I. American Pipe Tolling and the Problem of Protective Filings
In American Pipe & Construction Co. v. Utah, 1 414 U.S. 538 (1974). the Supreme Court wisely rationalized class action law and policy under Rule 23 by holding that the filing of a class action complaint “suspends the applicable statute of limitations as to all asserted members of the class who would have been parties had the suit been permitted to continue as a class action.” 2 Id. at 554. The Court subsequently clarified that American Pipe’s protective rule applies not just to class members who intervene in the would-be class representative’s original suit but to “all members of the putative class,” including those who file individual lawsuits after certification is denied. Crown, Cork & Seal Co. v. Parker, 462 U.S. 345, 353-54 (1983). A contrary rule, the Court warned, would impair the “efficiency and economy of litigation” by inducing potential class members who want to proceed independently if class certification is subsequently denied to move to intervene or file entirely separate but essentially duplicative actions. 3 Am. Pipe, 414 U.S. at 553. Tolling thus avoids putting injured parties to an unnecessary and unfair Hobson’s choice: file a costly and duplicative action or risk surrendering their rights.
That longstanding rule is now under threat by a case currently before the Court, California Public Employees’ Retirement System v. ANZ Securities Inc., 4 In re Lehman Bros. Sec. & Erisa Litig. (CalPERS), 655 F. App’x 13 (2d Cir. 2016), cert. granted sub nom. Cal. Pub. Emps.’ Ret. Sys. v. ANZ Sec. Inc., No. 16-373, 2017 WL 125670 (U.S. Jan. 13, 2017). that asks whether American Pipe tolling applies to so-called “statutes of repose” in the securities laws. Virtually all federal securities causes of action have a two-tiered time bar: a shorter statute of limitations period governed by a discovery rule 5 The “discovery rule” applicable to securities class actions delays accrual of a cause of action until the plaintiff discovers, or with due diligence should have discovered, the “facts that will form the basis for an action.” Merck & Co. v. Reynolds, 559 U.S. 633, 644-46 (2010) (quoting 2 Calvin W. Corman, Limitation of Actions § 11.1.1 (1991)). and a longer limitations period, running from the violation, that is sometimes referred to as a “statute of repose.” 6 Lampf, Pleva, Lipkind, Prupis & Petigrow v. Gilbertson, 501 U.S. 350, 359-60 (1991). Were the CalPERS Court to affirm the Second Circuit’s decision below rather than the Tenth Circuit’s contrary approach, 7 Joseph v. Wiles, 223 F.3d 1155, 1166-69 (10th Cir. 2000). it would render American Pipe’s protective rule inapplicable to these latter repose periods.
This Essay offers a conceptual and empirical analysis of a key issue that overhangs the case: the plausible quantity of wasteful protective filings—including interventions and separately filed lawsuits—that putative class members might make if the Court were to hold that American Pipe tolling does not apply to repose periods. 8 See Police & Fire Ret. Sys. of Detroit v. IndyMac MBS, Inc., 721 F.3d 95, 109-10 (2d Cir. 2013), cert. dismissed sub nom. Pub. Emps.’ Ret. Sys. of Miss. v. IndyMac MBS, Inc., 135 S. Ct. 42 (2014) (noting the concern about protective filings).
In Part II we discuss one potential approach—the “natural experiment” approach used in the social sciences and in empirical legal studies scholarship 9 See generally, e.g., Jonah B. Gelbach & Jonathan Klick, Empirical Law and Economics (U. Penn. Inst. for Law & Econ. Research Paper No. 14-39, 2014) —to estimating the expected number of protective filings. As we explain, the obstacles to such an approach make it unworkable in the present context. We also point to significant flaws in the primary study to which CalPERS respondent ANZ Securities points in an effort to blunt the concern raised here—that denying application of American Pipe tolling to repose periods would uncork a substantial flow of protective filings.
