In Re Clearly Canadian Securities Litigation, 966 F. Supp. 930 (N.D. Cal. 1997)

*931 Patrick J. Coughlin, William S. Lerach, Henry Rosen, John J. Stoia, Jr., Milberg, Weiss, Bershad, Hynes & Lerach L.L.P., San Diego, CA, Alfred G. Yates, Jr., Law Office of Alfred G. Yates, Jr., Pittsburgh, PA, for Melody J. Harrison, John Alan.

Neil A. Goteiner, Claudia A. Lewis, Farella, Braun & Martel, L.L.P., San Francisco, CA, for Douglas Mason.

Ann G. Daniels, Neil A. Goteiner, Claudia A. Lewis, Farella, Braun & Martel, L.L.P., San Francisco, CA, for Bruce D. Horton, Glen D. Foreman, Stuart Ross, Michael Gustavson.

Claudia A. Lewis, Farella, Braun & Martel, L.L.P., San Francisco, CA, for Boyce Butler, Woodall, Elliot Ewing, Ron Kendrick, Clearly Canadian Beverage Corporation.

Jonathan C. Dickey, Brobeck, Phleger & Harrison, Palo Alto, CA, for Stephen J. Kahn, Robert Dickerson, Spencer Leyton, Alan Henricks, Douglas Antone, Richard Schwartz, Stephen M. Dow, Joseph Howell.

Robin L. Filion, Tod L. Gamlen, Baker & McKenzie, Palo Alto, CA, Charles H. Dick, Baker & McKenzie, San Diego, CA, for Borland International, Inc.

WALKER, District Judge.

Currently pending before the court are motions for preliminary approval of settlement in these securities class actions. Because the two motions raise many of the same issues, the court will address them both in this order. Both proposed settlements suffer similar defects and must be disapproved.

The factual background of the Clearly Canadian ("CLCDF") litigation is set forth in detail in the court's order of February 7, 1995, In re Clearly Canadian Securities Litigation, 875 F. Supp. 1410 (N.D.Cal.1995), and will not be fully recounted here. It is sufficient to note that plaintiffs allege that defendants engaged in a course of conduct intended artificially to inflate the price of CLCDF stock during the period August 2, 1992, to July 19, 1993. The alleged conduct included misleading statements regarding a stock offering in Europe and baseless revenue and sales projections.

The two Borland actions, Kaplan v. Kahn, C 95-2295 VRW ("Borland I") and Crook v. Kahn, C 95-0699 VRW ("Borland II"), concern two separate series of events in 1991-1992 and 1994, respectively. Borland International, Inc. ("BORL") is a computer software concern based in Scott's Valley, California. Over the years, BORL has had a number of major product lines, including Quattro Pro, dBase and Paradox. As with a number of firms in the software industry, BORL's fortunes have waxed and waned. From modest beginnings in 1983, the company grew by 1991 to be the nation's third largest maker of software for personal computers. Mark M. Colodny, Borland's Takeover Program, Fortune, Aug 12, 1991, at 93. The price of BORL stock rose to more than *932 $85 per share by late 1991. Since then, the company's fortunes have deteriorated and its shares have dropped to the $6-$8 range.

The original complaint in Borland I was filed on January 8, 1993. A 160 page First Amended Complaint ("FAC") followed on May 7, 1993. Judge Whyte granted defendants' motion to dismiss the FAC on February 2, 1994, finding that plaintiff's allegations violated FRCP 8(a), requiring a short and plain statement of the claim, or FRCP 9(b), requiring fraud to be plead with particularity, or both.

Plaintiffs filed a Second Amended Complaint ("SAC") on March 11, 1994. The 59 page SAC claims that the company made a number of false statements and material omissions beginning in March 1991 and continuing until December 1992. The principal alleged misstatements and omissions concerned the purchase of Ashton-Tate, Inc, a competing software company, and projected release dates and sales figures for Windows versions of BORL's main software products. Plaintiffs assert that the alleged misstatements artificially inflated BORL's stock price, allowing the company and various insiders to make unwarranted profits on stock sales in 1991 and 1992. Judge Aguilar denied defendants' motion to dismiss the SAC on October 25, 1994.

