Alaska Electrical Pension Fund v. Flowserve Corp.

PER CURIAM: Plaintiffs-Appellants Alaska Electrical Pension Fund and Massachusetts State Carpenters Pension Fund (collectively, “Alaska”) brought a putative class action against Defendants-Appellees Flowserve Corporation, Scott Greer, Renee Hornbaker, Banc of America Securities LLC, PricewaterhouseCoopers LLP, and Credit Suisse First Boston LLC (collectively, “Flowserve”) for violations of §§ 10 and 20 of the Securities Exchange Act of 1934 (the “Exchange Act”), 15 U.S.C. §§ 78j and 78t, and §§ 11 and 15 of the Securities Act of 1933 (the “Securities Act”), 15 U.S.C. §§ 77k and 77o. The district court denied Alaska’s motion for class certification, which ruling is before us in appeal number 07-11303, and granted the defendants summary judgment on all claims, which ruling is before us in appeal number 08-10071. In 07-11303, Alaska claims that the district court erred when it refused to certify Alaska’s proposed class, because in reaching its decisions the court (1) conducted the Federal Rule of Civil Procedure 23 certification hearing under a preponderance-of-the-evidence standard, rather than under a reasonable-trier-of-fact standard, and (2) applied an incorrect standard of loss causation to Alaska’s Exchange Act claims, resulting in an erroneous “predominance” determination under Rule 23(b)(3). In 08-10071, Alaska asks us to reverse the district court’s grant of Flow-serve’s motion for summary judgment on the merits of Alaska’s actions, contending that the court (1) was precluded from act *225 ing on the merits of Alaska’s claims during the pendency of its class-certification appeal in 07-11303, (2) erroneously concluded that the merits of Alaska’s Exchange Act claim were resolved by the court’s own resolution of loss causation for class-certification purposes, and (3) ignored the presumption of loss causation for Securities Act claims. Flowserve counters that even if the district court thus erred, its judgment should be affirmed because (1) the safe-harbor provision of the Private Securities Litigation Reform Act (the “PSLRA”), 15 U.S.C. § 78u-5(c)(l)(A)-(B), applies to the forward-looking projections at issue, and (2) the Securities Act’s one-year statute of limitation, 15 U.S.C. § 77m, bars Alaska’s §§ 11 and 15 claims. We reverse in part, vacate in part, and remand for further proceedings in the district court. I. FACTS AND PROCEEDINGS Flowserve Corporation manufactures pumps, seals, and valves, along with providing flow management services. In January 2000, the company made C. Scott Greer its CEO. Any honeymoon for Greer was brief: On February 9, 2000, Flow-serve reported net earnings of $12.2 million for fiscal year (“FY”) 1999, a 70% decline from its 1998 earnings, after which its stock closed at a 52-week low. To help address its financial difficulties, Flowserve undertook acquisitions in 2000 and 2002 that were designed to create synergies and to yield other benefits in a consolidating industry. Flowserve acquired Ingersoll Dresser Pump Company (“IDP”) in August 2000 for $775 million and the Flow Control Division of Invensys pic (“IFC”) in May 2002 for $535 million. To finance these acquisitions, Flowserve entered into a $1.1 billion credit agreement and twice offered public stock, once in November 2001 and again in April 2002. 1 The parties’ versions of Flowserve’s conduct diverge from there. Alaska claims that Greer, Renee J. Hornbaker, the CFO during the class period, and Flowserve Corp. through its officers schemed to misrepresent the company’s financial condition by making “inextricably related” false statements concerning (1) earnings forecasts, (2) historical financial performance, (3) past and future integration costs and savings related to the acquisitions of IDP and IFC, and (4) debt-covenant compliance. Flowserve disputes this characterization. As an example of this alleged scheme, Alaska points to a press release issued on the date that it selected to begin the putative class period, February 6, 2001. Flow-serve’s release reported earnings of $13.2 million for FY2000, in which its true earnings were $5.4 million (140% lower than reported), a fact not known publicly until Flowserve restated its earnings in 2004. The same press release announced reduced fourth-quarter 2000 earnings and a FY2001 projection lower than analysts’ expectations. According to Alaska, the FY2001 projection included fraudulent misstatements designed to increase expectations about integration synergies from the IDP acquisition; yet, Flowserve’s stock price declined 8% in response to the inflated earnings estimate. In addition to a variety of Flowserve’s other public statements that Alaska claims were fraudulent, it highlights an April 24, 2001 announcement of positive first-quar *226 ter 2001 earnings that overstated earnings and understated costs, in response to which Flowserve’s stock climbed 8%. The second-quarter 2001 earnings statement, released on July 24, 2001, was inflated as well, but also reduced 2001 earnings guidance, causing the stock to slip 10.8% (although it rebounded by 7.7% the next day). According to Alaska, Flowserve again engaged in fraud when it knowingly released overly optimistic FY2002 earnings guidance on October 22, 2001, the same day that it released inaccurate third-quarter 2001 results. 