When the emergence of big data and supercomputing drew complex statistical analyses—event studies, regression analyses, ANOVA methods and the like—out of the classroom and into the marketplace, those paying attention knew that courtrooms would not be far behind. Today, these complex statistical analyses give market-watchers (potential Plaintiffs) unparalleled ability to identify unnatural market movement and ferret out manipulation. When incorporated into a complaint, these analyses—and the reasonable inferences drawn therefrom—allow market manipulation cases to clear the motion to dismiss stage and progress to discovery, where the machinations behind those unnatural movements are (theoretically) laid bare.
Now, a suit in the Southern District of New York alleging gold price manipulation has brought the discoverability of those statistical analyses—and particularly, the analyses that potential plaintiffs perform, but ultimately do not rely upon in their complaints—into focus. Plaintiffs were ordered to produce the preliminary, uncited analyses performed by their consulting expert, and they have promised to petition the Second Circuit Court of Appeals for a writ of mandamus enjoining enforcement of that order. The Second Circuit’s ruling on that petition threatens to have far-reaching effects on the use of statistical analyses in a complaint and the scope of the attorney work product protection.
Summary of the Allegations
This case concerns an alleged coordinated effort to manipulate the gold benchmark. Plaintiffs, a number of individuals, businesses and funds, allege that Defendants, several large broker-dealer banks including UBS, Barclays, Deutsche Bank, Bank of America, and HSBC, manipulated the daily gold benchmark—the “fix”—by, inter alia, making spoof bids/asks or wash trades during the time period that the fix was calculated (the “PM Fixing”) to steer the fix in a specific direction, usually down. Plaintiffs’ theory of market manipulation stemmed from statistical analyses demonstrating that gold prices acted differently around the PM Fixing than they did at any other time of day. Specifically, Plaintiffs’ proffered statistical analyses showed that that gold prices went down around the PM Fixing more than they went up, and that when prices fell, they fell further than they increased. Moreover, when the gold price dropped during the PM Fixing, Defendants’ gold spot quotes were closer to one another’s quotes than other market participants and lower than other market participants. See Third Amended Complaint, In re Commodity Exch., Inc., No. 14-md-02548 (VEC) ¶¶ 123-283 (S.D.N.Y.) (ECF No. 183). These analyses supported the reasonable inference that Defendants were conspiring to manipulate the PM Fix.
The statistical analyses underlying the complaint’s theory of manipulation were the lynchpin of Plaintiffs’ claims. Without them, Plaintiffs’ claims could not have survived Defendants’ motion to dismiss. When it sustained Plaintiffs’ Consolidated Second Amended Class Action Complaint in October 2016, the Court recognized the importance of the analyses, observing, “[w]hether the detailed statistical analyses contained in the Complaint reveal ground truth about the activities of the Defendant banks who participated in the Gold Fix or are on the ‘lies, damn lies and statistics’ side of the dichotomy remains to be seen.” In re Commodity Exch., Inc., 213 F. Supp. 3d 631, 641 (S.D.N.Y. 2016).
The Discovery Demand
Consistent with their discovery obligations, Plaintiffs produced all the underlying data and analyses on which their complaint was based, including complete datasets, native copies of all graphs or charts, and all data inputs and outputs used to develop any graphs, charts, or analyses cited in their complaints. But Defendants demanded more. It was not enough, Defendants argued, that Plaintiffs produce the data underlying the conclusions set forth in the complaints; Defendants needed everything Plaintiffs’ experts prepared, including all analyses not disclosed in the complaint, all iterations of those analyses, and all data underlying those analyses. See Ltr. Mot. to Compel, In re Commodity Exch., Inc., 14-md-02548 (S.D.N.Y.) (ECF No. 361).
Defendants based their broad demand for all analyses prepared by Plaintiffs’ experts—whether or not they were cited in the complaints—on the argument that in drafting their complaints, Plaintiffs selectively relied on their experts’ analyses, and discovery of the entire corpus of their experts’ work was necessary to avoid the prejudice resulting from that selective utilization of the data. Essentially, Defendants argued that “if you torture the data long enough, it will confess,” and Plaintiffs must produce the entirety of that torture, not just the confession.