In striking contrast to the complaints in Borland I, the complaint in Borland II is only 21 pages long. It was filed on February 28, 1995, and alleges that BORL officials misrepresented the release date and projected sales figures for a product called dBase for Windows during the period June 1994 to December 1994. There is no allegation of insider profits from the alleged fraud in the Borland II complaint.

Federal Rule of Civil Procedure 23(e) requires court approval for the settlement of any class action. In order to be approved, a settlement must be "fundamentally fair, adequate and reasonable." Torrisi v. Tucson Elec. Power Co., 8 F.3d 1370, 1375 (9th Cir. 1993) (quoting Class Plaintiffs v. Seattle, 955 F.2d 1268, 1276 (9th Cir. 1992), cert. denied, 506 U.S. 953, 113 S. Ct. 408, 121 L. Ed. 2d 333 (1992)), cert. denied, 512 U.S. 1220, 114 S. Ct. 2707, 129 L. Ed. 2d 834 (1994). Assessing a settlement proposal requires the court to balance a number of factors, potentially including:

the strength of the plaintiffs' case; the risk, expense, complexity, and likely duration of further litigation; the risk of maintaining class action status throughout the trial; the amount offered in settlement; the extent of discovery completed, and the stage of the proceedings; the experience and views of counsel; the presence of a governmental participant; and the reaction of the class members to the proposed settlement.

Torrisi, 8 F.3d at 1375 (quoting Officers for Justice v. Civil Serv. Comm'n of San Francisco, 688 F.2d 615, 625 (9th Cir.1982), cert. denied, 459 U.S. 1217, 103 S. Ct. 1219, 75 L. Ed. 2d 456 (1983)). Other factors may play a role, or even predominate, in certain circumstances. Torrisi, 8 F.3d at 1376.

Settlement of a securities class action premised on a "fraud on the market" theory seemingly sanctioned in Basic v. Levinson, 485 U.S. 224, 241-45, 108 S. Ct. 978, 988-91, 99 L. Ed. 2d 194 (1988), entails particular considerations. Settlement creates a common fund which all class members share in return for extinguishment of their claims; class counsel too must be paid from these funds. A court assessing the fairness of such a settlement must consider:

(1) the total consideration offered by defendants, in light of the potential recovery of the class and the risks of going to trial;

(2) the plan of allocation governing the distribution of the settlement proceeds; and

(3) the amount allocated to attorneys' fees and litigation expenses.

A fairness determination which neglects any of these factors necessarily fails to protect the interests of the class members as against the defendants, other potential class members or class counsel.

*933 The first two factors need to be analyzed separately, but are closely related issues. This follows from the circumstance that, under the fraud on the market theory liability and damages are related. Both the definition of the class and its potential recovery are a function of the impact of the alleged fraud on the price of the security. This impact is itself the product of the extent or degree to which the alleged fraud affected the trading price of the security involved and the period during which the trading price was affected by the fraud. Without establishing both the extent and timing of the price effect neither the class' recovery nor its allocation among class members can be determined. Thus, also, the adequacy of the total consideration offered by defendants and the risks of going to trial cannot be evaluated without analyzing how the consideration will be allocated among class members.

In his concurring opinion in Green v. Occidental Petroleum Corp., 541 F.2d 1335, 1341 (9th Cir.1976), Judge Sneed observed that a fraud on a securities market involves either a false statement or failure to disclose a fact which causes the price of the security to diverge from the value that would prevail in the absence of the fraud. Purchasers of the security who are damaged are those who pay an inflated price. This group includes purchasers who hold the securities acquired at an inflated price until the end of the fraudulently induced inflation ("retention purchasers") and also those who sell the shares acquired while the fraud still affects the trading price of the security but at a time when the degree of fraudulent price inflation is less than at the time of purchase ("in/out purchasers"). Sometimes, a purchaser may have both types of damages. The extent of the inflation and its timing both must be established to identify members of the class and calculate the total amount of damages.

The necessity of determining the extent or degree of price inflation to assess damages to the class should be obvious. It may be less obvious that ascertaining the timing of the inflation, and thus the class period, is equally important to the fairness determination. If the class period is too long, investors unaffected by the fraud will recover a windfall at the expense of injured parties. Likewise, an unduly short class period will result in relative overcompensation of those included in the class and no compensation to those left out.

