Court is wary that, without defining a relevant market, the model's baseline premise hinges on "the ipse dixit of the expert." See Gen. Elec. Co. v. Joiner, 522 U.S. 136, 146 (1997).

The Court will therefore consider Dr. Ordover's testimony reliable.

D. Dr. McCrary

Dr. McCrary's testimony is similarly reliable.

First, the plaintiffs argue that Dr. McCrary's testimony must be excluded because he failed to show that the "'potential' procompetitive benefits for Jimmy John's" are "what prompted the No-Poach Agreement." (Pls.' Mot. to Exclude at 12) (emphasis added). But "it is the effect or consequences which controls, not intent or motive." Wilk v. Am. Med. Ass'n, 719 F.2d 207, 225 (7th Cir. 1983) (citing Chi Bd. of Trade v. United States, 246 U.S. 231, 238 (1918)). In other words, while "it is useful to determine the setting in which the restraint was adopted," the "critical and sole factor" is the "true effect." Id. (emphasis added). To that end, Dr. McCrary did not have to show that the "procompetitive" justifications were the reasons for the No-Poach Provision—just as the plaintiffs need not show intent or motive. Indeed, "[t]he fact that Dr. McCrary focused primarily on what matters—the actual procompetitive effects of Section 7(d)—does not make his opinions unreliable. Quite the opposite." (Jimmy John's Resp. to Pls.' Mot. to Exclude 12, Doc. 196) (emphasis in original).

The plaintiffs also contend that Dr. McCrary's testimony must be excluded because it supposedly rests on insufficient facts. More specifically, they assert that his "report is devoid of documentary evidence" or "scientific analysis" showing that "the removal of the No-Poach Provision actually reduced any of his purported procompetitive benefits . . . ." (Pls.' Mot. to Exclude at 13). For support, they cite the Seventh Circuit case Gopralratnam v. Hewlett-Packard Co., where the court affirmed the exclusion of expert testimony that suffered from "myriad issues." 877 F.3d 771, 778 (7th Cir. 2017). (Id.). In that product-liability case, the plaintiff's expert opined, among other things, that an internal fault in a laptop's battery "was specifically caused by a manufacturing defect in the cell or a failure in the computer's electrical circuitry." 877 F.3d at 787 (emphasis added). Yet "despite concluding that a manufacturing defect led to the alleged internal fault," the expert "could not provide details as to what the specific defect was; why it transpired; when it occurred in the manufacturing process; or even where such manufacturing took place." Id. (emphasis in original). His opinion, therefore, was "simply too speculative to pass muster under Daubert and Rule 702." Id. The Seventh Circuit concluded by noting that the expert's testimony was plagued with "various shortcomings"; taken as a whole, the district judge did not abuse her discretion in excluding it. Id. Indeed, earlier in the opinion, the court revealed that the expert had "supported his premise only by citing to his own . . . article" and then unwittingly provided contrary sources. See id. at 785-87. By the time the court reached the expert's testimony about the specific cause of the fire, it came as no surprise that the expert failed to provide sufficient evidence leading to his conclusion. See id. at 787-88.

The plaintiffs' analogy to Hewlett-Packard is unpersuasive. True enough, during his deposition, Dr. McCrary was unable to point to specific documents showing when the No-Poach Provision first appeared in Jimmy John's Franchise Agreement. (Pls.' Mot. to Exclude at 13; McCrary Dep. 157:15-158:15, Doc. 185-3). But Dr. McCrary was not tasked with determining when the No-Poach Provision first appeared—for that, he relied on the plaintiffs' allegations in the Amended Complaint. (McCrary Dep. at 239:10-21; see, e.g., McCrary Report at 4-5). The Court rejects the plaintiffs' attempts to mischaracterize Dr. McCrary's deposition testimony and downplay the extent of his research by analogizing with the clearly unsupported expert testimony at issue in Hewlett-Packard. Cf. Petrogradsky Mejdunarodny Kommerchesky Bank v. Nat'l City Bank of N.Y., 170 N.E. 479, 483 (N.Y. 1930) (Cardozo, J.) ("[O]pinion has a significance proportioned to the sources that sustain it.").