2 Flowserve’s stock rose 6.8% in response to these releases. After repeating the same FY2002 guidance on February 4, 2002 — the same day that it had released misstated fourth-quarter 2001 results, including a 600% understatement of loss — Flowserve downwardly revised its FY2002 guidance in July and September 2002. 3 After the July downward revision of the FY2002 guidance, Flowserve’s stock declined 37.4%, and declined another 38.3% after the September disclosure. Both releases blamed the reduced earnings on industry and market factors and did not disclose facts that Alaska contends were the real reasons for the decline, namely Flowserve’s failure to comply with debt covenants, misstated historical financial performance, and problems in extracting synergies from Flowserve’s acquisitions. Alaska introduced expert testimony to show that statistically significant proportions of those declines in share price were attributable to company-specific factors, not to market-wide factors. On February 3, 2004, Flowserve announced that it would downwardly restate earnings for 2000-2003 by $11 million. No statistically significant decline in share price occurred after this disclosure. The actual amount of the eventual restatement totaled almost $60 million, and Flowserve admitted that it had not been in compliance with its debt covenants. Alaska filed suit in federal district court against Flowserve Corp., Greer, and Horn-baker on August 7, 2003. An amended complaint filed on May 14, 2004 named as additional defendants (1) Banc of America and Credit Suisse First Boston, two of Flowserve’s underwriters for the stock offerings made during the class period, and (2) PricewaterhouseCoopers, Flowserve’s auditor during the class period. Alaska filed its final amended complaint on February 22, 2005, suing Flowserve, Greer, and Hornbaker under § 10 of the Exchange Act and 17 C.F.R. § 240.10b — 5; suing Greer and Hornbaker under § 20 of the Exchange Act and § 15 of the Securities Act as controlling persons; and suing all defendants under § 11 of the Securities Act for making stock offerings under a defective registration statement. All defendants moved to dismiss Alaska’s complaint under Federal Rules of Civil Procedure 9(b) and 12(b)(6). The district court denied these motions and, in a detailed opinion, denied the defendants request to file, an interlocutory appeal. Ryan v. Flowserve, 444 F.Supp.2d 718, 730 (N.D.Tex.2006). Alaska then moved to certify a class on May 30, 2006. It defined the class as “[a]ll persons who purchased the publicly-traded equity securities of Flowserve Corporation between February 6, 2001 and September 27, 2002,” subject to the customary exclusions. The district court conducted a class-certification hearing after which it denied certification in an opinion filed on November 13, 2007. Ryan v. Flowserve, 245 F.R.D. 560, 577 *227 (N.D.Tex.2007). In that same opinion, the court granted the defendants’ pending summary-judgment motions, thereby resolving the case on the merits as well. 4 On December 26, 2007, we granted Alaska’s timely motion in 07-11303 for permission to appeal the denial of class certification under Rule 23(f). On January 4, 2008, the district court entered final judgment on the merits, dismissing all of Alaska’s claims, from which judgment Alaska timely appealed in 08-10071. We consolidated the two appeals here. II. ANALYSIS We review the district court’s decision not to certify the putative class for abuse of discretion, but legal questions implicated by that decision are considered de novo. Gene & Gene LLC v. BioPay LLC, 541 F.3d 318, 325 (5th Cir.2008). We review a grant of summary judgment de novo, applying the same legal standard as the district court. Miller v. Gorski Wladyslaw Estate, 547 F.3d 273, 277 (5th Cir. 2008). Summary judgment is appropriate if the record demonstrates that “there is no genuine issue as to any material fact and that the movant is entitled to judgment as a matter of law.” Fed. R. Crv. P. 56(c). A. Class certification. Alaska’s Exchange Act claim requires it to show, in connection with the purchase or sale of a security, that the defendants made (1) a misstatement or omission, (2) of a material fact, (3) with scienter, (4) on which Alaska relied, and (5) that proximately caused Alaska’s loss. Greenberg v. Crossroads Sys., Inc., 364 F.3d 657, 661 (5th Cir.2004). Alaska’s class certification appeal concerns the application of Rule 23(b)(3)’s “predominance” requirement to the reliance element of Alaska’s Exchange Act claims. 