In Part III we conduct an empirical study using data drawn from a comprehensive dataset of securities lawsuits. We count the number of cases in which class certification proceedings overrun repose periods. These are cases for which a narrowing of American Pipe’s reach plausibly could induce putative class members to make protective filings, whether in the form of interventions or newly filed lawsuits. We estimate that certification proceedings extend beyond the repose period in as many as half of securities class actions that reach a court order as to class certification and as many as one-quarter of all filed securities class actions. Of course, not all cases in which class certification proceedings extend beyond the repose period would yield protective filings. But simple math shows that even if protective filings are made in only a small share of cases where they are possible, the ultimate result would be a substantial spike in litigation in federal courts. Our analysis thus makes clear that the Court’s affirmance of the Second Circuit’s approach risks undermining the core purposes of the American Pipe rule: to promote the “efficiency and economy of litigation.” 10 Am. Pipe & Constr. Co. v. Utah, 414 U.S. 538, 553 (1974).
II. Research Design Considerations in Assessing the Importance of Protective Filings
A. The Challenges of an Ideal “Natural Experiment” Approach
One way to estimate the expected quantum of protective filings were American Pipe tolling held inapplicable to repose periods would be to use the “natural experiments” approach referenced previously. 11 See supra text accompanying note 9. The Second Circuit’s June 2013 IndyMac decision—the first case in which that court held American Pipe inapplicable to repose periods 12 IndyMac, 721 F.3d at 101. —would seem to provide an opportunity to use this approach. In principle, one could compare the quantum of protective filings across a “comparison” set of cases filed prior to the IndyMac decision in which the statutory repose period expired before IndyMac and a “treatment” set of cases that were filed before IndyMac but in which the statutory repose period did not expire until after IndyMac. 13 The resulting “straddle” method—examining cases filed before a rule change, and then comparing the incidence of litigation events that occur before or after that change—is a common means of mitigating selection bias. See William H.J. Hubbard, Testing for Change in Procedural Standards, with Application to Bell Atlantic v. Twombly, 42 J. Legal Stud. 35, 37-40 (2013).
But there are several obstacles to successfully deploying this approach. First, IndyMac concerned only claims brought under sections 11 and 12 of the Securities Act and so was merely the opening salvo among the Second Circuit’s holdings limiting American Pipe’s reach. It was only quite recently, in 2016, that the Second Circuit expanded its rule to the far more numerous claims brought under section 10(b) of the Securities Exchange Act and SEC Rule 10b-5. 14 See, e.g., SRM Glob. Master Fund Ltd. P’ship v. Bear Stearns Cos., 829 F.3d 173, 175 (2d Cir. 2016). Consequently, many putative class members in cases filed in district courts within the Second Circuit would have been uncertain as to whether they would enjoy American Pipe’s protection if class certification were denied. Such uncertainty would blunt any “treatment” effect of Second Circuit case law.
A second obstacle comes at the intersection of the current circuit split on American Pipe’s application to repose periods and the liberal rules governing personal jurisdiction and venue for claims brought under the federal securities laws. While the Sixth and Eleventh Circuits recently joined the Second Circuit in limiting American Pipe’s reach, 15 See Stein v. Regions Morgan Keegan Select High Income Fund, Inc., 821 F.3d 780, 794-95 (6th Cir. 2016); Dusek v. JPMorgan Chase & Co., 832 F.3d 1243, 1247-49 (11th Cir. 2016). the Tenth Circuit long ago took the opposite position, 16 Joseph v. Wiles, 223 F.3d 1155, 1168 (10th Cir. 2000). and the remaining circuits have yet to decide one way or the other. This is important because the nationwide service-of-process and permissive venue provisions in the federal securities laws grant plaintiffs who wish to pursue individual actions outside of class proceedings liberal choice of fora in which to bring suit. 17 See Securities Act § 22, 15 U.S.C. § 77v(a) (2015); Securities Exchange Act § 27, 15 U.S.C. § 78aa(a) (2015). As to personal jurisdiction, most courts agree that plaintiffs suing under either the Securities Act (for instance, section 11 or 12 claims) or the Exchange Act (for instance, Rule 10b-5 actions) need show only that the defendant has minimum contacts with the United States as a whole rather than individual states. See, e.g., SEC v. Ross, 504 F.3d 1130, 1139-40 (9th Cir. 2007); In re Fed. Fountain, Inc., 165 F.3d 600, 601-02 (8th Cir. 1999); SEC v. Sharef, 924 F. Supp. 2d 539, 544 (S.D.N.Y. 2013). As to venue, section 27 is especially permissive, rendering venue properly laid “in the district wherein any act or transaction constituting the violation occurred.” 15 U.S.C. § 78aa(a). The absence of significant jurisdictional hurdles further blunts the treatment effect of the Second Circuit’s changes in case law. If litigants who might wish to pursue separate actions can duck the Second Circuit’s holding by filing suit in the Tenth Circuit or in the circuits that have not yet considered American Pipe’s reach, then an empirical analysis keyed to case filings within the Second Circuit will understate, perhaps substantially, the impact that a Supreme Court decision limiting American Pipe would have across the entire federal system.