In order to define the class properly and measure potential recovery, the court must identify the material misrepresentations or misleading omissions that plausibly might have affected the price and trading volume of the security. The court must also pinpoint the times at which the fraudulent misunderstanding in the market occurred and when it was corrected. only with these events fixed can the class period be defined and the reasonableness of a settlement assessed. Short of trial, of course, these facts will not be determined with certainty. In order for the court to evaluate the adequacy of a settlement and thus fulfill its obligations under FRCP 23(e), however, the court must at least be able to make reasonable estimates of them.

In the open and efficient securities market envisioned in Basic, the divergence of a security's actual price from its non-fraud induced value is demonstrated by evidence of a statistically significant change in the price of the security and volume of trading either (1) immediately after a misrepresentation is made, or (2) immediately after the existence of a misleading omission is revealed. The effect of a misleading omission is the more difficult of the two to measure because it can only be deduced from price and trading information at the time the true facts become known. The time at which the market becomes aware of true but fraudulently omitted facts may often be uncertain because corrective information may not enter the market at one time, but comes in bits and pieces. Typically, the effects of a misrepresentation or misleading omission are established through what for most courts are rather abstruse statistical presentations. See, Jon Koslow, Note, Estimating Aggregate Damages in Class-Action Litigation Under Rule 10b-5, 59 Fordham L Rev 811 (1991); Ronald B. Lee, Note, The Measure of Damages Under Section 10(b) and Rule 10b-5, 46 Md L Rev *934 1266 (1987); Jared Toben Finkelstein, Note, Rule 10b-5 Damage Computation: Application of Financial Theory to Determining Net Economic Loss, 51 Fordham L Rev 838 (1983). The accuracy and reliability of these methods has been questioned. See, e.g., Kenneth R. Cone and James E. Laurence, How Accurate are Estimates of Aggregate Damages in Securities Fraud Cases, 49 Bus Lawyer 505 (1994). These methods are, however, the only means by which courts can apply the Basic theory.

The task of the court in gathering and evaluating all of the relevant information is further complicated by the conflicting interests of the class, class counsel and the defendant. At the time of settlement, defendants in these cases have an interest in seeking the longest class period possible in order to maximize the number of claims precluded by res judicata. The plaintiffs' counsel has no particular interest in narrowing the class definition, because a long class period tends to maximize total dollar recovery and thus enhance the claim for attorney fees.^[1]

Furthermore, class members' interests are not congruent with those of their counsel. Individual class members do not benefit by a larger rather than smaller aggregate recovery. Instead, individual class members benefit by showing the maximum price inflation on the date they purchased and, if in/out purchasers, minimum price inflation on the date they sell.

Class members themselves generally have little involvement in the case and have mixed interests. The conflicts among class members can be divided into two groups: (1) conflicts between traders who have suffered actual damage and those who have not; and (2) conflicts among traders who have suffered actual damages. Those class members who have truly been harmed by the fraud are aided by a narrow class period, which reduces administrative costs and maximizes the recovery to those who were actually damaged. Class members without meritorious claims are, naturally, interested in an expansive class definition that will allow them to receive a windfall.

Even among those class members who have actually suffered damages, there are conflicts. First, where either misrepresentations or curative disclosures were allegedly made on multiple dates, plaintiffs suffering retention damages have different interests from those suffering in/out damages. See In re Seagate Technology II Securities Litigation, 843 F. Supp. 1341, 1359-67 (N.D.Cal. 1994). Those conflicts result from different plaintiffs' interests in formulating a value line for the security which maximizes their own recovery. Second, potential class members who currently own the stock have an interest in preserving the assets of the company against the claims of all other class members.

The effect of some of these conflicts on the class certification process has previously been explored. See id. At the time of settlement, the combination of intra-class conflicts and diverging interests of class members and class counsel creates a circumstance in which the court's normal confidence in the truth seeking capacity of the adversarial system would be misplaced. Determining that a proposed settlement treats class members fairly requires, therefore, that the court be shown at least some evidence supporting the factual predicate for the plan of allocation. In the absence of such evidence, the court has no assurance that the plan is not simply a post hoc rationalization of the settlement reached by defendant and class counsel, which unduly expands the class period for the benefit of the defendant and at the expense of those plaintiffs who have genuinely suffered harm.