5 Class certification requires that “questions of law or fact common to class members predominate over any questions affecting only individual members.” Rule 23(b)(3)(c). As to the reliance element of Alaska’s claims, “[i]f the circumstances surrounding each [class member’s] alleged reliance” on the purported misrepresentations differ, “then reliance is an issue that will have to be proven by each plaintiff, and the proposed class fails [the] predominance requirement.” Unger v. Amedisys Inc., 401 F.3d 316, 322 (5th Cir.2005). Alaska can “establish a class wide rebuttable presumption of reliance” using “the fraud on the market theory” of collective reliance. Id. at 322. “Under this theory, ... it is assumed that all public information concerning a company is known to the market and reflected in the market price of the company’s stock.” Greenberg, 364 F.3d at 661 n. 6. Thus, “when someone purchases a company’s stock,” a court “can presume that he relied on the supposition that the market price [was] validly set and that no unsuspected manipulation ha[d] artificially inflated the price.” Id. (internal quotation marks omitted). To establish the rebuttable presumption, the Supreme Court has explained, a plaintiff must prove that: “(1) the defendant made public material mis *228 representations, (2) the defendant’s shares were traded in an efficient market, and (3) the plaintiffs traded shares between the time the misrepresentations were made and the time the truth was revealed.” Id. at 661 (discussing Basic Inc. v. Levinson, 485 U.S. 224, 108 S.Ct. 978, 99 L.Ed.2d 194 (1988)). We have “tighten[ed] the[se] requirements.” Oscar Private Equity Invs. v. Allegiance Telecom, Inc., 487 F.3d 261, 265 (5th Cir.2007). Most notably, wé require a plaintiff additionally to “prove that the defendant’s non-disclosure materially affected the market price of the security.” Nathenson v. Zonagen Inc., 267 F.3d 400, 414 (5th Cir.2001) (quoting Abell v. Potomac Ins. Co., 858 F.2d 1104, 1120-21 (5th Cir.1988)). As we have explained, this essentially requires a plaintiff to demonstrate “loss causation.” Oscar, 487 F.3d at 265. To establish loss causation, a plaintiff must prove “(1) that the negative ‘truthful’ information causing the decrease in price [was] related to an allegedly false, non-confirmatory positive statement made earlier and (2) that it is more probable than not that it was this negative statement, and not other unrelated negative statements, that caused a significant amount of the decline.” Greenberg, 364 F.3d at 666. The district court concluded that Alaska failed to prove loss causation and that Alaska thus was not entitled to the rebut-table presumption of class wide reliance. Consequently, it held, Alaska did not satisfy Rule 23(b)(3)’s predominance requirement. It is this holding that Alaska challenges on appeal. We conclude that the district court failed to apply the proper legal standard, and we remand the case so that it may do so. 1. Alaska was required to prove loss causation by a preponderance of the evidence. Before addressing the merits of Alaska’s class certification appeal, we address its argument that the district court applied an erroneous standard of proof. The district court required Alaska to prove loss causation by a preponderance of the evidence. Alaska contends that the court should instead have inquired whether a reasonable trier of fact could have found loss causation by a preponderance of the evidence' — -a substantially less stringent standard. Our precedents foreclose that position. In Oscar Private Equity Investments v. Allegiance Telecom, Inc., we held “that loss causation [as an issue of predominance] must be established at the class certification stage by a preponderance of all admissible evidence.” 487 F.3d, at 269; see also Teamsters Local 445 Freight Division Pension Fund v. Bombardier, Inc., 546 F.3d 196, 202 (2d Cir.2008) (“[T]he preponderance of the evidence standard applies to evidence proffered to establish Rule 23’s requirements.”); In re Hydrogen Peroxide Antitrust Litigation, 552 F.3d 305, 320 (3d Cir.2008) (“Factual determinations necessary to make Rule 23 findings must be made by a preponderance of the evidence.”). And “the plain text of Rule 23 requires the court to find ... the facts favoring class certification,” Unger, 401 F.3d at 321, not to inquire whether a jury reasonably could do so. Alaska contends that reading our cases “so broadly” would create “conflict with Rules 23 and 56” and would have “the potential for intruding on plaintiffs’ [Seventh Amendment] right to a jury trial.” Appellants’ Opening Brief 63. Its argument proceeds as follows: When assessing a defendant’s Rule 56 motion for summary judgment, a district court is required to inquire whether a reasonable jury could find loss causation by a preponderance of the evidence. But a plaintiff cannot present that evidence to the jury if the district *229 court denies class certification under Rule 23 for lack of predominance. Thus, if the district court may assess predominance at the class certification stage under a higher standard of proof than that required under Rule 56, the plaintiff may be denied the opportunity to present sufficient evidence to the jury. This argument fails because it conflates the issue of loss causation for purposes of establishing predominance under Rule 23 with the issue of loss causation on the merits. “ ‘In determining the propriety of a class action, the question is not whether the plaintiff or plaintiffs have stated a cause of action or will prevail on the merits, but rather whether the requirements of Rule 23 are met.’ ” Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 178, 94 S.Ct. 2140, 40 L.Ed.2d 732 (1974) (quoting Miller v. Mackey International, 452 F.2d 424, 427 (C.A.5 1971) (Wisdom, J.)); see also Appellants’ Opening Brief 64 (“ ‘an evaluation of the probable outcome on the merits is not properly part of the certification decision’” (quoting Fed. R. Civ. P. 23(c) advisory committee’s notes)). The Rule 23 analysis “must focus on the requirements of the rule, and if findings made in connection with those requirements overlap findings that will have to be made on the merits, such overlap is only coincidental.” Bell v. Ascendant Solutions, Inc., 422 F.3d 307, 312 (5th Cir. 2005). The denial of class certification does not prevent a plaintiff from proceeding individually. And “the court’s determination for class certification purposes may be revised (or wholly rejected) by the ultimate factfinder.” linger, 401 F.3d at 323; see also Oscar, 487 F.3d at 269 n. 40 (“[L]oss causation, as an element of a 10b-5 claim, can[ ] be reexamined at summary judgment.”). In short, Alaska bore the burden of establishing loss causation by a preponderance of the evidence in order to obtain certification of its proposed class. This burden was independent of Alaska’s obligation to establish reliance on the merits, and the Rule 23 requirement does nothing to compromise Alaska’s opportunity to meet that distinct obligation. 2. The district court applied an improper standard for assessing loss causation. Loss causation requires proof of a causal connection between a misstatement and a subsequent decline in a stock’s price. In Dura Pharms., Inc. v. Broudo, the Supreme Court explained that, though necessary, evidence that a stock was purchased at a fraudulently inflated price is insufficient to prove causation. 544 U.S. 336, 342, 125 S.Ct. 1627, 161 L.Ed.2d 577 (2005). A plaintiff must also establish that after the purchase and before the' loss there was a disclosure of negative “truthful” information that was “related to [the] allegedly false, non-confirmatory, positive statement made earlier.” Oscar, 487 F.3d at 266; see Lormand v. US Unwired, Inc., 565 F.3d 228, 256 n. 20 (5th Cir.2009) (explaining that our “related to” test is in harmony with Dura’s “relevant to” standard). The loss must be caused because this truth “ma[de] its way into the marketplace,” not as a result of “changed economic circumstances, changed investor expectations, new industry-specific or firm-specific facts, conditions,” or other factors independent of the fraud. Dura, 544 U.S. at 342-43, .125 S.Ct. 1627. The central dispute between the parties concerns the definition of relevant corrective information for purposes of assessing loss causation. Flowserve argues, and the district court largely agreed, that a plaintiff must show a “fact-for-fact” disclosure of information that fully corrected prior misstatements. Joint Brief of Appellees Flowserve Corporation, C. Scott Greer, *230 and Renee J. Hornbaker 23; see also Ryan v. Flowserve Corp., 245 F.R.D. 560 (N.D.Tex.2007) (memorandum order). In other words, under this theory a fraud causes a loss only if the loss follows a corrective statement that specifically reveals the fraud. The only statements that meet this definition were made in 2004, when Flowserve restated its earnings and announced that it had not been in compliance with its debt covenants. Because no statistically significant decline in Flow-serve’s share price followed those disclosures, they did not cause Alaska any loss. Thus, under Flowserve’s view of relevant disclosures, Alaska has failed to establish loss causation. Alaska contends that loss causation may result when the “true financial condition” of a company becomes known — regardless of whether the disclosure of the company’s true financial condition corrects past misstatements. So long as the difference between the market’s erroneous perception of the financial condition of the company and the company’s true financial condition is a consequence of the fraud, Alaska argues, a loss that results from a revelation of the company’s true financial condition is caused by the fraud. Under this view, any disclosure of Flowserve’s weakening financial condition would relate to its earlier