defeating class certification by attacking plaintiffs' classwide...
TRANSCRIPT
Defeating Class Certification by Attacking
Plaintiffs' Classwide Damages Models
Today’s faculty features:
1pm Eastern | 12pm Central | 11am Mountain | 10am Pacific
The audio portion of the conference may be accessed via the telephone or by using your computer's
speakers. Please refer to the instructions emailed to registrants for additional information. If you
have any questions, please contact Customer Service at 1-800-926-7926 ext. 1.
TUESDAY, FEBRUARY 20, 2018
Presenting a live 90-minute webinar with interactive Q&A
Jeremy M. Creelan, Partner, Jenner & Block, New York
Aaron Dolgoff, Vice President, Charles River Associates, Boston
Cedric D. Logan, Partner, Wheeler Trigg O'Donnell, Denver
Tips for Optimal Quality
Sound Quality
If you are listening via your computer speakers, please note that the quality
of your sound will vary depending on the speed and quality of your internet
connection.
If the sound quality is not satisfactory, you may listen via the phone: dial
1-866-869-6667 and enter your PIN when prompted. Otherwise, please
send us a chat or e-mail [email protected] immediately so we can address
the problem.
If you dialed in and have any difficulties during the call, press *0 for assistance.
Viewing Quality
To maximize your screen, press the F11 key on your keyboard. To exit full screen,
press the F11 key again.
FOR LIVE EVENT ONLY
Continuing Education Credits
In order for us to process your continuing education credit, you must confirm your
participation in this webinar by completing and submitting the Attendance
Affirmation/Evaluation after the webinar.
A link to the Attendance Affirmation/Evaluation will be in the thank you email
that you will receive immediately following the program.
For additional information about continuing education, call us at 1-800-926-7926
ext. 2.
FOR LIVE EVENT ONLY
Program Materials
If you have not printed the conference materials for this program, please
complete the following steps:
• Click on the ^ symbol next to “Conference Materials” in the middle of the left-
hand column on your screen.
• Click on the tab labeled “Handouts” that appears, and there you will see a
PDF of the slides for today's program.
• Double click on the PDF and a separate page will open.
• Print the slides by clicking on the printer icon.
FOR LIVE EVENT ONLY
Predominance:
Comcast and the World That Came
Before
Jeremy Creelan
February 20, 2018
6
Pre-Amchem: Increasing Scrutiny of Certification
• As class actions used increasingly to bring mass tort claims in 1980s, courts increased scrutiny of certification on predominance and other Rule 23 grounds. For example:
– Castano v. Am. Tobacco Co., 84 F.3d 734, 740 (5th Cir. 1996) (tobacco): “The district court erred in its analysis in two distinct ways. First, it failed to consider how variations in state law affect predominance and superiority. Second, its predominance inquiry did not include consideration of how a trial on the merits would be conducted.”
– In re Am. Med. Sys., Inc., 75 F.3d 1069, 1085 (6th Cir. 1996) (penile implants): “As this case illustrates, the products are different, each plaintiff has a unique complaint, and each receives different information and assurances from his treating physician. Given the absence of evidence that common issues predominate, certification was improper.”
– Boughton v. Cotter Corp., 65 F.3d 823 (10th Cir. 1995) (hazardous emissions from uranium mill)
7
Amchem Prod., Inc. v. Windsor (1997):
“Cohesiveness” Test
• Purported class action on behalf of all those exposed to
defendant’s asbestos
• Settlement class – neither party intends to litigate
• District Court certified class based on shared experience of
asbestos exposure & interest in prompt compensation; held
that settlement was fair
• 3d Cir. vacated for lack of predominance under Rule
23(b)(3)
8
Amchem Prod., Inc. v. Windsor (1997):
“Cohesiveness” Test
• Ginsburg for Majority (6)
– “The Rule 23(b)(3) predominance inquiry tests whether proposed
classes are sufficiently cohesive to warrant adjudication by
representation.”
– Predominance readily met in consumer or securities fraud as well
as antitrust cases
– Predominance may be met in mass tort cases arising from common
cause or disaster
– Fact that all members had been exposed to asbestos was
insufficient to meet predominance, as different members were
exposed to different products for different amounts of time in
different ways, and differences in state law compounded those
disparities
– Settlement class – lower bar?