*935 Requiring that the plan of allocation be supported by evidence sufficient for plaintiff to survive a summary judgment also resolves one of the potential conflicts among potential class members. A settlement that pays damages to individuals whose claims could not survive summary judgment would be unfair to class members who, by virtue of their current ownership of defendant's stock, have an interest in protecting the company's assets. As long as each class member's claim could survive summary judgment, the company has an exposure which justifies paying something to resolve the matter.

Even a plan of allocation which has evidentiary support may, of course, be constructed so as unfairly to shift recovery from individuals with meritorious claims of damages to others. In a complex securities fraud case, there may be dozens of different theories which would survive summary judgment. Unfortunately, the information asymmetries between the court and counsel make it impossible for the court to assure itself that some unwarranted shifting of recovery has not occurred. In the absence of other indicia of reliability, such as participation as class representative by a substantial, knowledgeable and sophisticated investor, demanding that the allocation plan be supported by the more than a scintilla of evidence necessary to survive summary judgment will, however, protect against blatant abuses. Because the presentations made on behalf of the present settlements fail that standard, the court declines to give them its preliminary approval.

The parties have proposed a settlement in the CLCDF litigation that would entail a cash payment of $2.5 million, plus interest. The class recovery will be reduced by attorneys' fees, the cost of administering the fund and possibly some litigation expenses. In contrast, the parties assert that total potential damages are $37,970,518, including $31,192,567 in retention damages and $6,777,951 in in/out trading damages. Exh C to Declaration of Mark S. Kangas re Plan of Allocation ("Kangas Declaration").

The parties have not provided the type of evidentiary support for the plan of allocation that this order indicates is necessary. The absence of such an evidentiary showing would be sufficient reason to deny the motion for preliminary approval. The plan also contains, however, a number of troubling characteristics separate and apart from its lack of evidentiary support. In the interest of maximizing the likelihood of approval of any future settlement proposal in this action, the court will address those substantive concerns as well.

The plan of allocation is premised on the notion that the class can be divided into two periods. The first period ("Period 1") runs from August 3, 1992, through December 22, 1992. The second period ("Period 2") runs from December 23, 1992, through July 19, 1993. During the Period 1, the plan assumes the fraud resulted in a constant twenty-five percent overvaluation of the stock. During Period 2, in contrast, the plan assumes a constant true value of six dollars for the stock.

In reference to Period 1, the Kangas Declaration cites four allegedly fraudulent statements giving rise to plaintiffs' claims. Those statements are:

(1) a July 31, 1992, announcement to the effect that CLCDF would be making a stock offering in Europe during the fall of 1992^[2];

(2) a November 4, 1992, statement attributing a portion of a third-quarter sales increase to inventory accumulated for the summer that would be depleted in the fourth quarter;

(3) a November 16, 1992, article on the Dow Jones Newswire attributing a recent rise in the price of CLCDF stock to company executives "touting the stock"; and

*936 (4) a December 5, 1992, statement by CLCDF's chief operating officer predicting fourth quarter earnings of approximately $30-35 million, when they turned out to be $15 million.

The declaration cites only one additional alleged misrepresentation affecting Period 2. Each of these events will be addressed in turn.

Publicly available data indicate a potentially significant spurt in the volume of CLCDF trading during the period surrounding the July 31 announcement, and an increase in price from 14 on July 28 to 17¾ on August 4. The July 31 announcement may well, therefore, have been material. The longevity of any price effect resulting from that announcement is, however, doubtful. Examination of the performance of CLCDF in the weeks following the announcement indicates that the price drifted back down to 15 by August 25, 1992, and had dropped to 13¾ by September 1, 1992. Of particular note is a steep drop in the stock price on August 31, 1992, which accompanied extremely high trading volume of over two million shares. Even ignoring these events, the latest date on which the July 31 announcement could possibly have affected the price of the stock was September 20, 1992, when CLCDF announced the cancellation of the European stock offering. The trading day immediately following the disclosure of the cancellation, however, saw only modest trading, at 342,000 shares, and the price was down only 3/8 . This trading pattern indicates that the inflationary effect of the July 31 announcement was either very small or had basically been eliminated prior to the September 20 disclosure.