Even if these problems somehow could be overcome, there is a third fatal problem: insufficient data. 18 Stanford Securities Litigation Analytics (SSLA), which comprehensively tracks federal securities class actions, graciously provided data. See Stanford Securities Litigation Analytics, https://sla.law.stanford.edu (last visited Mar. 6, 2017). SSLA also provided the data for the analysis performed in Part III below. As an initial matter, claims brought under sections 11 and 12 of the Securities Act are not sufficiently numerous to generate reliable empirical estimates. 19 From 2007 to the present, district courts within the Second Circuit saw roughly seven lawsuits per year asserting section 11 and 12 claims. Claims under section 14 of the Securities Exchange Act were also sparse, having only recently increased from a dozen per year nationwide to a few dozen per year more recently, fueled by a rise in merger-objection suits. See Svetlana Starykh & Stefan Boettrich, Recent Trends in Securities Class Action Litigation: 2015 Full-Year Review, NERA Economic Consulting 5 fig.4 (Jan. 25, 2016), http://www.nera.com/content/dam/nera/publications/2016/2015_Securities_Trends_Report_NERA.pdf. But even if we were to use the more numerous cases asserting claims under section 10(b) and Rule 10b-5, the short time since the Second Circuit’s June 2013 IndyMac decision would severely limit the available treatment sample. For instance, in 8 of the 75 cases asserting section 10(b) claims filed in the Second Circuit between June 2011 and June 2013, the repose period expired as to at least some putative class members even before IndyMac was decided, once more blunting the treatment effect. And in 13 of the 75 cases, the entire case terminated before IndyMac, leaving no possibility for any treatment effect at all. 20 One case falls into both categories, in that the repose period would have run as to at least some putative class members, but the case terminated prior to IndyMac. We also face what statisticians call a right-censoring problem: in 32 of the 75 cases asserting section 10(b) claims filed over the same June 2011 to June 2013 span, the repose period has not yet expired even as of this writing for at least some putative class members.
Restricting our treatment sample to only those cases filed after the IndyMac decision would fare no better. Indeed, in all but 2 of the 84 cases asserting section 10(b) claims filed in the Second Circuit from June 2013 to June 2015—the two-year span after the Second Circuit’s IndyMac decision—the repose period still has not yet run for all putative class members as of this writing. And in 21 of these 84 cases, the repose period has not yet run as to any putative class member, again eliminating any possibility of a treatment effect. These problems grow more acute—and afflict more of our case observations—as we move backward or forward in time from the Second Circuit’s June 2013 IndyMac decision, significantly limiting the overall quantity and quality of available data observations.
B. The Weaknesses of “Opt-Out” Studies
This discussion shows that the most obvious “natural experiment” approach suffers so many problems as to be practically useless. In Part III below, we offer an alternative empirical approach. But before turning to that alternative, we discuss a different study—published by Cornerstone Research 21 See Amir Rozen et al., Cornerstone Research, Opt-Out Cases in Securities Class Action Settlements: 2012-2014 Update (2016), https://www.cornerstone.com/Publications/Reports/Opt-Out-Cases-in-Securities-Class-Action-Settlements-2012-2014; Amir Rozen et al., Cornerstone Research, Opt-Out Cases in Securities Class Action Settlements (2013), https://www.cornerstone.com/Publications/Reports/Opt-Out-Cases-in-Securities-Class-Action-Settlements. —which respondent ANZ Securities cites for the proposition that the flow of protective filings will be trivial. 22 Brief in Opposition to Petition for a Writ of Certiorari at 21-22, Cal. Pub. Emps.’ Ret. Sys. v. ANZ Sec. Inc., No. 16-373 (U.S. Jan. 13, 2017) (citing Cornerstone study). The Cornerstone study suffers from numerous flaws, at least when deployed in support of Respondent’s claims about protective filings.