9
Amchem Progeny:
Predominance Requirement Powerful
• Predominance requirement satisfied:
– Smilow v. Sw. Bell Mobile Sys., Inc., 323 F.3d 32, 39–40 (1st Cir.
2003)
– Gunnells v. Healthplan Servs., Inc., 348 F.3d 417 (4th Cir. 2003)
– In re Visa Check/MasterMoney Antitrust Litig., 280 F.3d 124 (2d Cir.
2001)
• Predominance requirement NOT satisfied:
– Danvers Motor Co. v. Ford Motor Co., 543 F.3d 141, 148 (3d Cir.
2008)
– Broussard v. Meineke Disc. Muffler Shops, 155 F.3d 331 (4th Cir.
1998)
– Jackson v. Motel 6 Multipurpose, Inc., 130 F.3d 999 (11th Cir. 1997)
10
Amchem Progeny:
Predominance Requirement Unclear
• Individualized damages a bar to certification
– Randall v. Rolls-Royce Corp., 637 F.3d 818 (7th Cir. 2011)
– Blades v. Monsanto Co., 400 F.3d 562, 570-71 (8th Cir. 2005)
• Individualized damages NOT a bar to certification (majority)
– Yokoyama v. Midland Nat’l Life Ins. Co., 594 F.3d 1087, 1089 (9th
Cir. 2010)
– In re New Motor Vehicles Canadian Exp. Antitrust Litig., 522 F.3d 6
(1st Cir. 2008)
– Allapattah Servs. v. Exxon Corp., 333 F.3d 1248, 1261 (11th Cir.
2003)
11
Comcast v. Behrend (2013):
“Rigorous Analysis” of Predominance
• Purported class action on behalf of 2M Comcast
subscribers
• Sherman Act claims challenging alleged “clustering”
scheme to eliminate competition in Philadelphia area
• District Ct rejected 3 of Ps’ 4 theories of anti-competitive
impacts; accepted “overbuilder deterrence” theory b/c
“capable of measurement on classwide basis” and “capable
of classwide proof”
• Ps’ expert damages analysis included all 4 impacts; did not
isolate “overbuilder” theory
12
Comcast v. Behrend (2013):
“Rigorous Analysis” of Predominance
• Scalia for Majority (5)
– “Rigorous analysis” required at class certification to confirm Ps’
damages case consistent with liability case under FRCP 23(b)(3)
– Overlap with merits analysis
– Ps’ expert did not tie supra-competitive prices to “overbuilder
deterrence” theory damages incapable of measurement on a
classwide basis if based on Ps’ expert analysis
13
Comcast v. Behrend (2013):
“Rigorous Analysis” of Predominance
• Ginsberg for Dissent (4)
– Court should dismiss cert. as improvidently granted b/c question
presented whether expert evidence admissible at certification
– “no new ground” on FRCP 23(b)(3) standards
– Class may be certified even if damages not provable in aggregate
– Ps’ expert analysis did allow classwide proof of damages because
Comcast agreed that Ps’ 3 other theories had no impact on prices
14
Comcast v. Behrend (2013):
“Rigorous Analysis” of Predominance
• Key Takeaways
– Part of pantheon of recent SCOTUS cases shifting merits analyses
earlier in life of a case
– Lingering difficulties for appeals courts trying to review half-formed
factual record from certification stage
– Strategic Note: Identify mismatch between Ps’ theory and damages
model before class cert. consideration
15
Comcast v. Behrend (2013):
“Rigorous Analysis” of Predominance
• Questions
– Would courts take Scalia at his word? • Or limit Comcast to its facts?
• Or limit Comcast to nexus between damages theory and liability theory?
– Comcast addressed combined liability-damages class
certification: impact on liability-only or damages-only
classes?
– What relationship between Comcast and Daubert analysis?
– Would individualized damages calculations block
certification?
– Was Comcast about reliability of damages analysis or case
management or Due Process?
