JAMES G. CARR, Senior District Judge.
This securities fraud class action arises out of the collapse of the Dana Corporation in late 2005 and early 2006.
I previously certified a class of all persons who purchased Dana stock between April 21, 2004, and October 7, 2005. Plumbers & Pipefitters Nat'l Pension Fund v. Burns, 292 F.R.D. 515 (N.D.Ohio 2013). I reserved ruling on whether: 1) the class could include purchasers of Dana's bonds; and 2) if so, whether the claims of lead plaintiffs, who purchased only Dana stocks, were typical of the bondholders' claims. Id., 528-29.
The principal issue vis-a-vis the bondholders is whether the Dana bonds traded in an "efficient market." An efficient market is one in which a security's "market price fully reflects all publicly available information." In re PolyMedica Corp. Sec. Litig., 432 F.3d 1, 10 (1st Cir. 2005).
If the market for Dana bonds was efficient, the class members may invoke the fraud-on-the-market presumption to prove the reliance element of their claims. See Basic, Inc. v. Levinson, 485 U.S. 224, 242-245, 108 S.Ct. 978, 99 L.Ed.2d 194 (1988). In that case, certifying a bondholder class would be appropriate under Fed.R.Civ.P. 23(b)(3), which requires a showing that questions common to the class predominate over questions affecting only individual class members. But if the market for
To resolve this issue, I held an evidentiary hearing on July 10-11, 2013.
For the reasons set forth below, I am satisfied that the Dana bonds traded in an efficient market. I also conclude that lead plaintiffs' claims are typical of the bondholders' claims. Thus I grant the motion for class certification in full.
Plaintiffs seek to represent purchasers of nine different series of Dana bonds. Each bond has a unique alphanumeric code called a CUSIP number. For example, the CUSIP number for one of the Dana bonds is 235811AH9. Referring to the bond by the two-letter portion of their CUSIP numbers, the bonds at issue here are the AH, AJ, AK, AL, AU, AX, AY, BA, and MAB bonds.
The parties submitted conflicting expert testimony on the question whether these bonds traded in an efficient market. Courts consider five factors — the Cammer factors, after Cammer v. Bloom, 711 F.Supp. 1264 (D.N.J.1989) — when deciding if a security traded in an efficient market.
Plaintiffs argue that there are important structural differences between the corporate bond market and the stock market. They contend that, once I account for those differences, their evidence on each Cammer factor shows that the Dana bonds traded in an efficient market.
Defendants respond that only the fifth Cammer factor (whether there is a cause-and-effect relationship between the release of new, unexpected information about a company and the price of its securities) provides reliable economic evidence of market efficiency. Defendants argue not only that plaintiffs failed to introduce such evidence, but also that the defense expert proved the Dana bond market was inefficient.
Deciding whether to certify a bondholder class depends principally on which expert's testimony I find most apt and persuasive. With these arguments in mind, I discuss each party's evidence about the efficiency, or lack thereof, of the bond market in light of the Cammer factors.
Plaintiffs' expert was financial economist Jane D. Nettesheim. Nettesheim is the Vice-President of Stanford Consulting, which provides litigation support to plaintiffs in securities fraud cases. Nettesheim holds an M.B.A. and has testified previously in market-efficiency cases, but she has not authored any articles, peer-reviewed or otherwise, on market efficiency.
Nettesheim described the principal differences between the corporate bond market and stock market. In her view, these differences should affect how an economist determines whether a bond market is efficient.
In contrast to stocks, which trade on open exchanges like the New York Stock Exchange, corporate bonds trade in an over-the-counter market dominated by institutional investors. Bonds also trade less frequently and in greater volumes than stocks. The average bond trades only fifty-two days per year, and bond trades "tend to be at least 50 times as large as a typical stock exchange transaction." (Doc. 166-5 at 33). On a typical day in 2005, there were 22,000 transactions in the bond market worth $18 billion in par value.
Nettesheim testified that stock and bond prices do not react to the same types of information about the issuing company. Stock prices generally respond to short-term, issuer-specific events like dividend announcements and earnings reports. Bond prices typically react to macro-like events, such as fluctuations in market-wide interests rates, or issuer-specific information affecting the company's ability to make timely interest and principal payments.
Nettesheim also testified that negative information about a company, while affecting its stock price, may not significantly impact bond prices. This is so because an "equity cushion" may absorb the brunt of bad news.
Nettesheim reached her conclusion that Dana's bonds traded in an efficient market after: 1) gathering evidence relevant to each Cammer factor; and 2) conducting an event study that tested whether the bonds reacted to new, unexpected information.
The first Cammer factor is whether the security has a large average weekly trading volume. Cammer, 711 F.Supp. at 1286. An average weekly trading volume of two percent or more of outstanding shares triggers a "strong presumption" of market efficiency, while a one-percent average warrants a "substantial presumption." Id.
Nettesheim found that eight of the bonds met or exceeded the two-percent benchmark during the class period; the average weekly trading volume for these bonds varied from two percent (AX bond) to 18.7% (AY bond). The ninth bond (MAB bond) had an average weekly trading volume of one percent. Nettesheim also opined that these trading volumes indicated that the market for Dana bonds was liquid.