To that end, the plaintiffs' contention that Dr. McCrary's testimony is unreliable because he did "no scientific analysis that the removal of the No-Poach Provision actually reduced any of his purported procompetitive benefits" also misses the mark. A central theme to Dr. McCrary's testimony is that Jimmy John's, like other franchisors, employs a variety of intrabrand restraints that work in tandem with one another, the No-Poach Provision being just one potentially procompetitive tool (once) at its disposal. Given that the bulk of his report featured an extensive discussion of labor economics and key concepts, Dr. McCrary's testimony is neither unreliable nor unhelpful. Indeed, he aptly demonstrated that his premise is "consistent with the labor economics literature on human capital investment and internal labor markets . . . ." (See McCrary Report at 57- 60). At the very least, his testimony would aid the Court in assessing the plaintiffs' assertions about the "Jimmy John's system," the "economic theory," and the "anticompetitive effects" discussed in their Motion for Class Certification. (See Pls.' Mem. in Support of Class Cert. at 4-18). That Dr. McCrary did not conduct another scientific analysis showing that removing the No-Poach Provision reduced the procompetitive benefits goes to his testimony's weight.

The plaintiffs also object to the record evidence that Dr. McCrary used, specifically the franchisee-declarations, or "unreliable facts." (Pls.' Mot. to Exclude at 16). The plaintiffs point to an inconsistency with the testimony provided by one of the franchisees—in his declaration, the franchisee said that he had never denied an employee's release request, even though an email suggested that he once had. (Id. at 17). But this argument too goes to the weight given to Dr. McCrary's testimony, as it challenges the soundness of the factual underpinnings of his methodology. See Manpower, 732 F.3d at 810. Even so, Dr. McCrary cited that declaration among four others to support his assertion: "Numerous franchisees testified that they did not enforce the No-Poach Provision, and others testified that they were not aware of" it. (McCrary Report at 49-50). At bottom, this dispute goes to the quality of Dr. McCrary's data, not his methodology. It is also a serious charge to allege that the declarations were "rigged," "biased," and "spoon-fed" by Jimmy John's counsel. The Court agrees with Judge Dow of the Northern District of Illinois that "[t]here is nothing unusual or improper about an attorney drafting a witness's declaration, so long as the witness reviews the declaration, agrees with its contents, and has a basis for his or her agreement . . . ." Luxottica Grp. S.p.A. v. Light in the Box Ltd., No. 16-cv-05314, 2016 WL 6092636, at *4 (N.D. Ill. Oct. 19, 2016).

Next, the plaintiffs contend that Dr. McCrary's model purporting to show a decrease in the use of certified managers from 2014 to 2018 is misleading based on the graph's caption. (Pls.' Mot. to Exclude at 1). Dr. McCrary's Exhibit 10 is captioned, "Percent of all stores with 0, 1, and 2+ Certified Managers." (McCrary Report at 87). The plaintiffs note, however, that the study was not a count of the number of certified managers, but a count of the number of shifts "covered by any number of Certified Managers." (Pls.' Mot. to Exclude at 14). They thus argue that the caption to Exhibit 10 is misleading because "a store with zero 'Certified Managers' in Dr. McCrary's Exhibit could, in fact have up to four certified managers on staff." (Id. at 15) (emphasis omitted). The Court disagrees.

The plaintiffs' argument fails for two reasons. First, Dr. McCrary states explicitly in the text of his report that he measures "how the use of Certified Managers changed during the class period" (McCrary Report at 88) (emphasis added), as in the number of "shifts . . . covered by a Certified Manager" (id. at 114) (emphasis added). Exhibit 10's caption is misleading only when taken out of context. At any rate, it is still consistent with Dr. McCrary's opinion that the reduction in the number of shifts covered by a certified manager is a proxy for the number of certified managers overall:

The entire purpose of the investment in Certified Managers is to leverage the improved performance that Certified Managers generate, and, thereby, enhance brand standards. It would not make any economic or business sense for a store to invest in training four Certified Managers, and then not use

them for any shifts. Indeed, it is just the opposite. Basic economic reasoning tells us that the number of shifts covered by Certified Managers would be closely related to the number of Certified Managers being trained and employed.

(McCrary Rebuttal at 9). In other words, Dr. McCrary established a rational connection between the data and his opinion; the Court will not exclude his testimony based on surface-level confusion.