370 Seventeenth Street | Suite 4500 | Denver, Colorado 80202-5647 | P 303.244.1800 | F 303.244.1879 | wtotrial.com
SCRUTINIZING PROPOSED DAMAGES
MODELS UNDER COMCAST
Cedric D. Logan
February 20, 2018
Dukes dicta; Comcast implies Daubert applies
Circuit Courts agree that Daubert applies at class certification:
In re Blood Reagents Antitrust Litig., 783 F.3d 183 (3d Cir. 2015).
Unger v. Amedisys Inc., 401 F.3d 316, 323 n.6 (5th Cir. 2005).
Am. Honda Motor Co. v. Allen, 600 F.3d 813, 815-16 (7th Cir. 2010).
Sher v. Raytheon Co., 419 F. App’x. 887, 890 (11th Cir. 2011).
Ellis v. Costco Wholesale Corp., 657 F.3d 970 (9th Cir. 2011).
Modified Daubert:
In re Zurn Pex Plumbing Prods. Liab. Litig., 644 F.3d 604, 612 (8th
Cir. 2011) (“focused” Daubert analysis at class certification).
17
Question 1 Raised by Comcast: How
“Rigorous” Must the Analysis Be?
File a Daubert Motion or Not?
Tyson Foods v. Bouaphakeo, 136 S. Ct. 1036 (2016).
Donning and doffing case under the Fair Labor Standards Act.
Plaintiff introduced expert testimony with representative evidence
Held: “Representative evidence that is statistically inadequate or
based on implausible assumptions could not lead to a fair or accurate
estimate[.] Petitioner, however, did not raise a challenge to
respondents’ experts’ methodology under Daubert.”
Concurrence: What about the completely uninjured class members?
Dissent: District Courts must ensure continuing compliance with Rule
23.
18
Question 2: Must Damages be Measureable on A
Classwide Basis to Meet Predominance?
Certified Question:
Whether a district court may certify a class action without resolving whether
the plaintiff class had introduced [evidence] to show that the case is
susceptible to awarding damages on a class-wide basis.
Majority:
[R]espondents’ model falls far short of establishing that damages are capable
of measurement on a classwide basis.” “If the model does not even attempt to
[measure such damages], it cannot possibly establish that damages are
susceptible of measurement across the entire class for purposes of Rule
23(b)(3).”
Dissent:
“[T]he decision should not be read to require, as a prerequisite to class
certification, that damages attributable to a classwide injury be measureable
‘on a class-wide basis.’”
19
Early Circuit Court Decisions
In re: Rail Freight Fuel Surcharge Antitrust Litigation, 725 F.3d
244 (D.C. Cir. 2013)
“No damages model, no predominance, no class certification.”
Wallace B. Roderick Revocable Living Trust v. XTO Energy, 725 F.3d
1213 (10th Cir. 2013)
The Tenth Circuit reversed certification and remanded to the district
court to “consider the extent to which material differences in damages
determinations will require individual inquiries.”
20
Must Damages be Measureable on a
Classwide Basis? (Cont’d)
The need for individual damages calculations alone does not
defeat a finding of predominance:
Leyva v. Medline Industries, Inc., 716 F.3d 510 (9th Cir. 2013)
In re Whirlpool Corp., 722 F.3d 838 (6th Cir. 2013)
Butler v. Sears, 727 F.3d 796 (7th Cir. 2013)
In re Deepwater Horizon, 739 F.3d 790 (5th Cir. 2014)
In re Nexium Antitrust Litig., 777 F.3d 9 (1st Cir. 2015)
Roach v. T.L. Cannon, 778 F.3d 401 (2d Cir. 2015)
Neale v. Volvo Cars of N. Am., LLC, 794 F.3d 353 (3d Cir. 2015)
Brown v. Electrolux Home Products, 817 F.3d 1225 (11th Cir. 2016)
21
But the Door is Still Open
Fifth, Sixth and Seventh Circuits: those cases involved liability
only classes.
First, Second and Ninth Circuits all noted that there was an easy
way to calculate individual damages in those cases.
Still an open question after these cases about how to evaluate a
putative class where damages are not measureable on a
classwide basis and individual damages are not readily calculable.
22
Consider The Ninth Circuit
Individual calculations cannot defeat class certification.
Leyva v. Medline Industries, Inc., 716 F.3d 510 (9th Cir. 2013).