Nettesheim identified more than 200 analyst reports issued during the class period discussing Dana's securities. Although most reports focused on Dana stock, Nettesheim opined that even equity reports discussing "the financial condition and creditworthiness of the issuer (an important indicator as to whether the issuer would be able to pay interest, repay principal, and make those payments on time) ... [is] relevant to the market for bonds." (Doc. 197-9 at 14).
In addition, two credit ratings agencies, Moody's and S & P, rated all the Dana bonds throughout the class period. Fitch, another ratings agency, rated all but the MAB bond. Nettesheim opined that information provided by the ratings agencies helps investors evaluate the riskiness of a company's debt securities.
The third Cammer factor is whether the security had numerous market makers. Cammer, 711 F.Supp. at 1286-1287. Nettesheim explained that this factor is "concerned with whether market makers and the potential for arbitrage exists to facilitate trading." (Doc. 166-5). "Arbitrageurs are traders who identify and eliminate differences between [t]he price [of a security] and [its] value." Sullivan & Long, Inc. v. Scattered Corp., 47 F.3d 857, 862 (7th Cir.1995).
When a company first issues a bond, underwriters bring the bond to market and act as intermediaries between buyers and sellers. The Dana bonds had between two underwriters per bond (MAB bond) to fifteen underwriters (AX bond).
Nettesheim also found that "[n]umerous market makers were trading in the Bonds throughout the Class Period[.]" (Doc. 166-5 at 42). She cited data from the Financial Industry Regulatory Authority (FINRA) showing that "over 200 separately identifiable market makers traded in the Bonds during the Class Period[.]" (Id.). The FINRA data also showed how many times each market maker transacted in the bonds.
The fourth Cammer factor is whether the company was eligible to file an S-3 Form with the SEC. Cammer, 711 F.Supp. at 1287. An S-3 form is a shortened registration form available to companies that have filed SEC reports for twelve consecutive months and have a $75 million market-capitalization level. Dana last filed an S-3 in November, 1998, from which Nettesheim inferred Dana was eligible to file an S-3 during the class period.
The fifth factor is whether empirical evidence shows "a cause and effect relationship between unexpected corporate events or financial releases and an immediate response in the [security's] price." Cammer, 711 F.Supp. at 1287.
Nettesheim found that: 1) the bonds "reacted to material new information about Dana that was value-relevant with respect to the Dana Bonds"; and 2) the price movements "demonstrate[d] that the market for the ... Bonds was efficient during the Class Period." (Doc. 166-5 at 45).
As relevant here, Nettesheim's event study proceeded in three steps.
Nettesheim's first step was identifying the events to study. The first event she selected was Dana's announcement on September 15, 2005, that it: 1) was reducing its earnings projections for 2005 by more than fifty percent; and 2) may have to write down its deferred tax asserts. The second event was Dana's announcement on October 10, 2005, that it would: 1) restate its financial statements for fiscal year 2004 and the first half of fiscal year 2005; and 2) write off more than $740 million in deferred tax assets. Nettesheim characterized this information as unexpected, "cataclysmic," and relevant to bondholders. (Tr. 362).
As her second step, Nettesheim identified a control period starting April 21, 2004, and ending August 31, 2005. For each day in the control period, Nettesheim identified each Dana bond's daily return (i.e., the price change) and the daily return for an index of comparably rated bonds. Nettesheim then calculated the difference between each Dana bond's return and the comparable index return, giving her the premium return. Next, she calculated the difference between each bond's daily premium return and the mean premium bond return during the control period, which gave her the bond's premium excess return.
In her third step, Nettesheim measured whether any of the premium excess returns for the Dana bonds were statistically significant. She first identified the standard deviation of each bond's premium excess returns, and then divided each bond's daily premium excess return by the standard deviation. These calculations gave Nettesheim a figure called the t-statistic. If the t-statistic was greater in value than 1.97, Nettesheim concluded that the bond's premium excess return was statistically significant at the ninety-fifth-percent level of confidence.
Nettesheim determined that the Dana bonds had negative, statistically-significant premium excess returns after the September 15 and October 10 disclosures. Nettesheim found that:
In light of these returns, Nettesheim opined that the market for Dana bonds was informationally efficient.
To calculate the bonds' premium excess returns and determine their statistical significance, Nettesheim needed to know each bond's price on each day of the class period. But the bonds did not trade every day. Therefore, on days when no transaction prices were available, Nettesheim used matrix prices to complete her calculations.
A matrix price is an estimated price for a security generated, in the absence of a transaction price, by a proprietary formula.
Nettesheim also testified that investment professionals use matrix prices to value their holdings, and that parties in securities fraud cases rely on them for the same reasons she did here.
Nettesheim opined that matrix prices were a reliable proxy for transaction prices because matrix prices for the Dana bonds generally moved in the same direction as actual transaction prices. She based that opinion partly on a visual comparison between the movements of the two price sets. She also relied on calculations showing that, on 68.3% of all nine bonds' trading days, the matrix price was within $.50 of that day's high or low transaction price. Nettesheim selected the $.50 benchmark because, in her view, that figure represents a reasonable bid-ask spread in the bond market.
As noted, Nettesheim's event study focused only on Dana's release of "cataclysmic" information on September 15 and October 10, 2005. However, an exhibit to her expert report included more than 200 news items about Dana released during the class period. Among those items were documents in which defendants allegedly misrepresented Dana's financial health.