The statements made in November and December concerning future sales and revenues do not seem to have caused a shift in the market from which one could infer materiality. The Kangas Declaration does not specify the time of day when the November 4, 1992, statement was made. Assuming it was made while the market was open, it apparently did not have the buoyant affect suggested by the plan of allocation; the price of the stock closed down 1 1/8 . Alternatively, if the statement was made after the market closed, it appears not to have affected traders; there was no change in the closing price from November 4 to November 5.

It is difficult to see how the November 16, 1992, Dow Jones Newswire story could be construed as a representation by the company, true or false. The parties have not provided a copy of the article for the court to review, thus depriving the court of any opportunity to evaluate any company representations that may have been relayed by the story. From the representations made in the Kangas Declaration, it might be inferred that the newswire had attributed the rise in the stock price from $10 3/8 on November 11, 1992, to 11 1/8 on November 13, 1992, to representations made by company executives. While it is possible that a change of¾ was statistically significant for CLCDF stock in this period, that proposition is not self-evident and Kangas provides no support for it.

The December 5, 1992, Barron's article which reported that CLCDF's chief operating officer had projected earnings of $30-35 million in the fourth quarter was actually followed by a substantial drop in the stock price, from 11 to 9 5/8 . Because this statement allegedly constituted both a misstatement and an omission of adverse information, however, measuring its impact requires an examination of the market reaction to the release of the true information. The Kangas Declaration does not provide the date on which the market learned of CLCDF's actual fourth quarter earnings, so the court cannot use the relevant price and volume information to assess materiality.

The foregoing discussion makes clear that the plan of allocation regarding Period 1 is patently inadequate. Even if the dubious allegations regarding November and December earnings predictions were taken seriously, the plan utterly fails to explain how purchasers during the period September 20, 1992, to November 4, 1992, were affected by any alleged fraud. The resulting windfall to these individuals at the expense of legitimate claimants is, standing alone, a sufficient basis to reject the proposed settlement as unfair. The absence of any plausible theory supporting the inclusion of purchasers from November *937 4 to the end of Period 1 buttresses that conclusion by creating the possibility that the class recovery will be further diluted by including purchasers without a genuine damage claim.

The Kangas Declaration appears to assume that the misstatements from Period 1 continued to affect the stock price in Period 2. The only other alleged misrepresentation affecting Period 2 was a statement made by CLCDF's chief operating officer concerning new distribution agreements and aggressive expansion plans for early 1993. The stock price rose by 1 on the day this statement was reported by the Dow Jones Newswire. Again, an increase of this size may have been statistically significant, but the Kangas Declaration does not demonstrate that it was.

A far more remarkable aspect of the plan of allocation for Period 2 is that it assumes that the true value of CLCDF stock was a constant $6 for the entire period. While, as an abstract matter, a constant value for a security over six months is possible, it must be a rare occurrence and no evidence for it has been provided in this case. During Period 2, the trading price of the stock ranged from more than 10 to less than 7. The supposed inflation, therefore, ranged from more than 4 to less than 1 without any allegation of further misrepresentations or corrective disclosures.

The Kangas Declaration premises this rather remarkable proposition on the fact that CLCDF stock traded at roughly 6 during the period following full disclosure. This approach is flawed on two grounds. First, no account is taken of changes in the market-place from December 1992 to July 1993. For the bulk of Period 2, the stock traded between 8 and 10. If the value line is constant at six, what explains this variation. Applying the existing plan of allocation, a purchaser paying 10 would receive twice as much as a purchaser paying 8, with no explanation for the distinction. Furthermore, the claim in the Kangas Declaration that CLCDF traded at roughly 6 for several months after July 20, 1993, is not supported by the facts. Within a month after the full disclosure, the stock closed as low as 4½, 25 percent below the 6 dollar true value claimed by Kangas. A price change of 25 percent is significant in any event, but appears positively apocalyptic compared with the unswerving consistency the plan of allocation assumes for Period 2.