misstatements. Neither party is correct; the true standard lies in the middle. Flowserve’s position effectively does away with the fraud-on-the-market theory of reliance. If a fact-for-fact disclosure were required to establish loss causation, a defendant could defeat liability by refusing to admit the falsity of its prior misstatements. In re Worlds of Wonder Sec. Litig., 35 F.3d 1407, 1422 (9th Cir.1994). And if a “complete” corrective disclosure were required, defendants could “immunize themselves with a protracted series of partial disclosures.” Freeland v. Iridium World Commc’ns, Ltd., 233 F.R.D. 40, 47 (D.D.C. 2006). Thus, “[t]o be corrective, [a] disclosure need not precisely mirror [an] earlier misrepresentation.” In re Williams Sec. Litig. — WCG Subclass, 558 F.3d 1130, 1140 (10th Cir.2009). Alaska’s position similarly is untenable, because undisclosed information cannot drive down the market price of a stock. Only information known to the market can cause a loss. For this reason, only information known to the market is relevant under the fraud-on-the-market theory of class wide reliance. As we have explained, to establish loss causation this disclosed information must reflect part of the “relevant truth” — the truth obscured by the fraudulent statements. Our cases illustrate the application of this standard. For example, in Greenberg v. Crossroads Systems, Inc., 364 F.3d 657 (5th Cir.2004), the plaintiffs argued that they relied to their detriment on several alleged misstatements. Among them were (1) statements reporting the defendant’s financial results for the first quarter of Fiscal Year 2000 and detailed analysts earnings estimates for the fiscal second quarter, and (2) statements pertaining to the defendant’s third quarter earnings. Id. at 667-69. These statements caused a substantial loss, plaintiffs argued, when the stock price fell in part because of a subsequent press release predicting a significant revenue shortfall for the defendant’s third fiscal quarter. We explained that the press release was related to the latter category of statements, but not the former. As to the former, “there [was] no relationship between” the allegedly false statements and the subsequent release. Id. at 668. The release did not “report any concern that [the defendant’s] first and second quarter earnings [were] incorrect.” Id. To the contrary, it “ma[de] no *231 reference at all to [the] first and second fiscal quarters.” Id. Thus, the first category of statements could not support a presumption of loss causation. The second category of purported misstatements pertained to third quarter earnings. “Because the ... release clearly concerned a significant revenue shortfall for [the] third fiscal quarter, the plaintiffs [had] shown the requisite link between the [release] and th[ose] earlier statements.” Id. at 669. Lormand v. US Unwired, Inc., 565 F.3d 228, provides additional guidance. The defendant in Lormand v. US Unwired, Inc., uttered two general classes of misstatements. First, it touted becoming a Type II affiliate of Sprint, when in fact it had been forced into that arrangement by Sprint — and Type II affiliation posed significant financial risks that were not disclosed. Id. at 240-41. Second, US Unwired misled the public about the expectations it had for success after Sprint forced the company to enter the subprime cellular service market. Id. at 240-44. We discussed a series of disclosures through which some truth leaked out. Generally, the disclosures made increasingly clear that there were substantial problems in the subprime market and that expected subscriber growth in that market was unlikely to materialize. Id. at 259. Upon detailed examination of these disclosures, we concluded that none related to US Unwired’s conversion to Type II affiliation, and that US Unwired thus failed to establish loss causation with respect to the misstatements pertaining to that conversion. Id. at 260. On the other hand, the disclosures did relate to US Unwired’s misstatements about the subprime market. Id. at 260-62. For this reason, the plaintiffs could demonstrate that their losses were caused by those misstatements when the relevant truth leaked out. Id. at 262. Here, the district court appears to have demanded more of Alaska than is required by Greenberg or Lormand. For example, the district court held that the reductions in July ($1.70 — $1.90 per share) and September 2002 ($1.45-$1.55 per share) of the FY2002 earnings guidance did not reveal the “relevant truth” concerning the inaccuracy of the October 2001 FY2002 earnings projection ($1.90-$2.30 per share). It must have reasoned that the July and September 2002 statements need not have simply reduced Flowserve’s earnings-per-share guidance, but had directly to reveal that the October 2001 guidance was fraudulent. That is not required; it was enough that the market learned that the October 2001 guidance was wrong and