Nettesheim did not identify any of these "misrepresentation dates" as events for purposes of her event study. She acknowledged that, in conducting event studies in past cases, she had studied events like the earnings announcements in which defendants allegedly made false statements. But Nettesheim explained that, unlike this case, the past cases involved the market for stocks, not bonds.
Asked how analyzing the misrepresentation dates would affect her opinion on market efficiency, Nettesheim responded that "it wouldn't." (Tr. 181). Although "earnings announcements are certainly important to bond investors," Nettesheim did not expect "much of an impact on the bond price" until the news "makes its more uncertain as to the future payoffs of the bonds." (Id.).
Financial economist Andrew Roper testified for the defense. Roper holds a Ph. D. in finance and is a principal at the consulting firm Cornerstone Research. He is also a lecturer in law at Stanford Law School, where he teaches a course in securities fraud. Roper has previously opined in four securities fraud cases, concluding in each one that the plaintiff failed to demonstrate market efficiency. This is Roper's first case opining on the bond market.
In Roper's view, plaintiffs failed to produce reliable economic evidence that the Dana bonds traded in an efficient market. On the contrary, Roper found that the market was inefficient.
Dr. Roper based his opinion partly on what he deemed infrequent trading in the Dana bonds. He testified that active trading "is central to the question" whether a market is efficient because active trading fosters liquidity. (Tr. 242). Active trading also provides cover to arbitrageurs seeking to trade quickly and inconspicuously, so as to not move the security's price and eliminate their profit.
The heart of Roper's opinion was that Nettesheim did not, in fact, conduct an
Roper first criticized Nettesheim for studying only the corrective disclosures on September 15 and October 10. Because Nettesheim's goal was to identify a cause-and-effect relationship between new information and bond prices throughout the class period, Roper testified that Nettesheim should have studied additional events besides those at the end of the class period. However, Roper did not identify any specific event during the class period that he believed Nettesheim should have studied. Nor did Roper conduct his own event study.
Roper did not dispute that the September 15 and October 10 disclosures were associated with negative, statistically-significant returns across the Dana bonds. But he disagreed that those returns were relevant to market efficiency for the duration of the class period.
Second, Roper faulted Nettesheim for relying on matrix prices. In Roper's view, matrix prices cannot reliably substitute for transaction prices because there is no way to know whether matrix prices reflect how a company's securities react to new, company-specific information.
Roper testified that matrix prices are "algorithmic prices" dependent on inputs into a proprietary algorithm. (Tr. 275). Although secret, the inputs generally include changes in interest rates and the prices of comparably rated bonds.
According to Roper, the problem with relying on the price of a comparably rated bond (Bond A, for example) to generate a matrix price for Bond B is that, in general, the price of Bond A will not react to company-specific information concerning the issuer of Bond B. Thus, matrix prices cannot establish whether the price paid for a security reflects new, unexpected information about the security's issuer.
Roper also disputed Nettesheim's conclusion that matrix prices and transaction prices moved in the same direction. According to his calculations, the two price sets moved in the same direction only twenty-two to forty-five percent of the time, depending on the bond.
Roper agreed that one could infer from those calculations that matrix prices and transaction prices moved in the same direction between fifty-five percent and seventy-eighty percent of the time. However, he explained that that range would include not only instances when the price sets moved together, but also instances when the matrix price changed but the bond price did not (because there was no trade). Roper did not provide a breakdown between those two categories.
Roper further disagreed with Nettesheim's finding that matrix prices are reliable because of the frequency with which a matrix price was within $.50 of a high or low transaction price.
According to Roper, Nettesheim's decision to examine the relationship between a matrix price and a trading day's high or low transaction price biased her results in favor of finding matrix prices reliable. For example, the matrix price for the AH bond on April 6, 2005 ($99.22) was within $.50 of the day's low transaction price ($99.50). But the matrix price was nearly two dollars removed from the high transaction price ($101, at which three trades occurred) and more than $1.25 removed from the median transaction price ($100.50, at which four trades occurred).
Roper also testified that the $.50 benchmark used by Nettesheim is irrelevant to market efficiency. Rather, the economic literature from which Nettesheim drew
In a weak-form efficient market, a security's price fully and rapidly reflects only the security's historical price data. Thus, if a market is weak-form efficient, one cannot predict how a security will trade in the future based solely on how it traded in the past.
Roper opined that the market for Dana bonds was not even weak-form efficient. To reach this conclusion, he conducted a regression analysis commonly accepted by financial economists. Roper's analysis showed that knowing the Dana bonds' returns on the prior trading day helped predict their returns on the next trading day. This result is known as a lead-lag relationship or autocorrelation.
Roper agreed that, in theory, non-synchronous trading can produce spurious autocorrelation findings. Non-synchronous trading refers to a discrepancy between the time at which a security's price is reported (every day at market's close, for example) and the time when the security actually traded (earlier in the day, for example). Calculating a security's return on the false assumption that the return measures a price change between two equally-spaced intervals "can create a false impression of predictability in price changes and returns even if true price changes or returns are statistically independent." (Doc. 197-9 at 23-24) (quoting Campbell, et al., The Econometrics of Financial Markets 85 (Princeton Univ. Press 1997))
Roper testified that, in practice, non-synchronous trading is unlikely to explain a finding of autocorrelation. Based on a test he performed, Roper determined that non-synchronous trading could not explain the autocorrelation he detected among the Dana bonds. However, Roper did not discuss that test, its methodology, or its results in either of his two expert reports.