Plaintiffs opposed Defendants’ demand. Plaintiffs argued that the materials sought were subject to attorney work product protection, and that protection was not waived because Plaintiffs did not put those materials in issue. Moreover, Plaintiffs argued that since they did not intend to rely upon those analyses for class certification, summary judgment, or trial, Defendants’ demand sought materials that were irrelevant to the litigation going forward. See Ps’ Opp’n, In re Commodity Exch., Inc., No. 14-md-02548 (S.D.N.Y.) (ECF No. 365). As to Defendants’ claims that Plaintiffs utilized only a misleading selection of evidence, Plaintiff’s insisted that Defendants had everything they needed to challenge the verity of the conclusions set forth in their complaint; disclosure of materials beyond that was a bridge too far.
The Court’s Order
On February 25, 2019, Judge Caproni granted Defendants’ motion to compel. See Order, In re Commodity Exch., Inc., No. 14-md-02548 (S.D.N.Y.) (ECF No. 377). The Court considered the factors commonly considered in this jurisdiction when determining whether a party has waived work product privilege under Rule 502(a). Specifically, a waiver applies “to an undisclosed communication or information in a federal or state proceeding only if: (1) the waiver is intentional, (2) the disclosed and undisclosed communications or information concern the same subject matter, and (3) they ought in fairness to be considered together.”
The Court held that Plaintiffs, by making selective use of their experts’ analyses, waived work product protection as to the entirety of the experts’ work because it all concerned the “same subject matter” of gold price fixing.
The Court relied heavily on fairness concerns, largely echoing Defendants’ concerns that Plaintiffs were presenting a misleading slice of the analyses: “the withholding of information that would tend to undermine key statistical conclusions alleged in a complaint would, in this Court’s view, result in ‘a selective and misleading presentation of evidence to the disadvantage of the adversary.’ Fed. R. Evid. 502(a) Advisory Committee Note; Seyler v. T-Sys. N. Am., Inc., 771 F. Supp. 2d 284, 288 (S.D.N.Y. 2011). Intentional disclosure of only favorable statistical results is, by definition, selective and misleading. And in this case, such a selective disclosure would benefit Plaintiffs and prejudice Defendants, as Plaintiffs’ statistical presentation was central to the Court’s decision that Plaintiffs have plausibly alleged Defendants’ participation in price manipulation and an antitrust conspiracy.”
The Mandamus Petition
Just last week, Plaintiffs asked the district court to stay enforcement of its order to produce until a forthcoming petition for a writ of mandamus to the United States Court of Appeals for the Second Circuit can be resolved. Plaintiffs argue that mandamus is necessary to determine whether, under Federal Rule of Civil Procedure 26 and Federal Rule of Evidence 502(a), a plaintiff that includes non-testifying consultants’ economic analyses and data in a complaint retains work-product protection over those consultants’ economic analyses and data that were not utilized to support the allegations in the complaint.
Plaintiffs argue that the district court erred when it held that expert analyses not cited in the amended complaint were waived simply because they concern the same general subject of gold price fixing. See Mot. for Stay, In re Commodity Exch., Inc., No. 14-md-02548 (S.D.N.Y.) (ECF No. 380). Such a hair-trigger application of the subject-matter work-product waiver would subvert the Advisory Committee’s admonition that subject matter waiver be “reserved for . . .unusual situations” in which it is “necessary to prevent a selective and misleading presentation of the evidence.” Fed. R. Evid. 502 (a) Advisory Committee Note to 2011 Amendment. More importantly, it would chill litigants’ ability to discuss their case freely and critically with consulting experts. See Mfg. Admin. & Mgmt. Sys., Inc. v. ICT Grp., Inc., 212 F.R.D. 110, 118 (E.D.N.Y. 2002) (non-testifying expert is one “to whom an attorney may speak freely about litigation strategies and opinions without falling prey to the powerful jaws of mandatory disclosure.”).
Given the increased reliance on statistical analyses to both identify cases of suspected market manipulation and allow those cases to clear the motion to dismiss hurdle, Second Circuit guidance on the discoverability of those analyses will have far-reaching consequences in the investigation and litigation of market manipulation cases. Denial of Plaintiffs’ petition for a writ of mandamus threatens to upend the role that consulting experts play in preparation of the statistical analyses that underlie these and similar complaints, and it may seriously compromise the ability of attorneys to ferret out misconduct like the kind alleged here. On the other hand, granting Plaintiffs’ petition risks providing potential plaintiffs an undiscoverable forum to “torture” the market data until it yields an actionable case. Either way, this dispute over the uncited, unreferenced expert analyses merits close attention going forward.
This post was written by Peter J. Sluka.