that other negative information unrelated to the reduced FY2002 guidance did not cause the decline in Flow-serve’s share price. See In re Williams Sec. Litig. — WCG Subclass, 558 F.3d at 1140 (“Loss causation is easiest to show when a corrective disclosure reveals the fraud to the public and the price subsequently drops — assuming, of course, that the plaintiff could isolate the effects from any other intervening causes that could have contributed to the decline.”). Alaska offered expert testimony on the latter point, and the district court did not find that Alaska had failed to segregate loss caused by the “relevant truth” leaking out from loss caused by unrelated factors. Instead, the district court found that the July and September 2002 guidance did not relate to (or, in Dura’s terms, was not “relevant” to) the October 2001 guidance or any other alleged misstatements. This conclusion was erroneous. Flowserve complains that our approach turns the Exchange Act into a “scheme of investor insurance.” Dura Pharm., Inc. v. Broudo, 544 U.S. at 345, 125 S.Ct. 1627 *232 (internal quotation marks omitted). It ignores the Exchange Act’s scienter requirement. Flowserve was free to be wrong in its October 2001 earnings guidance and even for such error to cause investors loss when it was revealed in July and September 2002 — so long as Flowserve did not commit fraud. Only if Flowserve’s October 2001 guidance (or another alleged misstatement) was fraudulent would any loss it caused Alaska be actionable. The Exchange Act is not investor insurance, but neither is the possibility of recovery for fraud tantamount to insurance. The market may never have learned the “relevant truth” of Flowserve’s allegedly fraudulent statements concerning its past financials, acquisition synergies, and debt-covenant compliance from the July and September 2002 reduction-in-earnings guidance. Clearly, the market understood that there was some problem with Flow-serve’s business (or its business environment) when the company reduced earnings guidance and the market reacted by driving down Flowserve’s share price. But loss caused solely by a general impression in the market that “something is wrong” is insufficient to establish causation. See In re Williams Sec. Litig. — WCG Subclass, 558 F.3d at 1138 (“The inability to point to a single corrective disclosure does not relieve the plaintiff of showing how the truth was revealed; he cannot say, ‘Well, the market must have known.’ ”). Were this not so, we could have concluded that the subprime market information in Lormand related to the misstatements about Type II affiliation because that affiliation was the reason US Unwired was enmeshed in the subprime cellular market in the first place. Similarly, we could have concluded in Greenberg that the disclosure of reduced third-quarter earnings related to misstatements about first- and second-quarter earnings because the earlier shortfalls were surely some cause of the third-quarter shortfall. If Alaska cannot prove by a preponderance of the evidence that the market learned more than that Flow-serve’s earnings guidance was lower and so its business seemed less valuable, it cannot establish that its loss was caused by Flowserve’s misstatements pertaining to past financials, acquisition synergies, or debt-covenant compliance. Finally, the district court concluded without analysis that twenty-one of Flow-serve’s misstatements were confirmatory and therefore not actionable. We have held that “[i]f the market price was not actually affected by the statement, reliance on the market price does not of itself become reliance on the statement.” Nathenson, 267 F.3d at 419. We have also said that “information already known to the market” will not change a stock’s price. Greenberg, 364 F.3d at 663. But, those of Flowserve’s statements to which the market “actually” responded by pushing up the price of Flowserve’s stock are not confirmatory under this standard. Id. at 665. If Alaska has proved positive “actual effect” on the price of the stock after a statement issued, that statement is not merely confirmatory because the market’s valuation demonstrates that the statement revealed new, positive information. We leave it to the district court on remand to apply this framework in order to determine which statements were confirmatory and which were not. Because the district court conducted its Rule 23 hearing under an erroneous legal standard for proving loss causation, and we cannot discern the basis of its holding that some of Flowserve’s statements were confirmatory, we vacate its denial of class certification on Alaska’s Exchange Act claims and remand for a new class certification hearing. *233 B. Summary judgment. 