Although Roper found a high degree of predictability among the bond returns, he did not attempt to determine whether investors had used the lead-lag relationship to profit on the bonds.
Roper testified that only the fifth Cammer factor — the cause-and-effect relationship between new, unexpected information and a security's price — provides relevant, economically reliable evidence that a market is efficient.
Plaintiffs invoke the fraud-on-the-market theory to show that they can prove reliance — an essential element of their claims, see Stoneridge Inv. Partners, LLC v. Scientific-Atlanta, 552 U.S. 148, 159, 128 S.Ct. 761, 169 L.Ed.2d 627 (2008) — on a class-wide basis. They can do so, and thus become entitled to a class-wide presumption of reliance, if they prove, inter alia, the Dana bonds "traded in an efficient market." Erica P. John Fund v. Halliburton, ___ U.S. ___, 131 S.Ct. 2179, 2185, 180 L.Ed.2d 24 (2011).
In resolving this issue on a motion for class certification, I must be satisfied "after a rigorous analysis" that the Dana bonds traded in an efficient market. Wal-Mart Stores, Inc. v. Dukes, ___ U.S. ___, 131 S.Ct. 2541, 2552, 180 L.Ed.2d 374 (2011).
For purposes of the fraud-on-the-market theory, market efficiency means that "prices will reflect all relevant information, a definition of efficiency known as informational efficiency." In re Diamond Foods, Inc. Sec. Litig., 295 F.R.D. 240, 247, 2013 WL 1891382, *6 (N.D.Cal.2013). "The fraud-on-the-market theory is concerned with whether a market
When the Supreme Court adopted the fraud-on-the-market theory, "it did so in the context of stocks trading on the NYSE." In re HealthSouth Corp. Sec. Litig., 261 F.R.D. 616, 630 (N.D.Ala.2009). However, courts evaluating the efficiency of bond markets "tend to apply the same factors as applied to common stock." In re DVI, Inc. Sec. Litig., 249 F.R.D. 196, 214 (E.D.Pa.2008), aff'd, 639 F.3d 623 (3d Cir.2011), abrogated on other grounds by Amgen, Inc. v. Connecticut Ret. Plans & Trust Funds, ___ U.S. ___, 133 S.Ct. 1184, 185 L.Ed.2d 308 (2013).
The expert testimony in this case highlighted the many structural differences between the market for stocks and the market for bonds. Nevertheless, I agree with the court in DVI, supra, 249 F.R.D. at 214, that "corporate bond investors cannot be categorically denied an opportunity to utilize the fraud on the market theory simply because of a structural difference in the way that debt securities are marketed and traded vis-a-vis equity securities."
Accordingly, and in line with the great weight of authority, I will consider the Cammer factors in my analysis. See, e.g., Countrywide, supra, 273 F.R.D. at 615; HealthSouth, supra, 261 F.R.D. at 633-634; DVI, supra, 249 F.R.D. at 214; In re Dynex Capital, Inc. Sec. Litig., 2011 WL 781215, *4-6 (S.D.N.Y.).
The parties agree that the most important Cammer factor is the cause-and-effect relationship between new, unexpected information and bond prices. I therefore begin my analysis with Nettesheim's event study and Dr. Roper's criticisms of it.
At the outset, plaintiffs object that Roper did not disclose, before the hearing, his opinion that Nettesheim studied too few events. Defendants respond that Roper discussed that topic in his first expert report. They also argue that, to the extent plaintiffs object to cross-examining Nettesheim about the number of events studied, no authority restricts the scope of cross-examination of an adverse expert to the opinions in the cross-examining party's expert's report.
Turning to the latter issue first, Fed.R.Evid. 611(b) allows a party conducting cross-examination to inquire on matters touching "the subject matter of the direct examination and matters affecting the witness's credibility." Cross "may embrace any matter germane to direct examination, qualifying or destroying it, or tending to elucidate, modify, explain, contradict or rebut testimony given by the witness." Villanueva v. Leininger, 707 F.2d 1007, 1010 (8th Cir.1983).
The Rules of Evidence "place[] the burden of exploring the facts and assumptions underlying the testimony of an expert witness on opposing counsel during cross-examination." Stecyk v. Bell Helicopter Textron, Inc., 295 F.3d 408, 414 (3d Cir.2002).
In their examination of Nettesheim, plaintiffs inquired: 1) generally about the steps in her event study; 2) whether she based the event study on a commonly accepted methodology; 3) how the bonds reacted to the events she identified; and 4) how the bonds would have reacted if "the market is efficient and no new information is received into the market." (Tr. 56).
These questions put in issue the validity of Nettesheim's event study and the reliability
Moreover, even assuming Roper deviated impermissibly from his report,
Accordingly, I will consider Roper's opinion that Nettesheim studied too few events.
"An event study attempts to determine whether new information correlates with a price movement[.]" Countrywide, supra, 273 F.R.D. at 614.
The events "should be selected using criteria that are as objective as possible," and the criteria "determined before looking at the result to be studied." Id. at 618. However, "many decision points in designing an event study require some subjectivity: identifying news, categorizing which news is `material,' and determining whether news should have a certain ... magnitude of positive or negative influence on price are all subjective determinations." Id.