A proposed class damages model can pass muster even if it is “an
approximation . . . uncertain, contingent, or difficult of
ascertainment” for certain California state claims.
Pulaski & Middleman, LLC v. Google, Inc., 802 F.3d 979, 989
(2015).
“Class wide damages calculations under the UCL, FAL, and CLRA
are particularly forgiving.” Lambert v. Nutraceutical Corp., 870
F.3d 1170 (2017)
23
Consider The Ninth Circuit (cont’d)
So is Comcast toothless? Not so fast . . .
Doyle v. Chrysler Group, LLC (9th Cir. 2016) (unpublished):
Reversed certification because plaintiff had not demonstrated that
damages could be measured on a classwide basis.
Distinguished Leyva and Pulaski because those cases involved “a
common methodology for calculating damages”.
24
When Do Individual Damages Calculations
Tip the Predominance Scale?
Brown v. Electrolux Home Products, 817 F.3d 1225 (11th Cir.
2016)
Noted that the predominance issue is qualitative, and the individual
issues raised by damages determinations are often easy issues to
resolve when damages calculations are formulaic.
Recognized that an “exception” to the blackletter rule when
computing individual damages will be complex and fact-specific.
Remand to consider whether common issues of liability and damages
together predominate over individual issues.
25
Question 3 Raised by Comcast: How
Precisely Must the Liability Theory and
Damages Model Align?
Comcast: Class liability and damages theories must align:
“[A] model purporting to serve as evidence of damages in this class
action must measure only those damages attributable to that theory.”
“Calculations need not be exact . . . but at the class certification stage
(as at trial) any model supporting a ‘plaintiff’s damages case must be
consistent with its liability case’”
Model does not establish that class damages are capable of
measurement if the model “identifies damages that are not the result
of the wrong.”
26
Liability and Damages Must Align
From the defense perspective, this is probably where the most
fruitful application of Comcast has been, but the lower courts are
not aligned on what this requirement actually means in practice.
Compare:
Hughes v. Ester C. Co., NBTY, Inc., 2016 WL 6092487 (E.D.N.Y.
Sept. 30, 2016); In re 5-Hour Energy Mktg. & Sales Practices Litig.,
2017 WL 2559615, at *9–11 (C.D. Cal. June 7, 2017)
Goldemberg v. Johnson & Johnson Consumer Companies, Inc., 2016
WL 5817012 (S.D.N.Y. Oct. 4, 2016).
27
Consider The Ninth Circuit (cont’d)
Brazil v. Dole Packaged Foods, LLC, 2014 WL 5794873 (Nov. 6,
2014)
Plaintiff’s expert’s asserted that he could calculate classwide
damages: Class certified.
Proposed model was carried out, and failed: Class decertified.
Regression model did not account for advertising spending, prices of
competing products, and did not verify reliability of his assumptions
regarding competing products’ labels
Affirmed, 660 Fed.Appx. 531 (9th Cir. 2016): “Because Brazil did
not explain how this premium could be calculated with proof
common to the class, the district court did not abuse its discretion
by granting Dole’s motion to decertify.”
28
Consider the Ninth (again)
But see Briseno v. ConAgra Foods:
Plaintiffs’ expert asserted that he could use regression
analysis to determine damages, but could not isolate damages
flowing solely from allegedly false aspect of “100% natural”
claim.
Court granted amended motion for class certification that relied
on regression + conjoint analyses to isolate price premium.
Affirmed: “Not an abuse of discretion for the district court to
conclude that Plaintiffs’ proffered model tracked their theory of
liability and was therefore sufficient to survive class certification.”
29
Liability and Cognizable Damages
Theory Must Align
Harnish v. Widener University School of Law, 833 F.3d 298 (3d
Cir. 2016):
Not only must damages model measure damages attributable solely
to Plaintiffs’ theory of liability, but it must measure a legally cognizable
theory of damages pursuant to underlying state law.
A flawed damages model can prevent certification, even without
Comcast.
Did not cite Comcast.
30
Three Key Points
First, the “rigorous analysis” of damages evidence includes
considering whether class damages experts satisfy FRE 702 and
Daubert.
Second, although individual damage calculations alone will not
defeat predominance in most circuits, courts are increasingly
willing to consider the need for such inquiry as a factor in deciding
whether common issues outweigh individual issues.