Along those lines, the plaintiffs argue that Dr. McCrary's testimony must be excluded because he "did not even bother to control for the trend that his index share of shifts covered by Certified Managers was already in decline well before the Washington AG Settlement in July 2018." (Pls.' Mot. to Exclude at 15). They point to a regression model made by Dr. Singer purporting to show that, after adding "the relevant control variables from [his] wage regressions such as time trends" (Singer Rebuttal at 73), "the presence of the No-Poach Provision is associated with less training . . . , precisely the opposite of what Dr. McCrary predicts" (id. at 74) (emphasis in original).

Naturally, Jimmy John's argues that it is Dr. Singer's model, not Dr. McCrary's, that suffers from an "elementary error." (Jimmy John's Resp. to Pls.' Mot. to Exclude at 14). "The fundamental flaw in Dr. Singer's regression models is how he chooses to control for the time trend in Certified Manager usage." (McCrary Rebuttal at 18). In brief, "Dr. Singer inappropriately controls for a nonlinear time trend that he estimates using all the data, including data from after his events in interest . . . ." (Id. at 20) (emphasis in original).

For all that, the plaintiffs' challenge to Dr. McCrary's model goes to the weight of his testimony. As mentioned, "the exclusion of major variables or in the inclusion of improper variables may diminish the probative value of a regression model. But such defects do not generally preclude admissibility, and courts allow use of a regression model as long as it includes the variables accounting for the major factors." In re Urethane Antitrust Litig., 768 F.3d 1245, 1260-61 (10th Cir. 2014) (citing Bazemore v. Friday, 478 U.S. 385, 400 (1986)). That said, the Court is persuaded by Dr. McCrary's rebuttal, particularly the following graph, which depicts Dr. Singer's error and "how Certified Manager usage would ha[ve] evolved assuming the same trend after April 2016 as before April 2016":

[Redacted] (McCrary Rebuttal at 20). "The effect of this error is that it is impossible for Dr. Singer to obtain a reliable estimate of the impact of the events he is studying because his control variables for the time trend improperly heat the decline in Certified Manager shift coverage after the changes in the No-Poach Provision as part of a 'pre-existing' trend." (Id. at 21). After removing the "time trend variables and instead control[ling] for a pre-existing time trend," Dr. McCrary found that Dr. Singer's model affirmed his findings. (Id. at 22-23). The plaintiffs did not attempt refuting this in their reply brief. (See Pls.' Reply in Support of Its Mot. to Exclude 5, Doc. 206). At any rate, Dr. McCrary's defense holds water, and remaining doubts will go to his testimony's weight.

Next, the Court will address three arguments advanced by the plaintiffs about Dr. McCrary's release analysis. Recall that Dr. McCrary asserts that "documents describing [Redacted] requests for release of individual employees" from the No-Poach Provision reveal that the requests "were granted 88 percent of the time." (McCrary Report at 44) (emphasis in original). First, the plaintiffs contend that whether the No-Poach Provision "was enforced is a question for the jury." (Pls.' Mot. to Exclude at 18). In response, however, Jimmy John's properly cites Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338, 350-51 (2011), and Priddy v. Health Care Service Corp., 870 F.3d 657, 660 (7th Cir. 2017). (Jimmy John's Resp. to Pls.' Mot. to Exclude at 18). In those cases, the appellate courts reaffirmed that, in determining whether common questions predominate, the Court's analysis at the class-certification stage may overlap with the merits—that is, whether the No-Poach Provision was enforced. Indeed, the plaintiffs dedicate a section of their Motion for Class Certification to the question. (See Pls.' Mem. in Support of Class Cert. at 10-13). But see Messner, 669 F.3d at 811 ("[T]he court should not turn the class certification proceedings into a dress rehearsal for the trial on the merits.").

Second, the plaintiffs argue that the Court should exclude Dr. McCrary's testimony as "incomplete and unhelpful" because he supposedly could not "set[] forth any objective standard as to what it even means to 'enforce' the agreement, such that anyone could replicate his analysis." (Pls.' Mot. to Exclude at 18). Yet again, however, they mischaracterize Dr. McCrary's statements and ignore his report. For one, counsel for Jimmy John's appropriately objected to the vague questioning, evidenced by Dr. McCrary's confused response:

PLAINTIFFS' COUNSEL: The question was: Do you see no enforcement? And so that's a yes or no answer.