Plaintiffs argue Nettesheim's event study provides reliable evidence that the Dana bonds fully and rapidly absorbed new, unexpected information. They contend Nettesheim: 1) based her study on a commonly accepted model and performed the study in keeping with how, according to Dr. Roper, an event study should be conducted; 2) identified both relevant events for bondholders; and 3) properly relied on of matrix prices.
Defendants respond that Nettesheim's event study was not a proper event study because she did not examine a sufficient number of events. They also argue that price reactions to the corrective disclosures, which occurred at the end of the class period, are irrelevant to whether the bond market was efficient throughout the class period. Defendants also maintain that matrix prices have no place in an event study. Finally, defendants contend that Roper's autocorrelation findings prove the market for Dana bonds was inefficient.
Nettesheim based her event study on a commonly accepted model described in reliable economic literature. See Handjinicolaou & Kalay, Wealth Redistributions or Changes in Firm Value: An Analysis of Returns to Bondholders and Stockholders Around Dividend Announcements, Journal of Financial Economics (March 1984).
Likewise, her methodology was consistent with Roper's testimony, Tr. 261-264, describing how to conduct an event study. Nettesheim: 1) identified events that were unexpected and "value relevant" to Dana bondholders (Tr. 261); 2) calculated the Dana bonds' premium excess returns while controlling for market forces; and 3) identified statistically-significant returns.
As for the events Nettesheim selected, she testified — and Dr. Roper did not disagree — that the September 15 and October
While Dr. Roper criticized Nettesheim at length for failing to study additional events, he failed to identify a specific event, involving the release of new, unexpected information relevant to the bondholders, that Nettesheim should have included in her event study. Moreover, Roper acknowledged that, to be reliable, an event study need not consider a definite number of events; the number of events "depends on the facts and circumstances" of each case. (Tr. 328).
I am persuaded Nettesheim acted appropriately in identifying and studying only the corrective disclosures. Her methodology flows naturally from her view, which I find credible, that bond prices do not react to the same types of information that drive stock prices.
Indeed, that proposition is firmly entrenched in cases evaluating the efficiency of bond markets. See, e.g., In re Enron Corp. Sec., Derivative, & ERISA Litig., 529 F.Supp.2d 644, 747 (S.D.Tex.2006) ("There seems to be little, if any, dispute that the nature of news that would affect the markets for stock can be quite different [from] what would affect the market for bonds."); Countrywide, supra, 273 F.R.D. at 615 ("Debt and equity respond differently to some types of news ... Debt reacts to the risk of a change in interest rates ... It also reacts to default risk, which is the risk that the issuer will not pay as agreed."); HealthSouth, supra, 261 F.R.D. at 631-632 ("information that affects the price of stocks, such as the announcement of a dividend or earnings statement, would not be expected to affect the price of bonds").
For all these reasons, I reject defendants' arguments that Nettesheim's study is unreliable because she studied only two events.
As discussed, matrix prices are estimated prices "based on credit ratings and other important creditworthy events." Enron, supra, 529 F.Supp.2d at 760. They are "used by financial institutions that invest in bonds ... to accurately price the securities in their portfolios periodically, even daily, for regulators, life insurance companies and others." Id.
Matrix prices can reliably substitute for transactions prices in an event study, provided that the matrix prices "and transaction prices generally move[ ] in the same direction[.]" Teamsters Local 445 Freight Div. Pension Fund v. Bombardier, Inc., 546 F.3d 196, 209 (2d Cir. 2008); see also Dynex, supra, 2011 WL 781215, *6 (same).
The weight of the evidence establishes that the matrix prices here are reliable proxies for transaction prices. First, Nettesheim's event study noted that: 1) on both September 15 and October 10, both price sets dropped; and 2) nearly all price drops were statistically significant. In this regard, I note that the court in Dynex, supra, 2011 WL 781215, *5-6, concluded that drops in the matrix and transaction prices after only two bond ratings downgrades "confirm[ed] the generally uniform
Second, I credit Nettesheim's testimony that examining whether the matrix price was within $.50 of either the day's high or low transaction price provides a reasonable representation of whether both prices move in the same direction.
As Nettesheim explained, on many days when a bond traded "there's more than one transaction" but only one matrix price. (Tr. 375). Accordingly, Nettesheim "look[ed] at the full range of transaction prices to see how close the matrix price is to any of the transaction prices during that day." (Id.). By focusing on the high and low trades, Nettesheim could "determine whether the matrix prices are related in some way to ... the actual transaction price." (Tr. 376). This is not, as Dr. Roper would have it, "cherry-picking" (Tr. 302), but a reasonable way of determining whether the matrix price is related to the prices at which the Dana bonds actually traded.
Third, while the precise inputs that generate matrix prices are unknown, I am comforted by the fact that investment professionals rely on matrix prices in performing similarly high-stakes work: valuing the holdings in their portfolios, whether for their own internal reporting requirements or in preparing regulatory filings. See Enron, supra, 529 F.Supp.2d at 760.
Fourth, Roper criticized matrix prices because his calculations showed that, depending on the bond, the matrix price and transaction price moved in opposite directions between twenty-two to forty-five percent of the class period's trading days. However, a corollary is that the price sets moved in the same direction (or the matrix price moved when there was no change in the transaction price) from fifty-five to seventy-eight percent of the trading days — a substantial number of trading days where the two price sets are closely related.
