SENTENCING OPINION RE: CALCULATION OF LOSS
BREYER, District Judge.
Defendant William Grabske (“Mr.Grabske”) was the Chief Executive Officer of Indus International, Inc. (“Indus”), a publicly traded software company. On May 29, 2002, Grabske was convicted of securities fraud and other related offenses in connection with the preparation of false financial statements which overstated Indus’s revenue for the third quarter 1999.
The parties agree that under the United States Sentencing Guidelines (“the Guidelines”), the base offense level is six.
See
U.S.S.G. § 2Fl.l(a) (1998). They also agree that the offense level should be increased by two levels for more than minimal planning,
see id.
§ 2Fl.l(b)(2), and an additional two levels for Grabske’s role in the offense.
See id.
§ 3Bl.l(c). While they agree that the offense level must also be enhanced based on the amount of investor loss caused by Grabske’s fraud,
see id.
§ 2Fl.l(b)(l), they vigorously disagree as to the amount of such loss and even the appropriate method for calculating it. Each party retained an expert to support its proposed method and critique the opposing expert’s method. Nonetheless, the parties ultimately agreed on what the out
come of the sentencing should be, and have stipulated to two alternative methods for reaching that result. Thus, the Court must resolve one critical issue: how to calculate the loss caused by Mr. Grabske’s securities fraud.
BACKGROUND
On October 28, 1999, Indus issued a press release announcing its third quarter 1999 financial results. Indus reported that it had met quarterly targets with revenues of $50.88 million and earnings of $3.52 million, or 10 cents per share on a diluted basis. Analysts estimated that fourth quarter revenue would be $52.3 million with earnings per share again at 10 cents. The next day Indus shares closed at slightly over $6.00. The financial report was false because Indus had improperly recognized over $2 million in revenue from two contracts with rights of cancellation and return. A couple of weeks later, Indus filed its Form 10Q with the Securities and Exchange Commission repeating the false financial information announced in the press release.
Over the next two months the price of Indus shares rose steadily. Then, on January 6, 2000, Indus announced that it would not meet analysts’ expectations for the fourth quarter and instead, that earnings per share would be zero cents on revenue of approximately $43 million. The price of Indus shares fell from $11.69 to $8.94 on the next day of trading.
Three weeks later, on January 27, 2000, Indus announced a fourth quarter 1999 loss of .24 cents per share. Indus also announced that it would be restating its third quarter 1999 financials and that the restatement would result in a drop of earnings from 10 cents per share to one cent per share. The next day Indus shares dropped from $9.69 to $7.63 per share. Four days later, however, shares of Indus rebounded to a price close to the price before the January 27, 2000 announcement. Indus did not make any public statements that would have contributed to the rebound.
On March 20, 2000, Indus issued its revised third quarter 1999 financial statement to report earnings of three cents per share, rather than the one cent per share predicted in late January and the 10 cents per share announced in late October.
THE BATTLE OF THE EXPERTS
One method of calculating damages in civil securities fraud cases is the “out-of-pocket” rule. The out-of-pocket rule fixes recoverable damages as “the difference between the purchase price and the value of the stock at the date of purchase.”
Green v. Occidental Petroleum Corp.,
541 F.2d 1335, 1344 (9th Cir.1976). Both parties’ experts employ an out-of-pocket method and agree that loss should be determined by multiplying the number of shares purchased during the fraud and held beyond the fraud’s disclosure by the price at which the shares would have traded but for the fraud. The parties have stipulated that the number of damaged shares is 2,742,963 but disagree as to the proper means for computing the value of the stock at the time of purchase.
A. The government’s expert (Michael Smith)
Economists often determine the amount of stock price inflation due to fraud through an “event study.” An event study looks to how the price of the stock changed after the fraud was disclosed as evidence of the amount by which it was inflated prior to disclosure. In other words, the out-of-pocket loss is the amount paid for the stock less the amount for which the investor sold the stock after the fraud was disclosed.
See In re Cendant Corp. Litigation,
264 F.3d 201, 242 n. 24 (3d Cir.
2001). Here, the date of disclosure was January 27, 2000; the next day the price of Indus shares fell by more than $2.00.