The proposed settlement of the two BORL cases calls for payment of $11.5 million, $10 million for Borland I and $1.5 million for Borland II. Of this amount, BORL's insurance carrier is to contribute $8.5 million. The various firms representing plaintiffs have requested attorneys' fees of 25 percent of the settlement plus expenses in the amount of $1,169,789.49. Thus, the plaintiff class will receive less than $7.5 million and the lawyers will collect close to $4.5 million. The court is unable to assess the reasonableness of the total recovery amount because the parties have not provided an estimate of the total damages to the class.

As in the case of the CLCDF proposal, the parties in these actions have not provided evidence to support the plan of allocation. Again, the absence of such evidence is sufficient reason to deny the motion for preliminary approval of settlement, but the court will address other problematic aspects of the settlement in the hope that future proposals will be adequate.

The plan of allocation in the BORL cases is outlined in the supplemental declaration of John B. Torkelson ("Torkelson Declaration"). As is appropriate, the plan addresses Borland I and Borland II separately. The plan for Borland I hypothesizes two "misrepresentations" resulting in an artificially high stock price. The first of these allegedly inflationary events occurred on March 5, 1991. On that date, BORL publicly exhibited its new Windows based products and issued a press release which stated:

`Borland is aggressively developing a full line of Windows products,' said Philippe Kahn, chairman, president and chief executive officer of Borland. `Our experience in object-oriented programming (OOP) enables *938 us to develop the highest quality products, and we develop from the ground up, rather than porting, to take full advantage of advanced Windows features.'

The price of BORL stock increased four dollars, or 7.92 percent, on the day of the demonstration and announcement. The Torkelson Declaration asserts that the price would not have increased that much had "full disclosure of the true state of [BORL's] new Windows-based software products" been made.

An examination of the Torkelson Declaration leaves the reader uncertain whether the significant event on March 5 was a misstatement or an omission. The declaration places great emphasis on the company's press release and measures inflation after March 5 as equal to the price increase on that day. These factors suggest a misstatement, but Torkelson also asserts that the omission of "full disclosure" influenced traders' behavior. In the absence of any clear characterization of the March 5 event, the court will assume that it should be viewed as a misstatement occurring simultaneously with an omission.

The March 5 press release contains three assertions of fact:

(1) "Borland is aggressively developing a full line of Windows products";

(2) BORL's "experience in object-oriented programming (OOP) enables [it] to develop the highest quality products;" and

(3) BORL "develop[s software] from the ground up, rather than porting, to take full advantage of advanced Windows features."

Neither the SAC or the Torkelson Declaration identifies which of these statements was allegedly false. The curative "disclosures" identified in the plan consist of (1) analysts downgrading the stock due to concerns over sales of BORL's Windows based products; and (2) BORL's announcement of significant layoffs and a large restructuring charge against earnings. Neither of these events appears to involve a revelation that any statement in the press release was false.

Even if it is assumed that a misrepresentation was made on March 5, 1991, its materiality is not clear. The price of BORL stock rose 7.92 percent on that date on trading volume of 471,600 shares. The Torkelson Declaration does not analyze either of these figures to determine whether they vary from the stock's normal behavior in a statistically significant way. One indication that the change in price may have been a response to broader market conditions, rather than the press release, is that the thirteen other software stocks Torkelson examined in connection with this case rose an average of 7.21 percent on that date. Admittedly, the sample is very small and there was substantial variety in the performance of the companies surveyed, but the trend was strongly up. An analysis which showed a statistically significant increase in BORL's stock price after adjusting for the movement of the market and the stock's normal performance relative to the market (quantified by a coefficient commonly known as the security's "beta") would be a far more meaningful indicator of materiality. See Seagate II, 843 F. Supp. at 1348.

To the extent the March 5 event might be perceived as an omission, it reveals several weaknesses in the allocation plan. First, the plan does not disclose what information was omitted. The Torkelson Declaration references the "true state of [BORL's] new Windows-based software products," but does not reveal what the state of those products was on March 5. The absence of this information makes it difficult to assess (1) whether the omitted information is material; and (2) how much inflation can be attributed to the omission. The plan of allocation attributes four dollars worth of inflation to the March 5 event based on the four dollar rise in BORL stock on that date. While a reasonable means of estimating the inflationary effect of a misrepresentation, this method makes little sense in the context of an omission.