For these reasons, I find plaintiffs have established that matrix prices are a reliable proxy for transaction prices. See Bombardier, supra, 546 F.3d at 209; Dynex, supra, 2011 WL 781215, *6.
Nettesheim's event study established that the bonds reacted to new, unexpected information. Both she and Dr. Roper agreed that the information disclosed on September 15 and October 10 was "cataclysmic" (Tr. 273, 362) — and thus precisely the type of information that puts bond values in jeopardy. See Countrywide, supra, 273 F.R.D. at 615 ("Until the financial situation becomes severe enough that the issuer is likely to default, there is relatively little effect on debt price.").
Eight Dana bonds traded on September 15; the price of each bond decreased, and five of those price declines — for the AH, AJ, AK, AL, and BA bonds — were statistically significant. After the October 10 disclosure, Nettesheim found that all the bonds experienced statistically significant price drops on either October 10 (the AK, AL, and BA bonds), October 11 (the AH and MAB bonds), October 13 (the AU bond), or October 14 (the AJ and AX bonds).
Defendants do not dispute these findings. Rather, they argue that price reactions to cataclysmic information released at the end of the class period cannot establish market efficiency throughout the class period. Relying on Unger v. Amedisys, 401 F.3d 316, 325 (5th Cir.2005), defendants contend that price reactions to catastrophic news are "insufficiently probative to determine based on empirical facts" whether a market is efficient.
In contrast, defendants have not identified any other event that Nettesheim should have studied — let alone an event that should have correlated, but did not correlate, with a statistically-significant return. Nor have defendants provided a countervailing event study.
Second, Nettesheim opined that, had the corrective disclosures been made earlier in the class period, the bonds would have reacted in the same way. This testimony suggests that the market was functioning efficiently throughout the class period. I credit that testimony, as it comports with plaintiffs' evidence on the remaining Cammer factors (which I discuss below) — all of which shows that the Dana bond market was open, developed, and liquid — and thus likely to incorporate all relevant information about Dana in the price of the company's securities — throughout the class period.
Third, defendants' reliance on the Fifth Circuit's decision in Unger, supra, is misplaced. In that case, the district court held that a company's stock traded in an efficient market based on raw data showing "the stock price rose following positive announcements issued by [the company]... and ... dropped the day the company announced that its earnings would be restated." 401 F.3d at 325. The appellate court vacated the district court's order, finding that, inter alia, the court "did not take into account the many other factors that could affect the [stock price]." Id. at 324.
Here, in contrast, Nettesheim did not rely on raw price movements in identifying a cause-and-effect relationship. Rather, she relied on the bonds' premium excess returns, which represent the portion of the return that is not attributable to market forces. Furthermore, nothing in Unger forbids relying on price returns after catastrophic corporate news when evaluating market efficiency.
Finally, my conclusion that Nettesheim's event study reliably shows a cause-and-effect relationship is consistent with HealthSouth, supra, 261 F.R.D. at 636-637.
Plaintiffs in HealthSouth introduced event studies that "examined numerous days" during the class period. Id. at 636. However, in holding that the bonds traded in an efficient market, the court emphasized that, "on three days when Health-South surprised market participants with adverse news, the price of bonds dropped dramatically." Id. (emphasis added). In the court's view, those studies demonstrated that "bond ratings and bond prices reacted promptly and dramatically to the unexpected disclosures of adverse financial information[.]" Id.
As in HealthSouth, Nettesheim's event study here shows that the Dana bonds reacted to unexpected, adverse information. Having rejected defendants' criticisms of that study, I note that "most of the[ir] challenges stem from the differences between how stocks and bonds respond to different efficiency." Health-South, supra, 261 F.R.D. at 636. However, I credit Nettesheim's opinion that an economist should bear those differences in mind when evaluating a bond market's efficiency.
Autocorrelation "generally refers to the correlation between two observations of the same series at different dates." (Doc. 207 at 6). "A security exhibits autocorrelation if the change in price of the security on a given day provides an indication of what the change in price for the security will be on the following day." DVI, supra, 249 F.R.D. at 213.
The presence of autocorrelation in a security's price may indicate that a security trades in an inefficient market. Id. However, autocorrelation does not, standing alone, prove market inefficiency. See In re Netbank, Inc. Sec. Litig., 259 F.R.D. 656, 675 (N.D.Ga.2009) ("even if Netbank's stock were to demonstrate negative autocorrelation, this would constitute only one factor weighing against a finding of efficiency"); DVI, supra, 249 F.R.D. at 213 (although stock "demonstrated some autocorrelation during the Class Period," court found market was efficient based on "assessment of the various factors [including Cammer factors] analyzed above, for which most weigh in favor of market efficiency").
Roper testified that the Dana bonds exhibited autocorrelation, which, in his view, meant the market for Dana bonds was not even weak-form efficient. However, Nettesheim's explanations of the flaws in Roper's analysis are convincing.
First, Roper's reports do not adequately account for the possibility that the autocorrelation he observed resulted from non-synchronous trading among the Dana bonds. Non-synchronous trading "is well known to financial econometricians to bias statistical estimates in favor of finding a significant relationship with lagged (i.e., historical) returns." (Doc. 197-9 at 23).