Third, proof of class damages must measure only those damages
attributable to Plaintiffs’ theory of liability.
31
Arguments Under Comcast That
Have Been Successful
1. Insufficiently developed model.
District court cannot perform its “rigorous analysis” without substance.
2. Model fails to eliminate the need for individual inquiry.
Courts have recognized that the need for such inquiries can defeat
predominance.
3. Model fails to measure damages flowing from the alleged injury.
Or the method does not measure damages in a way that is
cognizable under the underlying substantive law.
32
Comcast and the Application of
Damages Methods at Class Certification
February 20, 2018
Aaron Dolgoff
Vice President
200 Clarendon Street Boston, Massachusetts 02116-5092 617-425-3000
Topics covered
Using examples from securities, antitrust and labor/employment class actions:
• Briefly introduce a few damages methods Plaintiffs might use at Class
Certification phase
• Discuss various Plaintiffs and Defendant arguments supporting or rebutting
these methods
• Provide some lessons from case outcomes
34
Securities Litigation Damages Approaches
• In securities class actions every class member will necessarily have
individualized damages calculations based on the size and timing of their
respective trades
– Individual calculations do not preclude a common approach (i.e., individual calculations are not
necessarily a barrier to class certification)
• Two generic approaches to damages in securities suits:
(1) Stock price inflation: Actual purchase price – “But-for” purchase price
Economic models are used to estimate the “but-for” price
(2) Rescission of out-of-pocket losses: Actual purchase price – Actual sale price
[perhaps offset for unrelated losses]
35
Securities Litigation Comcast Decisions – Summary
• Some damage methods are easier to establish classwide application
– Claims securities “worthless”/ rescission claims (Cromeans v. Morgan Keenan, Dodona v.
Goldman Sachs)
– Statutory formulas (§11 of 1933 Securities Act) (NJ Carpenters Health v Residential Capital)
– Burden remains on plaintiff to offer evidence of damages methodology (e.g., Kosmos Energy Ltd.
Securities Litigation in which the court found plaintiffs offered no evidence in support of
predominance requirement for a §11 claim)
• Damage claims requiring estimation of stock price inflation (i.e., 10(b)-5) generally
require an economic method and expert opinion
– Event study method has become most commonly applied method to determine stock price
inflation
– Applying an event study to a damages formula is insufficient: plaintiffs need to show how their
damages methodology tracks their theory of liability (e.g., BP Securities Litigation 1 – class
denied)
36
Securities Litigation Event Study Method
1. Identify events
– Disclosure vs misstatement
– Cannot directly test impact of an omission
2. Define length of event windows
– Tradeoff: isolating window to the identified event vs statistical confidence
3. Calculate excess return using an appropriate economic model
– Control for market, industry and/or other exogenous factors (i.e., factors unrelated to
the event being tested)
4. Test statistical significance
37
Securities Litigation Issues in Measuring Value Impact of an Event
• Confoundment (i.e., the isolation problem)
– Defining the event window
– Isolating individual effects (example: controlling for earnings announcements)
– Impact on statistical significance
• How does the event information relate to plaintiffs’ theory of harm?
– Undisclosed risk vs. Realization of risk
• Is the information really “news”?