JIMMY JOHN'S COUNSEL: Object to form.

DR. MCCRARY: I'm not sure that's a yes/no answer.

(McCrary Dep. at 181:4-24). At any rate, Dr. McCrary sufficiently laid out the methodology underlying the release analysis:

I have performed a review of documents related to release requests that have been produced to date in this matter. Among [Redacted] unique release requests, I have identified [Redacted] that include clear responses to the request within the produced documents. Of the [Redacted] unique requests, [Redacted] were fully granted a release, leading to a release rate of 88 percent. Nearly always— [Redacted] out of [Redacted] times—employee releases were approved without conditions. Only [Redacted] approved employee releases were associated with conditions on reimbursement for training expenses. In other words, where there is sufficient evidence available to track the outcome of release requests, the requests were almost always granted, and almost always without cost.

(McCrary Report at 48).

And third, the plaintiffs suggest that Dr. McCrary's analysis is incomplete because he overlooked the fact that "Jimmy John's and numerous franchisees had blanket policies of not granting releases." (Pls.' Mot. to Exclude at 18). Except they provide no evidence showing that "a large fraction of stores employed such a policy with regard to releases or that such a policy was widely known by employees. Rather [they] simply assert this to be true." (McCrary Rebuttal at 33). Without more, the plaintiffs' argument amounts to mere speculation and does little to diminish the validity of Dr. McCrary's methodology.

The plaintiffs also challenge the economic principles underlying Dr. McCrary's opinion that Jimmy John's lacks incentives to suppress employee wages. Recall how Dr. McCrary suggests that in the textbook monopsony case, "revenue falls by virtue of the monopsonistic behavior, while profits rise." (McCrary Report at 55). So as labor demand decreases, so does output and revenue. (See id.). Because Jimmy John's collects royalties based on a percentage of franchisees' revenues, Dr. McCrary suggests that Jimmy John's would only lose money by engaging in monopsonistic behavior. (See id. at 55-56). The plaintiffs, on the other hand, argue that royalties are not Jimmy John's sole source of income. (Pls.' Mot. to Exclude at 19). Rather, they posit that reducing labor costs would increase franchisees' profits, thus attracting more franchisee-investors and permitting Jimmy John's to offset its losses through franchise fees. (Id.). For all that, the plaintiffs provide no support for this assertion. Dr. McCrary, on the other hand, notes at the onset of his Report that about "94 percent of the revenue [Jimmy John's] earned from franchisees was from revenue-based royalties paid by franchisees, and the remainder was from initial franchise fees." (McCrary Report at 53 n.181). The Court is not persuaded that Dr. McCrary's economic principles are unreliable.

Finally, the plaintiffs claim that Dr. McCrary's testimony is "pure theoretical speculation" because "he did not study the extent to which restaurant revenues were reduced by the No-Poach Provision." (Pls.' Mot. to Exclude at 20). Ironically, the inverse is true. Whereas the plaintiffs assume that "a reduction in output will also lead to a higher price," Dr. McCrary recognizes that such an argument would also suggest that Jimmy John's has monopoly power in the relevant market, which the plaintiffs make no attempt at showing. (See McCrary Rebuttal at 37). In other words, "even if a firm like [Jimmy John's] somehow was able to exercise monopsony power when hiring workers within its own brand through the use of an intrabrand restraint (like the No-Poach Provision), there is no way that such intrabrand restraints on hiring would somehow allow it to exercise monopoly power in the sale of its goods" given how vigorously QSRs compete "to sell their products at low prices." (Id.) (emphasis altered). Here too, the plaintiffs fall far short of casting doubt on Dr. McCrary's principles or methodology. Rather, his testimony "both rests on a reliable foundation and is relevant to the task at hand," the plaintiffs' Motion for Class Certification. See Daubert, 509 U.S. at 597.

For these reasons, the Court DENIES the plaintiffs' Motion to Exclude.

CONCLUSION

The Court GRANTS Jimmy John's Motion to Exclude Dr. Singer and DENIES the plaintiffs' Motion to Exclude Dr. Ordover and Dr. McCrary.

IT IS SO ORDERED.

DATED: February 16, 2021

/s/ _________

NANCY J. ROSENSTENGEL

Chief U.S. District Judge