First, the Cornerstone study tallies individual “opt-out” actions only in class actions that ultimately settled. The study therefore does not count protective filings in cases where class certification ultimately was denied or in cases that otherwise do not reach settlement (for instance, because of a dispositive motion). Such filings may be just as wasteful as individual filings in cases that settle.
Second, Cornerstone’s study considers only separate filings and does not appear to consider interventions filed by putative class members in the proceedings in which class certification was originally sought. This is important because interventions may be a significant way putative class members will preserve their rights if American Pipe tolling is unavailable and can consume substantial judicial resources. 23 Brief of Retired Federal Judges as Amici Curiae in Support of Petitioner at 16, CalPERS, No. 16-373.
Finally, the Cornerstone study is hampered by its limited time frame in relation to the relevant Second Circuit decisions. The Cornerstone data end in 2014, mere months after the Supreme Court dismissed the IndyMac case as improvidently granted, 24 135 S. Ct. 42 (2014). and two years before the Second Circuit extended its IndyMac decision to section 10(b) actions. 25 See SRM Glob. Master Fund Ltd. P’ship v. Bear Stearns Cos., 829 F.3d 173, 175 (2d Cir. 2016). There is thus virtually no opportunity for the effects of IndyMac to be reflected in the filings used in the study, creating a severe version of the problems afflicting the “natural experiment” approach described above.
III. How Many Cases Might Have Protective Filings in the Absence of American Pipe Tolling for Repose Periods?
A. Research Design
In this Part, we deploy an alternate methodology to measure the likely efficiency toll of a decision by the Court limiting American Pipe’s reach. To that end, we use historical data to count the number of securities class actions producing an order on a motion for class certification in which the court’s order granting or denying certification—or, in cases producing multiple certification orders, the last such order—came only after the applicable limitations period had expired. More specifically, we calculate the elapsed number of days between the first day of the class period specified in the operative complaint during class certification proceedings and either: (i) the date of the district court’s order on a motion for certification (or, in multi-certification-order cases, the last certification order) or (ii) the date of the district court’s order preliminarily approving the settlement class. 26 Keying this calculation to the start of the class period is consistent with section 13’s language, which states that the limitations period begins to run when the security was “bona fide offered to the public” (section 11 and 12(a)(1) claims) or upon the security’s “sale” (section 12(a)(2) claims). 15 U.S.C. § 77m (2015). This calculation permits us to tally the number of cases in which one or more putative class members would have needed, in the absence of tolling, to take protective action in order to preserve the right to proceed if class certification were later denied.
We constructed two datasets from a comprehensive database of securities case filings 27 See supra note 18. for the period 2002-2009. 28 We used 2002 as the front-end of our study window because data were not available for cases filed earlier. We used 2009 as our window’s back-end because, at the time the data were collected, it was the most recent year for which nearly the entire inventory of filed cases had been conclusively resolved. One data set contains all cases asserting only claims under sections 11 or 12 of the Securities Act over that period (as in CalPERS); 29 Section 11 of the Securities Act provides an express cause of action for investors in a public offering of securities that suffered “damages by material misrepresentations or omissions” contained in the company’s registration statement. Thomas Lee Hazen, Treatise on the Law of Securities Regulation § 7:2 (2016); see 15 U.S.C. § 77k (2015). Section 12 provides an express cause of action to investors “who purchase securities that were sold in violation of the [Securities] Act’s registration requirements.” Hazen, supra, § 7:2; see 15 U.S.C. § 77l. there were 86 such cases. The other contains cases asserting claims under section 10(b) of the Securities Exchange Act and SEC Rule 10b-5, whether or not those cases also asserted other types of claims (including claims under sections 11 and 12) filed during the same period; there were roughly 1200 section 10(b) cases filed during 2002-2009, from which we drew a random sample of 500 cases for analysis. We isolate cases asserting only section 11 and 12 claims because those claims are subject to the three-year limitations period in section 13 of the Securities Act, 30 15 U.S.C. § 77m. while section 10(b) claims are subject to a five-year limitations period. 31 28 U.S.C. § 1658(b).