The failure of the Torkelson Declaration plausibly to explain the significance attached to March 5, 1991, in this settlement cannot be overlooked by the court because the next alleged inflationary event identified in the plan does not occur until October 30, 1991. Thus, almost eight months of purchasers will be included in the settlement class based *939 solely on allegations relating to March 5. Such an expansion of the class would significantly dilute the recovery of other potential claimants, without a plausible justification.

The second alleged misrepresentation, on October 30, 1991, occurred at a conference for securities analysts. Plaintiffs have alleged that BORL falsely represented to the analysts that Quattro Pro for Windows, Paradox for Windows and dBase for Windows would ship during the March 31, 1992 quarter. The following day, BORL's stock price rose $9¾, or 16.92 percent, on trading volume of almost three million shares.

Unlike the March 5 event, the statements made at October 30 conference can plausibly be viewed as material representations of fact. Again, Torkelson did not determine the statistical significance of either the trading volume or the price increase, but there is evidence to suggest that such an analysis would demonstrate their significance. Trading volume for BORL stock met or surpassed the October 31, 1991, level on less than two percent of the trading days surveyed in Exhibit B to the Torkelson Declaration. In contrast to March 5, the price increase on October 31 differed markedly from the performance of the fourteen other software stocks measured by Torkelson. BORL rose 16.92 percent while the other companies gained an average of just 0.82 percent. Although these measures are less than ideal, they suggest that material information about BORL did enter the market on October 30, 1991.

The plan of allocation in Borland I identifies three dates on which disclosures correcting the October 30 misrepresentation were supposedly made: October 7, 1992, October 8, 1992, and December 9, 1992. The first remarkable thing about these "disclosures" is that the first of them occurs nearly a year after the last identified misrepresentation. The plan of allocation, therefore, assumes that the true information was withheld from the market for nearly twelve months. This feat of corporate secrecy would be a noteworthy achievement in any case, but is particularly amazing given the nature of the alleged misrepresentation.

The only false statement Torkelson specifically identifies from the October 30, 1991, conference was that the Windows versions of BORL's main products would be released in the March 31, 1992, quarter. If this statement was false, its inaccuracy would necessarily be known to the market no later than March 31, 1992. Even recognizing that software releases are often delayed and the market may have taken the March 31 date as an estimate, each day thereafter added more corrective information to the market. This hypothesis is supported by the gradual decline in BORL's stock price from 53¼ on April 1, 1992, to the 38-42 range, where it traded for most of the period from June 25 until the October 7 analyst downgrade.

Even if the court were to accept the October 7-8 analyst downgrade as a corrective disclosure, the December 8, 1992, announcement could not be viewed as one. Arguably, the 6.52 percent decline in BORL stock price on December 9 was statistically significant, although Torkelson does not demonstrate that fact. By December 8, however, the stock was trading at only $23 per share. The notion that any trader was continuing to rely on positive information provided at the October 30, 1991, analyst's conference is simply not plausible. Moreover, the December 8 announcement of layoffs and a multimillion dollar restructuring charge does appear to correct any misstatement identified in the plan of allocation.

The seemingly unjustifiable extension of the proposed class period into the second half of 1992 has the same potential to dilute the recovery of genuinely defrauded traders as the unduly early start of the period. As noted above, the court cannot conclude that a settlement which unjustly enriches some potential class members at the expense of those who have actually suffered harm is "fundamentally fair, adequate and reasonable."

The Torkelson Declaration also provides an explanation of the plan of allocation for Borland II. In Borland II, the alleged fraudulent misrepresentations appeared in an article in PC Week magazine on June 6, 1994. The article stated that dBase for Windows *940 was "slated to ship" in June or July 1994 and reported that BORL's chief executive officer, Philippe Kahn had predicted a return to profitability in the quarter ending September 30, 1994. The declaration identifies two alleged corrective "disclosures," one on September 19, 1994, and the other on October 20, 1994. On September 19, a Dean Witter analyst "expressed his concerns" about sales of dBase and Paradox for Windows. After the close of trading on October 19, BORL announced its earnings for the quarter ending September 30 and a Dow Jones news report the next day speculated that the company was preparing analysts for the possibility of another loss in the following quarter.