– Public disclosure
– Dissemination mechanisms (e.g., analyst reports)
• Statistical confidence / reliability of model
38
Securities Litigation Does the Event Related to Theory of Harm? BP Securities Litigation II Example
0
10
20
30
40
50
60
70
80
90
30
-Jun
-07
31
-Jul-
07
31
-Aug
-07
30
-Sep
-07
31
-Oct-
07
30
-No
v-0
7
31
-De
c-0
7
31
-Jan
-08
29
-Fe
b-0
8
31
-Ma
r-0
8
30
-Apr-
08
31
-Ma
y-0
8
30
-Jun
-08
31
-Jul-
08
31
-Aug
-08
30
-Sep
-08
31
-Oct-
08
30
-No
v-0
8
31
-De
c-0
8
31
-Jan
-09
28
-Fe
b-0
9
31
-Ma
r-0
9
30
-Apr-
09
31
-Ma
y-0
9
30
-Jun
-09
31
-Jul-
09
31
-Aug
-09
30
-Sep
-09
31
-Oct-
09
30
-No
v-0
9
31
-De
c-0
9
31
-Jan
-10
28
-Fe
b-1
0
31
-Ma
r-1
0
30
-Apr-
10
31
-Ma
y-1
0
30
-Jun
-10
39
“Pre Explosion” or “Process Safety” Subclass
“Post Explosion” or
“Spill Severity” Subclass
Securities Litigation BP Securities Litigation II– Post-Explosion Subclass
• Plaintiffs’ damage method
– Damage theory: defendants withheld information about the severity of the oil spill
– Event study method applied to a set of disclosure dates
– Constant $ of inflation between disclosures
• Defendant arguments
– It is improper to carry-back stock price declines to the first alleged spill severity misstatement
because there is no evidence defendants had the same information at that earlier time
– Improper overlap of corrective disclosure events between Pre-Explosion and Post-Explosion
subclasses
• Subclass certified
– Defendants arguments concern loss causation – challenging the “fit” between alleged corrective
events and alleged fraudulent statements.
– Failure to prove loss causation is not an impediment to class certification
– “Plaintiffs’ task … is to present a legally viable, internally consistent and truly classwide approach
… Whether Plaintiffs have properly executed under the approach is a question for a different day.”
40
Securities Litigation BP Securities Litigation II– Pre-Explosion Subclass
• Plaintiffs’ damage method
– Damage theory: Consequential damages theory
• “Materialization-of-risk” - misstatements deprived investors the opportunity to divest prior to realization of risk
– Event study to measure price impact of realization of risk across 8 corrective events (reduced by amount known
to the market, i.e., risks voluntarily assumed by investors)
– Apportion across alleged misrepresentation dates (linear step-up over time)
• Defendant arguments
– Corrective events were not related to the process safety misstatements
– Arbitrary allocation (back-casting) method; can produce irrational results
– Failure to measure value of an undisclosed risk, as distinct from value impact from realization of that risk.
• Subclass not certified
– Selection of improper corrective events not relevant: loss causation need not be proven at class certification
– “Plaintiffs damage model [back-casting] need not be perfect. It need not be ‘correct.’ … [the] damages
methodology contains its flaws, but is not wholly arbitrary.”
– Fatal flaw denying certification: consequential damages theory cannot be applied uniformly
• Lumps together those who would have bought at heightened risk with those who would not
• Presumes investor reliance on factors other than price (i.e., risk), a theory inconsistent with fraud-on-the-market theory
41
Securities Litigation Alternatives to Event Study Method
• Indicators from factual record
– Contemporaneous analyst reports
– Plaintiff or other investor valuation analyses
• Direct valuation of alleged misstatement / omission
– E.g., discounted cash flow method to value an earnings misstatement regarding earnings
• Regression or other benchmarking techniques
– Statistically isolating effects of similar information disclosures in cross-sectional studies
– Issues
• Comparability
• Controls for case-specific factors
42
Antitrust
Damages Approach
• Antitrust damages may be measured by a model of overcharges, where actual
prices paid are compared to but-for prices
• Multiple regression is used to separate the impact of the alleged misconduct
from other supply and demand factors affecting price:
43
𝑷𝒓𝒊𝒄𝒆𝒊,𝒕 =∝ +𝜷𝑫𝒆𝒎𝒂𝒏𝒅 𝑭𝒂𝒄𝒕𝒐𝒓𝒔𝒊,𝒕 + 𝜸𝑺𝒖𝒑𝒑𝒍𝒚 𝑭𝒂𝒄𝒕𝒐𝒓𝒔𝒊,𝒕 + 𝜹𝑪𝒐𝒏𝒅𝒖𝒄𝒕𝒕 + 𝜺𝒊,𝒕
𝛿 = coefficient providing average overcharge estimate (by time period, or other
sub-class)
Antitrust Steps in Using Regression Models
44
1. Collect data
2. Specify the regression model equation
– Variables to represent supply and demand factors
– Controls for sub-classes (time periods, customer groups, etc.)