1. The district court had jurisdiction to enter summary judgment. Before turning to the merits of the district court’s grant of summary judgment, we must address Alaska’s argument that the court lacked jurisdiction to enter that order. “The filing of a notice of appeal is an event of jurisdictional significance — it confers jurisdiction on the court of appeals and divests the district court of its control over those aspects of the case involved in the appeal.” Griggs v. Provident Consumer Discount Co., 459 U.S. 56, 58, 103 S.Ct. 400, 74 L.Ed.2d 225 (1982). But “the district court may still proceed with matters not involved in the appeal.” Alice L. ex rel. R.L. v. Dusek, 492 F.3d 563, 564-65 (5th Cir.2007) (per curiam). Alaska asserts that after we granted it permission to appeal the district court’s denial of class certification, that court lost jurisdiction over the merits of the Exchange Act and Securities Act claims and thus had no authority to enter summary judgment. This argument rests on Alaska’s contention that Flowserve’s negative-causation defense to Alaska’s Securities Act claim was “involved in” Alaska’s unsuccessful effort to certify the putative class because certification was denied on the basis of Alaska’s failure to establish loss causation. In short, the district court lacked jurisdiction to address Flowserve’s negative-causation defense, Alaska posits, because the issue of loss causation was on appeal to this court in connection with the denial of class certification. We disagree. As a matter of law, a district court’s findings in connection with a holding on class certification do not resolve loss-causation issues on the merits, even when — as here — the two issues are practically identical. “The findings made for resolving [the] class action certification motion servefd] the court only in its determination of whether the requirements of Rule 23 [were] demonstrated.” Bell v. Ascendant Solutions, Inc., 422 F.3d 307, 312 (5th Cir.2005); see also Unger, 401 F.3d at 323 (“the court’s determination for class certification purposes may be revised (or wholly rejected) by the ultimate factfinder”). When it denied class certification, the district court did not touch on the issue of loss causation for purposes of the merits of Alaska’s claims. Thus, there is no issue on appeal of the summary judgment order that was “involved in” the denial of class certification. Cf. Ry. Labor Executives’ Ass’n v. City of Galveston, Tex., 898 F.2d 481, 481 (5th Cir.1990) (a district court can proceed to resolve the merits of a case during the pendency of an interlocutory appeal from the grant or denial of a preliminary injunction). 2. The entry of summary judgment was erroneous. (a) Exchange Act Claims The district court held that its “loss causation holding in its denial of class certification [was] dispositive of Plaintiffs’ Exchange Act claims as a matter of law.” Order at 1, Ryan v. Flowserve, 245 F.R.D. 560 (N.D.Tex.2007). For the reasons we have now twice explained, that holding is erroneous. Even if, contrary to our conclusion, we had found no error in the district court’s class-certification decision, its loss causation holdings under Rule 23’s preponderance requirement would not govern the merits of Alaska’s claims. Further, we are convinced that a genuine fact issue exists on the material element of loss causation under the Exchange Act because a reasonable trier of fact could at the least conclude that the October 2001 statement concerning Flowserve’s FY2002 earnings caused some portion of Alaska’s loss after the “relevant truth” began to leak out in July and September 2002. We therefore *234 reverse the district court’s grant of summary judgment in favor of Flowserve Corp., Greer, and Hornbaker on Alaska’s Exchange Act claims. This ruling does not hamstring the district court during the class certification proceedings on remand. The district court must make an independent finding on loss causation for purposes of class certification under the preponderance-of-the-evidenee standard. (b) Securities Act Claims Once a plaintiff establishes a prima facie case under the Securities Act, loss causation is presumed. McMahan & Co. v. Wherehouse Entm’t, Inc., 65 F.3d 1044, 1048 (2d Cir.1995). Section 11(e) provides a means of rebutting that presumption. 15 U.S.C. § 77k(e) (2006). If a defendant can show that “any portion or all of such damages represents other than the depreciation in value of [the] security resulting from [the material misstatement] of the registration statement,” id., the presumption is rebutted for so much of the loss as is not attributable to the misstatement. The burden on the defendants to prove the affirmative defense in § 11(e) is “heavy” and arises out of “Congress’ desire to allocate the risk of uncertainty to the defendants in these cases,” but it is not insurmountable. Akerman v. Oryx Commc’ns, Inc., 810 F.2d 336, 341 (2d Cir.1987). This defense is frequently referred to as “negative causation.” E.g., In re Adams Golf, Inc. Sec. Litig., 381 F.3d 267, 277 (3d Cir.2004). The district court erroneously placed the burden of proving loss causation for the