Regarding the alleged misstatement, the most significant concern is the possibility that there is no evidence to support a finding of scienter on the part of Kahn. If he believed that the products were going to ship in the following two months and profitability would return in the coming quarter, there is no basis for misstatement liability. Normally, concerns that a defendant might not be liable would not concern the court in approving a proposed settlement; it is assumed that the defendant would not be settling if it thought it could win a summary judgment motion. In this case, however, there are two reasons to be concerned.

The first reason one might question the bona fides of the Borland II settlement is the possibility is that the real exposure is in Borland II, but the company wants the settlement structured so claims are precluded over the largest possible time period. In that case, the settlement would be unfair to Borland I claimants who should be getting the money now going to Borland II class members. The second concern is that the real exposure in Borland II is for the failure of the company to correct the June 6 statement when it became aware that the statement was inaccurate. Even if Kahn believed his statements to be true when he made them, there may be liability for a material omission in failing to make a corrective announcement when the true facts became apparent. In that case, the beginning date of the Borland II class period would be later than June 6, and the current proposal is shifting recovery to individuals who have no legitimate claim.

The disclosure events in the plan of allocation also present certain difficulties. As to both events, the Torkelson Declaration fails to provide the analysis of statistical significance necessary to determine whether the events indicate materiality. This concern is particularly relevant in relation to September 19, when the change in price was smaller and the volume of trading was much lower than October 20.

The September 19 "disclosure" event also seems problematic because of its questionable relationship to the alleged misstatement. The Torkelson Declaration does not make clear how a Dean Witter analyst's expression of concern regarding the "weak sell-through" of two of BORL's products constituted a disclosure of the true facts regarding either the release dates of those products or the profitability of the company in the quarter ending September 30.

While the September 19 analyst statement does not directly address the claims of the June 6 statement, one might argue that weak sales of the Windows products were an indicator that the company would not turn a profit for the quarter. Accepting that suggestion makes September 19 appear more relevant, but calls into question the significance assigned to the October 19 earnings announcement. If the market had access to information as early as September from which low quarterly earnings might be inferred, it seems reasonable that such information may have cured the inflation prior to the earnings announcement. During the period August 30 to October 4, the stock gradually declined from 13½ to 10¾. The more dramatic drop in price on October 20 could be explained by the fact that the Dow Jones news story suggested that company officials were pessimistic about the following quarter. The market may well have already absorbed the bad news for the September 30 quarter, but the projections for the following quarter were new information.

For the foregoing reasons, the motions for preliminary approval of settlement in these *941 two actions are DENIED. Further status conferences in both cases are scheduled for June 20, 1997, at 2:00 pm. If the parties' calendars cannot accommodate this date, they are directed to consult with opposing counsel in their respective matters regarding mutually agreeable alternatives and then contact the clerk to reschedule.

IT IS SO ORDERED.

[1] Typically, securities class actions recover but pennies for every dollar of estimated class damages. Janet Alexander Cooper, Rethinking Damages in Securities Class Actions, 48 Stan L Rev 1487, 1500 (1996) (settlements average twelve percent of claimed losses); Richard M. Phillips and Gilbert C. Miller, The Private Securities Litigation Reform Act of 1995: Rebalancing Litigation Risks and Rewards for Class Action Plaintiffs, Defendants and Lawyers, 51 Bus Law 1009, 1029 (1996) (recoveries in range of seven to eleven percent of claimed damages). This has frequently sparked criticism of securities class actions. See, e.g. Alexander, Rethinking Damages, 48 Stan L Rev at 1500; Phillips and Miller, Private Securities Litigation Reform Act, 51 Bus Law at 1029. The phenomenon may not result so much from recoveries which are small relative to actual class damages as from over-large estimates of class damages.

[2] The Kangas Declaration actually discusses only an August 3, 1992, Wall Street Journal article reporting the announcement, but the text of the article makes clear that the company made the public announcement on July 31.