Here, however, government expert Michael Smith (“Smith”) argues that an event study is not useful because the fraud disclosure was not clean; that is, other negative information about the company was released contemporaneously with the fraud disclosure. Specifically, on January 27, 2000 Indus announced that it would restate its third quarter 1999 financials because of the fraud
and
that Indus had suffered a fourth quarter loss. As investors are arguably more interested in the more recent information — the fourth quarter information — than restated old information, Smith contends that the drop in the price of Indus’s shares the day after the January 27, 2000 disclosure cannot be attributed solely to the third quarter restatement. Smith rejects an event study for this reason.
Smith proposes an alternative method: the earnings response coefficient (“ERC”) method. To calculate the ERC, Smith first identifies past Indus quarterly earnings forecasts and ensuing earnings announcements. Second, he determines if the announcements were clean, that is, whether they contained only earnings information and not other information that might affect the stock price. Third, he calculates the change in Indus stock price associated with each unexpectedly low (or high) earnings announcement. Finally, Smith “regresses” the Indus stock price changes on the unexpected earnings announcements to estimate the ERC. Once the ERC is estimated, it is multiplied by the percent difference between the fi-audulently reported earnings and the earnings as they would have been absent the fraud.
Smith estimates Indus’s ERC as .290; that is, a negative earnings surprise of 100 percent (for example expected earnings of 10 cents per share and actual earnings of zero cents) results in a 29 percent drop in stock price. He concludes that for the third quarter of 1999, Indus’s earnings absent the fraud would have been six cents per share. Since analysts had forecasted third quarter earnings of ten cents per share, this represented a 40 percent surprise. Multiplying the 40 percent earnings surprise by .290 results in a 11 percent, or 70 cents per share, decline in stock price. Seventy cents per share times the stipulated number of affected shares equals a total loss amount of approximately $1.9 million.
B. Defendant’s expert (Allan Kleidon)
Defendant’s expert, Allan Kleidon, purports to use an event study. He, like Smith, notes that the price of Indus shares fell by $2.06 the day after the announcement of the restatement, but that on the same day Indus announced a .24 cent per share loss for fourth quarter 1999. He then posits that an accounting restatement of the previous quarter’s earnings provides no new information to the investing public; the public only cares about how the company is performing now and will perform in the future. He concludes, therefore, that the influence of the third quarter restatement on the $2.06 price drop was minimal.
To be sure, Kleidon does not claim that the fraud had no impact on the price of the stock. Rather, Kleidon concludes that since the third quarter restatement announced on January 27, 2000 reduced third quarter 1999 income per share by nine cents (from ten cents to one cent), the inflation due to the fraud is nine cents per share. He multiplies that figure by the stipulated number of harmed shares (2,742,963) for a total loss of approximately $247,000. He then reduces this number by one third because one third of the third quarter restatement was due to accounting mistakes rather than fraud. Accordingly, defendant submits that the amount of the loss was approximately $164,000.
THE STIPULATION
The government initially submitted a sentencing memorandum arguing that, based on its $1.9 million loss figure, the sentencing range for Mr. Grabske is 41 to 51 months. It urged the Court to sentence Mr. Grabske to 51 months because such a sentence “would take into consideration the severity of the offense and the loss to Indus shareholders.” Government Sentencing Memorandum at 1. Mr. Grabske, on the other hand, argued that the applicable guideline range is 24 to 30 months. After reviewing the parties’ submissions, the Court advised the parties that it was not satisfied with either side’s loss calculation and directed the parties to meet and confer to at least narrow their areas of disagreement.
The parties subsequently submitted a stipulation. While they still do not agree on either the amount of the loss or the method by which the loss should be calculated, they have concluded “that there is no established case law sotting forth the proper method for determining loss in criminal securities fraud cases.” They also now agree on the sentencing range that the Court should apply. Accordingly, the parties have stipulated that the government’s loss calculation of $1.9 million “may be reasonable here (although not necessarily the most reasonable in this case),” such that a total offense level of 22 is appropriate. They have also stipulated, however, that should the Court adopt the government’s loss figure, the Court should depart downward five levels on the ground that the government’s loss figure overstates the seriousness of the offense, for a total offense level of 17. In addition, the parties stipulate that Mr. Grabske’s loss figure of $164,000, like the government’s figure, “may be reasonable here (although not necessarily the most reasonable in this case).” Mr. Grabske’s loss calculation corresponds to a total offense level of 17. The government and Mr. Grabske thus agree that a sentencing range of 24 to 30 months is appropriate. Of course, the Court is not bound by the parties’ stipulation.