To be sure, Roper opined that the risk of spurious autocorrelation findings is minimal, and that a test he ran confirmed that non-synchronous trading could not explain his findings. However, Roper did not describe that test, its methodology, or its results in either of his expert reports. Thus, I cannot be sure — and plaintiffs have had no opportunity to contest — whether he conducted that test properly or that its putative results are valid and credible.
Second, Roper's autocorrelation findings imply that an investor who knew how the Dana bonds traded in the past could accurately predict — and therefore profit from — how the bonds would trade in the future. Nevertheless, Roper did not attempt to learn whether any investors had, in fact, used the autocorrelation he identified to profit on the Dana bonds. See In re Computer Sci. Corp. Sec. Litig., 288 F.R.D. 112, 121 (E.D.Va.2012) (rejecting expert's autocorrelation claim where expert failed "to show[ ] that the serial correlations could have been used to generate trading profits").
In view of the unanswered questions about Roper's autocorrelation findings, I reject his claim "that Dana Bond returns violated the weak-form of the efficient markets hypothesis during the Class Period." (Doc. 197-9 at 25).
"The first four Cammer factors indirectly assess market efficiency by evaluating whether the attributes of the market for that security are conducive to informational efficiency." Countrywide, supra, 273 F.R.D. at 613.
Defendants contend that Cammer factors one through four do not provide reliable economic evidence of market efficiency,
When, as in this case, a plaintiff introduces evidence of high average weekly trading volumes, heavy analyst coverage, or the presence of numerous market makers, a court can properly conclude that the market is open, liquid, and developed. Such a market, moreover, is likely to be one in which securities prices fully and rapidly incorporate new, unexpected information. Accordingly, I reject defendants' argument that Cammer factors one through four are irrelevant to the issue of the market efficiency, or lack thereof, for the Dana bonds.
As the court explained in Countrywide, supra, 273 F.R.D. at 613, "[t]he higher the trading volume, the stronger the showing of significant investor interest in the company and, therefore, likelihood that many investors are executing trades on the basis of newly available or disseminated corporate information."
One court has suggested "this Cammer factor may not always be well suited in evaluating the efficiency of the market for specific corporate bonds," given the lower trading frequency in the bond market. HealthSouth, supra, 261 F.R.D. at 634. In assessing the relevance of this factor, what matters is viewing trading volume "in the context of the bond market[.]" Id. (citing DVI, supra, 249 F.R.D. at 215). As stated in DVI, a low absolute trading volume "would not likely support a finding of an efficient market for equities, [but it may be] sufficient to support such a finding for corporate bonds." 249 F.R.D. at 215.
Here, Nettesheim found that eight of the nine Dana bonds had weekly trading volumes over the class period of at least two percent; the remaining bond's weekly trading volume was one percent of par value outstanding. Under Cammer, supra, 711 F.Supp. at 1285, these figures warrant a "strong" presumption that the AH, AJ, AK, AL, AU, AX, AY, and BA bonds traded in an efficient market, and a "substantial" presumption that the MAB did as well.
Nettesheim also noted all bonds except the MAB bond traded more than fifty-two days per year — or on twenty to twenty-one percent of all trading days in the class period. In contrast, the average bond in 2006 traded on only fifty trading days. Notably, Nettesheim and Roper agreed that trading in the Dana bonds increased substantially during the weeks in which the corrective disclosures occurred.
Defendants argue that these numbers are insignificant, and that a market cannot be efficient unless there is "active trading." See Freeman v. Laventhol & Horwath, 915 F.2d 193, 197 (6th Cir.1990) ("the fraud on the market theory rests on the assumption that the price of an actively traded security in an open, well-developed, and efficient market reflects all the available information about the value of a company"). But this argument reflects defendants' refusal to recognize that the bond market's structural characteristics are relevant to whether that market is efficient. And contrary to defendants' argument, the cases recognize a distinction — and the evidence here
Thus, while the Dana bonds traded less frequently than Dana's stock, the bonds' average weekly trading volumes warrant either a "strong" or "substantial" presumption of efficiency under Cammer. I therefore conclude that the average weekly trading volumes "demonstrate that the market for [Dana] bonds was developed and efficient[.]" Id.
The Cammer factor regarding analysts' coverage of a security or issuer "reflects the view that the greater number of analysts who follow and report on a company, the greater the likelihood that investors rely on information provided about the company." HealthSouth, supra, 261 F.R.D. at 634.
Nettesheim found that at least fifteen analysts covered Dana and issued numerous reports about its securities. Although the majority of these reports covered Dana's stock, the information helped investors evaluate the riskiness of Dana's debt securities. See DVI, supra, 249 F.R.D. at 215 (recognizing that "equity analyst coverage is not a perfect substitute for debt analysis coverage," equity reports "allow bond investors to better understand [a bond's] risk profile and its potential for default").
Moreover, two credit ratings agencies monitored the Dana bonds throughout the class period, while a third agency monitored all but one of the bonds. See Countrywide, 273 F.R.D. at 615 ("[c]overage by the major credit ratings agencies also provides information to the market and is relevant to a determination of a debt security's efficiency").
In light of the substantial analyst coverage, this factor weighs in favor of market efficiency.
"The presence of Market Makers supports the efficiency of a market because they react swiftly to company news and reported financial results by buying or selling [a security] and driving it to a changed price level." HealthSouth, supra, 261 F.R.D. at 635. "The more market-makers for a particular security ... the more reasonable it is to infer that the security is liquid, and, therefore, more likely the market for that security is efficient." Countrywide, 273 F.R.D. at 614.