– Functional form (linear or nonlinear effects)
3. Estimate regression model using actual data
4. Test statistical significance (e.g.,is 𝜹 statistically different from zero?)
5. Use regression model to estimate predicted prices absent the alleged
misconduct (“But-for price”)
Antitrust Potential Certification Issues in Use of Regression Models
• Is the model specification reliable?
• Are there data integrity, sample size
or other statistical problems?
• Are 𝜷 and 𝜸 the same across
plaintiffs or various subclasses?
45
𝑷𝒓𝒊𝒄𝒆𝒊,𝒕 =∝ +𝜷𝑫𝒆𝒎𝒂𝒏𝒅 𝑭𝒂𝒄𝒕𝒐𝒓𝒔𝒊,𝒕 + 𝜸𝑺𝒖𝒑𝒑𝒍𝒚 𝑭𝒂𝒄𝒕𝒐𝒓𝒔𝒊,𝒕 + 𝜹𝑪𝒐𝒏𝒅𝒖𝒄𝒕𝒕 + 𝜺𝒊,𝒕
• Does 𝜹 vary across plaintiffs or
subclasses?
• Does 𝜹 vary between class members
and non-class members?
• Does 𝜹 reflect plaintiffs’ liability theory
and not other factors or alternative
liability theories?
Antitrust In re Rail Freight Fuel Surcharge Antitrust Litigation
• Plaintiffs alleged 4 rail freight carriers conspired to increase freight prices through fuel surcharges
• Plaintiffs’ expert controlled for various factors believed to affect prices of all rail shipments and found a
significant change in the relationship between fuel prices and freight rates at the start of the class
period
• Appellate court decision focused on one of Defendant’s rebuttal arguments: “false positives”
– Defendants found plaintiff’s model estimated damages for customers with rates set by contracts
signed before the alleged misconduct.
46
Plaintiff Reply to Defendant Arguments Court
Pre-class shippers are not part of class;
relevant issue is whether the class paid
higher rates
Although false positives don’t disprove the model… Plaintiffs
misapprehend their burden… It is not enough to submit a
questionable model whose unsubstantiated claims cannot be refuted
through a priori analysis… we have no way of knowing the
overcharges the damages model calculates for class members is any
more accurate than the obviously false estimates it produces for
legacy shippers.
The conspiracy may have predated the class
period, thus tainting even legacy contracts
Plaintiffs failed to present any evidence of this possibility, a claim that
is also at odds with the district court’s factual findings.
Antitrust In re High-Tech Employee Antitrust Litigation (1 of 2)
47
Plaintiff model Defendant Critique Court Finding
Conduct Regression
• Show impact of anti-
solicitation agreements
and quantify damage by
each proposed class
• Controls for benchmark
period of two years before
and after the class period.
• Plaintiffs should have produced
disaggregated models which
would show dramatically different
results for each Defendant.
• If benchmark period is restricted
to just after the class period the
model shows over-compensation
rather than under-compensation.
• Failure to control for value of
equity compensation.
• Failed to control for the fact that
compensation within firms is
correlated, thus requiring the use
of “clustered standard errors.”
• Defendant expert does not appear to have created a
truly disaggregated model… moreover, his use of so
many variables may “minimize artificially” the effects
of the anti-solicitation agreements. Aggregation may
also be appropriate given limited sample size.
• Defendants fail to explain why it makes sense to limit
the benchmark period (e.g., by showing that the pre-
conduct period is not comparable to the post-conduct
period).
• Defendant’s control for equity compensation (S&P
500) does not reflect variations in Defendants’ stock
prices or compensation.
• The fact that when standard errors are clustered the
results are not significant at the 95% level does not
render the regression inadmissible / unreliable.
Court not persuaded that the Conduct Regression by itself provides plausible method of showing
detrimental effects were experienced by all or nearly all class members… Nevertheless, the Court is
persuaded that the regression provides reasonable method of … showing impact generally, and
providing a measure of class-wide damages.
Antitrust In re High-Tech Employee Antitrust Litigation (2 of 2)
48
Plaintiff model Defendant Critique Court Finding
Common Factors Analysis
• Statistical model to assess
firm wide compensation
structures and the
formulaic way in which total
compensation varied over
time” (evidence of “rigid
wage structure”)
• Almost all of the model’s “fit” is
explained by employer and job title
• Cannot establish that compensation
of different employees with different
job titles are correlated over time.