See
U.S.S.G. § 6B1.4(d).
DISCUSSION
In determining the amount of loss for the purposes of sentencing, the Court “need only make a reasonable estimate of the loss, given the available information.” U.S.S.G. § 2F1.1, commentary n. 9 (1998). “[T]he sentencing court should take a realistic, economic approach to determine what losses the defendant truly caused or intended to cause, rather than the use of some approach which does not reflect the monetary loss.”
United States v. Allison,
86 F.3d 940, 943 (9th Cir.1996).
A. The government’s loss figure
The Court does not find the government’s estimate of $1.9 million in losses realistic or reasonable. The government’s method is based on several unsupported and unexplained assumptions. For example, the government has not adequately explained how defendant’s fraud
caused
Indus investors to lose $1.9 million.
See
United States v. Yeaman,
194 F.3d 442, 457 (3d Cir.1999) (“Section lB1.3(a)(3) establishes a causation requirement when determining actual loss.”). As one court has explained, if the fraudulent statements “were successfully covered up, they were never disclosed to the market and therefore could not have caused a loss. Alternatively, if the cover-up did not succeed, and the market became aware at some time that the accounting statements were incorrect, then proof of the market’s awareness and proof of loss caused by such awareness is still required.”
In re Ikon Office Solutions, Inc. Sec. Litig.,
131 F.Supp.2d 680, 691 (E.D.Pa.2001). In other words, the government must prove that the fraud inflated the price of the stock
and
that the investors “lost” that inflation after the fraud was revealed. The government’s analysis focuses on the first step— the extent to which the fraud inflated the price of the stock — but does not adequately address the second. Under the government’s theory, the loss to investors would be $1.9 million even if the day after the fraud was disclosed the price of Indus shares went
up.
Such a result does not make sense.
In addition, the government’s analysis assumes that the price of the stock was inflated by the same amount during the entire period prior to the fraud’s disclosure, such that every investor who purchased during that period was affected to the same extent. Smith offers no explanation as to why it is reasonable to assume the stock inflation remained the same during the entire period of the fraud, even after Indus’s January 6, 2000= negative announcement about fourth quarter results.
See In re Clearly Canadian Sec. Litig.,
1999 WL 707737 (N.D.Cal. Sep.3 1999) (noting that the assumption that the stock inflation was constant for the period of the fraud was unsupportable).
The government’s stipulation that the loss figure of $1.9 million overstates the seriousness of Mr. Grabske’s offense, warranting a five level departure, is further evidence that the $1.9 million figure is not a reasonable estimate of loss. The Background to Guideline Section 2B1.1 (the successor to Section 2F1.1) explains the connection between amount of loss and seriousness of the offense:
[t]he Commission has determined that, ordinarily, the sentences of defendants convicted of federal offenses should reflect the nature and magnitude of the loss caused or intended by their crimes. Accordingly, along with other relevant factors under the guidelines,
loss serves as a measure of the seriousness of the offense
and the defendant’s relative culpability and is a principal factor in determining the offense level under this guideline.
U.S.S.G. § 2B1.1, Background (Nov.2001) (emphasis added). If Mr. Grabske’s crime actually caused investors to lose $1.9 million, or anything close to.that amount, an offense level of 22 would not overstate the seriousness of the crime because the Guidelines dictate that the seriousness of the crime is a function of the amount of the loss. At the very least, an offense level of 22 does not overstate the gravity of the offense to such an extent that a five-level departure is warranted. The Guidelines specifically contemplate that causing a $1.9 million loss is so serious as to warrant a 12-level enhancement.