Securities Act claims on Alaska, even though it recognized the affirmative nature of the defense. For example, it held that “[w]hile the evidence establishes that the July and September Releases had a negative effective on the price of Flow-serve stock, the Releases, standing alone, do not demonstrate a corrective effect on the price of Flowserve stock.” Ryan v. Flowserve Corp., 245 F.R.D. at 579. It later held that a “single question is not significant probative evidence and fails to satisfy Plaintiffs’ summary judgment burden. Providing a scintilla of evidence is not enough; there must be evidence on which the jury could reasonably find for the plaintiff.” Id. at 581 (internal quotation marks omitted). The July and September 2002 releases may not have had a corrective effect, but the defendants were required to prove that no reasonable juror could believe that any portion of Alaska’s July and September 2002 losses was caused by the defendants’ alleged misrepresentations in the registration statements, i.e., the losses were caused by another factor. This poses quite a different question than the one posed by the loss-causation issue under the Exchange Act, even at the Rule 56 stage of proceedings. In short, even if we were to accept that the defendants may have proved that the July and September 2002 releases did not cause loss attributable to any misrepresentations in the registration statements, that does not prove that the July and September losses were unrelated to the registration statements. For example, there is analyst commentary in the record (although not necessarily believed by the analyst discussing it) that suggests concerns about debt-covenant compliance and perhaps some concern about Flowserve’s financials. This might not have been enough evidence to avoid summary judgment if Alaska had the burden of proving loss causation, but it is sufficient to defeat the defendants’ argument that all reasonable jurors must conclude that the defendants were not responsible for any of Alaska’s loss. We therefore vacate the district court’s grant of summary judgment dis *235 missing Alaska’s Securities Act claims. We vacate, rather than reverse, because further proceedings might lead to a finer partitioning of culpability among the different defendants. For example, Greer, Hornbaker, and Flowserve Corp. engaged in different acts than those engaged in by Bank of America and Credit Suisse First Boston, which in turn engaged in acts different from those engaged in by PricewaterhouseCoopers. On the briefs and record before us, we are not positioned to investigate differences between these defendants’ acts. Consequently, to the extent that any defendant can shoulder the burden of proving negative causation as an affirmative defense at the summary-judgment stage, the district court is not precluded from reinstating its judgment on remand as to that defendant. 3. We decline to reach the defendants’ alternative grounds for affirmance. Defendants argue that Alaska’s Securities Act claims are barred by the statute of limitations and that any forward-looking statements they issued are protected by the Private. Securities Litigation Reform Act’s safe-harbor provision. Their arguments implicate disputed factual issues and are best addressed by the district court in the first instance. III. CONCLUSION To be successful, a securities class-action plaintiff must thread the eye of a needle made smaller and smaller over the years by judicial decree and congressional action. Those ever higher hurdles are not, however, intended to prevent viable securities actions from being brought. For the foregoing reasons, we REVERSE the district court’s grant of summary judgment on Alaska’s Exchange Act claims, VACATE the district court’s denial of class-certification and the remainder of its grant of summary judgment, and REMAND this case for further proceedings consistent with this opinion. Footnotes [1] The registration statements filed in connection with these stock offerings form the basis of Alaska’s claims against Banc of America, Credit Suisse First Boston, and PricewaterhouseCoopers, as well as the Securities Act claims against Flowserve Corp., Greer, and Hornbaker. [2] Flowserve projected that FY2002 earnings would range from $1.90-$2.30 per share. [3] The July revision projected FY2002 earnings of $1.70-$1.90 per share. The September revision projected $1.45-$ 1.55 per share. [4] In fact, the district court only granted Flow-serve summary judgment on the Securities Act claims. It determined that Alaska’s Exchange Act claims were ’’moot” because of the court’s loss-causation ruling during the class-certification hearing and dismissed them. [5] Although Alaska moved for class certification of both its Exchange Act claims and its Securities Act claims, the district court denied certification of only the Exchange Act claims. We thus leave the Securities Act claims to the district court on remand.