• The analysis shows where an employee works
and what an employee does plays a large role in
determining salary. Plaintiff expert fails to explain
how one can infer from that that Defendant’s
salary structures were so rigid that compensation
for employees with different titles would
necessarily move together
Compensation Movement
Charts
• Track compensation for top
10 positions at Google and
Apple over time (to show
co-movement)
• Plaintiff examined only a small
fraction of the class and focused on
Technical Class employees
• In light of Plaintiffs failure to provide a broader
sampling, the Court is not persuaded that the
[charts] are particularly probative of whether
salaries for all or nearly all Class members moved
together
“the Court has concerns about the capacity of Plaintiffs’ evidence and proposed methodology to
prove impact to the All Employee Class or the Technical Class. The Court is most concerned about
whether the evidence will be able to show that Defendants maintained such rigid compensation
structures that a suppression of wages to some employees would have affected all or nearly all
Class members. The Court is also concerned that Plaintiffs’ proposed classes may be defined so
broadly as to include large numbers of people who were not necessarily harmed by Defendants’
allegedly unlawful conduct.”
Labor and Employment Minor League Baseball (Senne, et al. v. Kansa City Royals, et al.)
• Plaintiffs alleged FLSA and state wage and hour law violations (minimum wage and
overtime of minor league players)
• The court conditionally certified FLSA class with more than 2,000 players opting-in to
the class.
– Plaintiffs motion for state law class certification
– Defendants motion to decertify FLSA class
• In support of class certification Plaintiffs submitted expert declarations for survey and
damages experts
– Survey expert designed survey to obtain estimates of player hours worked at various activities / times of
year (e.g., spring training, championship season, winter); pilot survey implemented using opt-ins to FLSA
class.
– Damages expert demonstrated use of pilot survey data combined with other data (e.g., game schedule,
travel times between cities) to calculate damages
– Defendants filed motion to exclude Survey expert declaration and testimony (damages expert relies on
survey results, so exclusion of survey would effectively preclude damages analysis)
49
Labor and Employment Example from Minor League Baseball (Senne, et al. v. Kansa City Royals, et al.)
50
Defendants’ Expert Rebuttal criticism Court opinion
Pilot survey sample bias (respondents
were all class opt-ins)
No reason to believe a survey conducted at a later stage will suffer from this
defect
Potential bias from non-response
(about 1/3 response rate)
Defendants have raised the possibility of non-response bias, but not
demonstrated any actual bias
Bias from ambiguity in questions
combined with predetermined
category ranges that might influence
responses
The use of closed-ended questions is an accepted practice in conduct surveys
of this nature.
Complicated survey questions invited
“best guess” answers
“Troubled” by format of the question flagged… while on its own this problem is
not sufficient to render the survey unreliable, when combined with assumptions
about players’ memories, the reliability is questionable.
Self-interest bias The methodology used in conducting the pilot is fundamentally flawed and
plaintiffs have not shown they can overcome these problems when they
conduct the more comprehensive survey. All minor league players have a
vested interest, regardless of whether they opted in to the FLSA class.
Plaintiffs have not pointed to any specific records that can be used to validate
survey results. Self-interest bias is compounded by the ability of players to
remember mundane events to come up with reliable answers.
* The above does not address Plaintiffs’ subsequent attempts for reconsideration based on additional expert declarations.
Summary Implications
• Important to empirically test assumptions about reliability of damages model across
sub-classes or compared to those outside the class
– Not enough to show potential flaws in plaintiffs’ expert opinions – need demonstrate those flaws are real
and have measurable impact
• At Class Certification stage courts are not looking for perfection from plaintiffs’
experts
– Damages calculations need not be correct
– An appropriate model incorrectly applied may still be sufficient for class certification
– Remediable errors in Plaintiffs’ methods are less likely to prevent class certification
• Expert opinions must address fundamental issues:
– Do plaintiffs provide an approach to estimate a common impact across plaintiffs or sub-classes?
– Is the damage model consistent with plaintiffs’ liability theory?
– Is the damage model confounded by alternative liability theories?
– Are there reliable data (measurable, sufficient data, unbiased) to implement the damages method?
51