At oral argument the government explained that it believes the $1.9 million overstates the seriousness of Mr. Grabske’s crime because the price of Indus shares “bounced back” shortly after disclosure of the fraud. Here, the fraudulent conduct did not lead to the company’s demise; according to the government, this case is not “Enron.” This argument, how
ever, is tantamount to an admission that Mr. Grabske’s fraud did not cause investors as much of a loss as the government contends. Because the stock price bounced back, investors did not lose as much money as they would have if, as in Enron, the price did not recover so quickly-
Application note 11 does not support the government’s position. That note provides that “[i]n a few instances, the loss ... may overstate the seriousness of the offense. This may occur, for example, where a defendant attempted to negotiate an instrument that was so obviously fraudulent that no one would seriously consider honoring it. In such cases a downward departure may be warranted.” U.S.S.G. § 2F1.1, commentary, n. 11 (2000). The example is one in which the
intended
loss — the face value of the check — overstates the seriousness of the offense because the
actual
loss is zero and there is no realistic ehance that the actual loss will equal the intended loss. By contrast, here the government contends that the
actual
loss to investors was $1.9 million.
Finally, the government has not cited a single case in which the ERC method was adopted as a method for computing damages in a criminal
or
civil securities fraud case. It is inappropriate to adopt a novel method for calculating the securities fraud loss in a criminal sentencing proceeding where, as here, that method results in a dramatically higher sentence than that derived by way of the defendant’s proposed method.
In sum, the government has not proved by a preponderance of the evidence that Mr. Grabske caused investors to lose $1.9 million.
See United States v. Hicks,
217 F.3d 1038, 1049 (9th Cir.2000) (“the government [is] obligat[ed] to prove by a preponderance of the evidence the amount of 'harm that resulted from the acts or omissions’ of the defendant.”).
B. Mr. Grabske’s loss figure
Mr. Grabske’s method is also unreasonable, although the loss figure arrived at by way of that method is closer to a reasonable amount than the government’s figure. Mr. Grabske’s expert concludes that in light of the January 27 disclosure of the fourth quarter loss, the incremental effect of the disclosure of the fraud is minimal or nonexistent. He offers no method, however, for quantifying that incremental effect; instead, he simply asserts that the loss is the difference between the income per share announced in October 1999 (10 cents) and the income per share announced when the fraud was disclosed (1 cent), and that this difference should be decreased by one third to account for the portion of the misstatement that was due to mistake rather than fraud. Nor does he explain how the quarterly income per share measure bears any relationship whatsoever to the price of Indus shares and thus the loss sustained by Indus’s investors.
C. An alternative loss figure
While “out-of-pocket” damages are the traditional measure of loss in securities
fraud cases, another measure is rescissory damages. “The rescissory measure of damages in theory contemplates a return of the injured party to the position he occupied before he was induced by wrongful conduct to enter the transaction.” As the Ninth Circuit explained in
In re Mego Financial Corporation Securities Litigation,
213 F.3d 454 (9th Cir.2000):
“Assuming a sale and purchase of stock, true rescission would involve a return, on the one hand, of the purchase price and, on the other hand, of the stock purchased.” In an open market setting the injured party “returns the stock” by selling it on the open market. The defendant “returns the purchase price” by compensating the injured party for any difference between the price that the injured party paid for the security and its trading price following the disclosure of the fraud. Sometimes the purchase price can be fully recovered on the open market and there is no need for the defendant to compensate for the difference.
To illustrate the application of the rescissory measure of damages in an open market setting assume that X purchased 100 shares of ABC stock at $5 per share during a period following a fraudulent misstatement. Further assume that following the disclosure of the fraud the price of ABC stock was $10 per share. X would need only $500 ($5/share x 100 shares) in return for his shares to recover his cost of the 100 shares. Since X could recover $1000 ($10/share x 100 shares) following the disclosure of the fraud simply by selling his shares on the open market no damages are recoverable from ABC.
However, were the trading price of ABC stock following the disclosure of the fraud $1 per share then X could only recover $100 ($l/share x 100 shares) by selling his shares on the open market. It follows that to be returned to the status he occupied before the fraudulent transaction X must recover damages in the amount of $400 from ABC — the difference between the purchase price X paid for the ABC stock and the trading price of that security following the disclosure of the fraud.
Id.
at 460.