As noted above, Nettesheim identified at least 200 market makers and the number of times each transacted in the Dana bonds. Although Roper criticized Nettesheim for conflating the concept of a market maker with a market participant, the FINRA data on which Nettesheim relied defines the "market maker side" of each transaction as the participant that executed the trade and has the reporting responsibilities. Thus, Nettesheim properly tallied the number of market makers.
Moreover, notwithstanding any difference between the SEC's definition of a market maker and those entities Nettesheim identified as market makers, the market participants Nettesheim identified facilitated trading in the Dana bonds — a process that contributes to an open, developed, and efficient market.
Accordingly, I conclude that the presence of market makers transacting in the Dana bonds weighs in favor of market efficiency.
SEC regulations permit companies to file a short-form S-3 registration form if they are sufficiently large and have previously filed periodic SEC reports. "Corporations
Nettesheim noted Dana had filed an S-3 form in 1998 and opined Dana remained eligible to file an S-3 during the class period. Roper did not dispute that contention. Accordingly, I find Dana's eligibility to file an S-3 form weighs in favor of market efficiency.
In sum, having analyzed the parties' evidence as to all five Cammer factors, I am satisfied that plaintiffs have proved that the Dana bonds traded in an efficient market. Thus, plaintiffs have satisfied the requirements of Rule 23(b)(3) for certifying the bondholder class because: 1) common questions will predominate over questions affecting only individual class members; and 2) a class action is superior to other ways of adjudicating the dispute — namely, individual suits for the same claims that can fairly and efficiently be resolved in a single proceeding.
To obtain class certification, the representative parties — here, the Plumbers and Pipefitters National Pension Fund and the West Virginia Laborers Pension Trust Fund — must also show their claims are "typical of the claims or defenses of the class." Fed.R.Civ.P. 23(a)(3).
The typicality requirement ensures that "a sufficient relationship exists between the injury to the named plaintiff and the conduct affecting the class, so that the court may properly attribute a collective nature to the action." Beattie v. CenturyTel, Inc., 511 F.3d 554, 561 (6th Cir. 2007). A plaintiffs claim is typical "if it arises from the same event or practice or course of conduct that gives rise to the claims of other class members, and if his or her claims are based on the same legal theory." In re Am. Med. Sys., Inc., 75 F.3d 1069, 1082 (6th Cir.1996).
"[T]here is no requirement that the claims of all plaintiffs and class members must be identical." Plumbers, supra, 292 F.R.D. at 527.
In certifying the stockholder class, I saw no reason "why Dana bond purchasers will present different factual allegations, legal issues, or legal theories of liability[.]" Id. Because I intended to hold a hearing on the efficiency of the bond market, I reserved ruling on the typicality question and invited the parties to present additional arguments in post-hearing briefs.
Plaintiffs argue that, even though they did not purchase Dana bonds, their claims are typical of the bondholder claims because all claims: 1) arise out of the same fraudulent conduct; 2) are based on the same cause of action; and 3) depend on showing that defendants misrepresented Dana's financial health.
Defendants contend that the bondholders' claims present different "market efficiency issues, [and] also relevance and materiality issues" than the stockholders' claims. (Doc. 225 at 13). They also argue that plaintiffs lack standing to assert claims on behalf of bondholders.
I agree with plaintiffs that their claims, as stock purchasers, "arise from the same... course of conduct that gives rise to the claims of" the bondholders — namely, the misrepresentations defendants allegedly made throughout the class period. Am.
To the extent there may be differences in proof on some elements of the bondholders' claims, those differences do not go to the heart of the claims: whether defendants: 1) with scienter; 2) made misrepresentations; 3) in connection with the sale or purchase of a security; 4) on which the class members relied; 5) resulting in economic loss. See Stoneridge, supra, 552 U.S. at 157, 128 S.Ct. 761. Any differences in proof on the materiality element or the market-efficiency question are insufficient to make lead plaintiffs' claims atypical. See Plumbers, supra, 292 F.R.D. at 526-27. I therefore conclude that plaintiffs' claims are typical of the bondholders' claims.
For largely the same reasons, plaintiffs have "class standing" to assert the bondholders' claims. A plaintiff has class standing if he "plausibly alleges (1) that he personally has suffered some actual injury as a result of the putatively illegal conduct of the defendant," and "(2) that such conduct implicates the same set of concerns as the conduct alleged to have caused injury to other members of the class by the same defendants[.]" NECA-IBEW Health & Welfare Fund v. Goldman Sachs & Co., 693 F.3d 145 (2d Cir. 2012).
The Sixth Circuit has already determined that plaintiffs's claims are plausible. See Frank v. Dana Corp., 646 F.3d 954, 958-963 (6th Cir.2011). And the same considerations showing why plaintiffs' claims are typical of the bondholders' claims establish that defendants' conduct "implicates the same set of concerns" for both stockholders and bondholders. NECA, supra, 693 F.3d at 162.
For the foregoing reasons, it is
ORDERED THAT plaintiffs' motion for class certification (Doc. 166) be, and the same hereby is granted in full.
So ordered.
17 C.F.R. § 240.15c3-1(c)(8) (SEC Regulation).