Figure 1 (open by clicking here) offers a graphical summary of an analysis of the 86 securities class actions that asserted claims only under sections 11 or 12 over the period 2002 to 2009. The results are striking: section 13’s three-year limitations period, denoted in Figure 1 as a horizontal dashed line, would have expired prior to a certification decision in 73% (38 of 52) of cases that reached a certification decision and in 44% (38 of 86) of all filed cases. To provide more detail on the 52 cases depicted in the Figure that reached a certification decision, section 13’s three-year limitations period would have expired before an order on a motion for class certification in 11 of the 12 cases reaching an order resolving that motion. And that period would have expired before an order preliminarily approving a proposed class settlement in 29 of the 42 cases reaching an order resolving that motion. 32 Two of the cases in the sample of section 11 and 12 cases produced both an order on a motion for certification and a preliminary order approving a class settlement beyond the three-year limitations period, which explains why the numbers reported for cases for which the limitations period would have expired sum to 40 (11 + 29) rather than 38 and why the numbers reported for cases reaching the two types of orders sum to 54 (12 + 42) rather than 52.
This same approach also permits characterization of the efficiency costs of the Second Circuit’s decision 33 See SRM Glob. Master Fund Ltd. P’ship v. Bear Stearns Cos., 829 F.3d 173, 175 (2d Cir. 2016). to further limit American Pipe’s reach in the context of the far more numerous claims brought under section 10(b) of the Securities Exchange Act and SEC Rule 10b-5, which are governed by a statutory five-year repose period. 34 See 28 U.S.C. § 1658(b). To that end, Figure 2 (open by clicking here) presents a graphical summary of the same basic analysis as above, this time performed on our random sample of 500 cases asserting section 10(b) claims. 35 As with the prior analysis, keying the calculation of elapsed time to the start of the class period is consistent with the weight of authority among lower courts that § 1658(b)’s five-year limitations period is subject to an event-accrual rule—that is, the date of the misrepresentation or the completion of (or commitment to complete) the purchase or sale of the security. See, e.g., McCann v. Hy-Vee, Inc., 663 F.3d 926, 932 (7th Cir. 2011) (holding that the five-year limitations period starts upon misrepresentation); Arnold v. KPMG LLP, 334 F. App’x 349, 351 (2d Cir. 2009) (explaining that the limitations period starts when parties commit to purchase or sell). The results are again striking: the five-year limitations period would have expired prior to a certification decision in 44% (135 out of 307) of cases that reached a certification decision and in 27% (135 out of 500) of all filed cases in the sample. 36 The margin of error for these estimates, calculated at the standard 95% confidence level, is ±5.5 percent for the first and ±3.9 for the second. In other words, the 95% confidence interval is 38-50% for the first estimate and 23-31% for the second. To provide more detail on the 307 cases depicted in Figure 2 that reached a certification decision, the five-year limitations period that applies to such claims would have expired prior to an order on a certification motion in 42 of 86 cases that reached an order on such a motion. And that period would have expired prior to an order preliminarily approving a settlement class in 97 of 227 cases reaching an order on such a motion. 37 Four of the sample cases produced both an order on a certification motion and a preliminary order approving a class settlement beyond the five-year limitations period. This explains why the numbers reported for cases for which the limitations period would have expired sum to 139 (42 + 97) rather than 135. An additional two cases produced both types of orders before the expiration of the five-year limitations period, so that there were six cases with both types of orders. This explains why the numbers reported for the total number of cases reaching the two types of orders sum to 313 (86 + 227) rather than 307. Using the above estimates and extrapolating to the 4355 securities class actions filed since 1996 provides a more general estimate for the set of cases filed over the twenty-year period from 1996 to 2016: Plaintiffs seeking to preserve their rights without American Pipe’s protection might have filed protective actions in as many as 1175 cases. 38 See Alexander Aganin, Cornerstone Research, Securities Class Action Filings: 2016 Year in Review 40 (2017), https://www.cornerstone.com/Publications/Reports/Securities-Class-Action-Filings-2016-YIR (reporting more than 4355 securities class action lawsuits between 1996 and 2016). The “1175 cases” figure was derived by multiplying the 4355 cases filed since 1996 by the above-reported 27% estimate of the proportion of cases in the 500-case sample. See supra text accompanying note 37. Had even a handful of potential class members in each case done so as the end of the relevant three- or five-year limitations period approached, total filings, whether interventions or separate lawsuits, would have easily numbered in the thousands. This conclusion would hold even if protective filings occurred in only a fraction of those cases in which they might have.