The loss calculation adopted by the district court in
United States v. Bakhit,
218 F.Supp.2d 1232 (C.D.Cal.2002), is a form of the rescissory method. In
Bakhit,
the defendant pled guilty to securities fraud, and, as in this case, the stock did not become worthless after the fraud was disclosed. To calculate the loss for purposes of sentencing, the court first determined the average selling price of the shares during the life of the fraud.
Id.
at 1241. Second, the court determined the average selling price after the fraud was disclosed. For this step he chose the period January 20, 1998 through May 17, 1998, because the next significant drop in stock price occurred on May 18, 1998 in response to an earnings announcement.
Id.
at 1241-42. Third, he multiplied the difference between the average selling price during the fraud and the average selling price after the fraud by the number of harmed shares.
Id.
at 1242.
Here, Mr. Grabske’s expert computes the average price of the stock during the fraud (October 28, 1999 through January 27, 2000) as $9.300. The period for computing the average price after disclosure is the period from January 28, 2000 until the next major announcement — namely, March 20, 2000, when Indus reported final numbers for the third and fourth quarters of 1999. Mr. Grabske’s expert calculates the average price for this period as $9.214. The difference between the two periods is $.086. The parties have stipulated to the number of harmed shares as 2,742,963. The total loss computed under this method thus equals $235,894.81. This amount is
not reduced to reflect the fact that part of the restatement was caused by mistake rather than Mr. Grabske’s fraud because the point of rescission is to restore the victims — Indus’s investors — to the position they were in before they purchased the tainted stock.
For several reasons, the Court finds that the above rescissory method of calculating loss is, based on the information available in this particular case, more realistic and reasonable than the two alternative methods proposed by the parties.
First, the Private Securities Litigation Reform Act (“PSLRA”) mandates such an approach in civil securities fraud cases. The PSLRA’s limitation of damages provides:
Except as provided in paragraph (2), in any private action ... in which the plaintiff seeks to establish damages by reference to the market price of a security, the award of damages to the plaintiff shall not exceed the difference between the purchase ... price paid ... by the plaintiff for the subject security and the mean trading price of that security during the 90-day period beginning on the date on which the information correcting the misstatement or omission that is the basis for the action is disseminated to the market.
15 U.S.C. § 78u-4(e). If the plaintiff sells his shares during the 90-day period, “the plaintiffs damages shall not exceed the difference between the purchase ... price paid ... by the plaintiff for the security and the mean trading price of the security during the period beginning immediately after dissemination of information correcting the misstatement or omission and ending on the date on which the plaintiff sells” the security.
Id.
“Section 78u-4(e) essentially caps a plaintiffs damages to those recoverable under the rescissory measure.”
In re Mego Financial Corp.,
213 F.3d at 461. In
Mego,
the district court denied an objection to the plan of distribution in a settled securities fraud class action. The plan had allocated the net settlement according to the rescissory measure of damages.
Id.
at 460. The court concluded that the district court did not “abuse its discretion in approving a methodology similar to the one that Congress has
mandated
in cases filed after the effective date of the PSLRA.”
Id.
at 461 (emphasis added).
The Court recognizes that section 78u-4(e) limits damages for individual members of a civil class action. The Guidelines, however, provide that the sentencing court’s loss estimate “may be based on the approximate number of victims and an estimate of the average loss to each victim, or on more general factors.” U.S.S.G. § 2F1.1, commentary, n. 8. It is thus appropriate to compute the loss based on the average purchase price during the fraud and the average price during a relevant period after the fraud.
Second, the rescissory method eliminates, or at least reduces, the complexity, uncertainty, and expense inherent in attempting to determine out-of-pocket losses
on a case-by-case basis. “Although it may be preferable for the district court to have the benefit of dueling experts and an extensive tutorial to determine the actual loss with exactitude, that is simply not practical in the vast majority of criminal fraud cases. Most defendants do not have the resources to hire an independent expert and the government has similar finan: cial restraints.”
Bakhit,
218 F.Supp.2d at 1240.
Moreover, one of Congress’s central purposes in adopting the Guidelines was to achieve “reasonable uniformity in sentencing by narrowing the wide disparity in sentences imposed for similar criminal offenses committed by similar offenders.” U.S.S.G. Ch. 1, Pt. A(3) (Policy Statement). This purpose is thwarted if sentencing courts simply choose a loss figure based on which party’s expert is more persuasive. Based on the government’s figure, the guideline range is 41 to 51 months, while under Mr. Grabske’s figure the range is 24 to 30 months. The same crime will thus be punished quite differently depending on which method of loss calculation the court adopts.
Congress mandated the rescissory approach in PSLRA cases so as to avoid precisely this type of disparity.
See Mego Financial Corp.,
213 F.3d at 461 n. 3 (stating that one of the purposes of the PSLRA was “to give statutory guidance to provide certainty in calculating damages in securities fraud cases.”). As the Senate report on the PSLRA observed:
The current method of calculating damages in 1934 Act securities fraud cases is complex, with no statutory guidance to provide certainty. As a result, there are often substantial variations in the damages calculated by the defendants and the plaintiffs. Typically, in an action involving a fraudulent misstatement or omission, the investor’s damages are presumed to be the difference between the price he or she paid for the security and the price of the security on the date the corrective information is disseminated to the market. Between the time a misrepresentation is made and the time the market receives corrected information, however, the price of the security may rise or fall for reasons unrelated to the alleged fraud. According to an analysis provided to the Securities Subcommittee, damages in securities litigation amount to approximately 27.7% of an investor’s market loss. Calculating damages based on the date corrective information is disclosed may substantially overestimate plaintiffs actual damages. The Committee intends to rectify the uncertainty in calculating damages by providing a “bounce back” period, thereby limiting damages to those losses caused by the fraud and not by other market conditions.
Senate Report 104-98 at 20 (1995),
reprinted in
1995 U.S.C.C.A.N. 679, 699 (1995). Since Congress believes the rescissory method is the best means of achieving uniformity and certainty in securities fraud civil cases, it follows that it is at least an appropriate method for criminal cases, where certainty and consistency are particularly important.
Finally, adopting a rescissory method is supported by the Guidelines:
“Loss” means the value of the property taken, damaged, or destroyed. Ordinarily, when property is taken or destroyed the loss is the fair market value of the particular property at issue.
Where the market value is difficult to ascertain or inadequate to measure harm to the victim, the court may measure loss in some other way, such as reasonable replacement cost to the victim.
U.S.S.G. § 2B1.1, commentary, n. 2. (1998) (emphasis added);
see also United States
v. West Coast Aluminum Heat Treating Co.,
265 F.3d 986, 991 (9th Cir.2001) (stating that Guideline section 2F1.1 cross-references the theft guideline (§ 2B1.1) in determining loss). Here, it is difficult to ascertain the amount lost, that is, the amount by which the stock was inflated during the period of the fraud and then lost upon disclosure. Loss should thus be measured in some other way, namely, rescission, a method which places the “victim” in the same position he or she was in prior to the fraud.
Both parties’ experts criticize the
Bakhit
method and, by implication, the rescissory approach to loss calculation. The Court agrees that such an approach is subject to criticism for the reasons proffered by the parties, as are the parties’ suggested approaches. The critical difference between the parties’ alternative methods and the method proposed by the Court, however, is that the Court’s method (1) has been expressly endorsed by Congress in the civil context; (2) does not require the Court to choose between battling experts where they propose equally flawed theories; and (3) provides for more uniformity and certainty in sentencing.
CONCLUSION
As the Court stated above, it agrees with the parties that the rescissory measure of loss is less than perfect; indeed, the Court believes that given the difficulty in calculating loss in fraud-on-the-market criminal ease,s a sentence based on factors other than loss may be more appropriate. For example, there has recently been an increase in the number of prosecutions based on inflated financial statements. In such cases a sentence based on the amount of the overstatement — for example, as a percentage of total revenue — may punish the crime more fairly and with greater certainty. Under such a scheme the defendant’s punishment is tied to his conduct and his intent rather than the fortuity of what happens to the price of the stock before or after the fraud is disclosed. Unless and until the Guidelines are revised, however, the Court must make a reasonable judgment as to the amount of loss caused by the defendant’s conduct. Based on the available information, the Court finds that the loss caused by Mr. Grabske’s fraud is approximately $230,000, resulting in an eight-level adjustment to his offense level.
IT IS SO ORDERED.