While some may try to argue that securities cases filed between 2002 and 2009 are somehow idiosyncratic, or that a sea change in the composition of the case pool going forward will render any backward-looking estimate an uncertain guide to the future, several considerations suggest that the above estimates are, if anything, conservative.
First, the estimates do not account for the fact that a case can generate protective filings even if it never produces a certification order, so long as the case is not dismissed until after the limitations period expires. Figures 1 and 2 both suggest the existence of a non-trivial number of such cases—these are cases denoted with dots that fall above the horizontal dashed line drawn at the relevant three- or five-year limitations period. In such cases, a certification motion may have been filed but not yet ruled upon when the court granted a pending motion for failure to state a claim, judgment on the pleadings, or summary judgment.
Second, the above estimates do not account for the fact that, under the Second Circuit’s approach, a potential class member’s rights can be cut off by the relevant three- or five-year limitations period because of any defect that is fatal to a class claim, not just denial of certification. 39 The IndyMac case is illustrative, as the attempted intervention came after the district court dismissed some class claims on standing grounds because the lead plaintiff had not purchased some of the securities in question. Police & Fire Ret. Sys. of Detroit v. IndyMac MBS, Inc., 721 F.3d 95, 103 (2d Cir. 2013), cert. dismissed sub nom. Pub. Emps.’ Ret. Sys. of Miss. v. IndyMac MBS, Inc., 135 S. Ct. 42 (2014). Without American Pipe’s protective rule, absent class members who lack complete confidence that they have canvassed all possible legal hurdles to recovery may make protective filings even after class certification has been granted.
A third reason the above estimates are likely conservative requires consideration of possible dynamic responses by litigants and judges to a decision by this Court limiting American Pipe’s reach. On the one hand, a decision limiting American Pipe would create perverse incentives for litigants to delay pre-trial and certification proceedings to cut off potential class members’ opt-out rights. After all, once the relevant three- or five-year limitations period has lapsed, a decision denying class certification would become a victory on the merits as to any potential class members who did not take protective action. On the other hand, a decision limiting American Pipe’s reach might lead district judges to speed up their consideration of securities cases—thus de-prioritizing other cases—in an effort to preserve the ability of absent class members to make meaningful decisions about how to pursue their rights.
Measuring the relative size of such competing effects is challenging. It is famously difficult, as empirical scholarship in civil procedure shows, to gauge behavioral responses to changes in procedural rules. 40 See, e.g., David Freeman Engstrom, The Twiqbal Puzzle and Empirical Study of Civil Procedure, 65 Stan. L. Rev. 1203, 1223-29 (2013); Jonah B. Gelbach, Can the Dark Arts of the Dismal Science Shed Light on the Empirical Reality of Civil Procedure?, 2 Stan. J. Complex Litig. 223, 228-29 (2014). Still, our evidence gives good reason to conclude that the effect of the litigant response will equal or even exceed the effect of the judicial response. Figure 1 provides especially strong evidence in this regard: cases that reached a certification decision before section 13’s three-year limitations period expired tend to cluster just below that cut-off, making strategic delay plausible without American Pipe tolling. By contrast, cases that reached a certification decision after section 13’s three-year limitations period tend to be more diffusely distributed above that cut-off: in more than half (23 out of 38) of these cases, a judge would have needed to accelerate pre-certification proceedings by more than a full year in order to reach a certification decision before section 13’s three-year limitations period expired.
Our estimates strongly suggest that a decision by the Supreme Court limiting American Pipe’s reach will undermine American Pipe’s goal of “efficiency and economy of litigation.” 41 Am. Pipe & Constr. Co. v. Utah, 414 U.S. 538, 553 (1974). Empirical claims to the contrary, particularly where based on pre-IndyMac tallies of opt-out actions in settled cases, are an unreliable guide to the likely efficiency toll were the Court to adopt the Second Circuit’s position.
David Freeman Engstrom is Professor of Law and Bernard D. Bergreen Faculty Scholar, Stanford Law School. Jonah B. Gelbach is Professor of Law, University of Pennsylvania School of Law. Both authors thank Peter Davis for skillful research assistance.