Summary
noting that defendant in "pump-and-dump" scheme is liable for full amount of investor's loss when "pumped" stock is for company with no assets or earnings
Summary of this case from United States v. SchwambornOpinion
03-CV-833 (NGG)
04-20-2012
UNITED STATES OF AMERICA, v. MYRON L. GUSHLAK, Defendant.
MEMORANDUM & ORDER
NICHOLAS G. GARAUFIS, United States District Judge.
Pursuant to the Mandatory Victim Restitution Act ("MVRA"), 18 U.S.C. § 3663A, the court must determine what amount of money, if any, Defendant Myron Gushlak must pay in restitution to the victims of his securities fraud conspiracy. For the following reasons, the court ORDERS Gushlak to pay $17,492,817.45 in restitution.
On July 22, 2003, Gushlak pled guilty to, inter alia, one count of conspiracy to commit securities fraud. (See Minute Entry for Plea Agreement Hearing (Docket Entry # 5); Information (Docket Entry # 3).) Gushlak admitted to owning, with his coconspirators, over five percent of the stock in a publicly traded telecommunications company called Global Net. (Plea Tr. (un-docketed) at 26.) Gushlak admitted that he paid kickbacks to certain brokers in exchange for the brokers' promise to aggressively push Global Net's stock on customers so that the price of the stock would rise. (Id.) These kickbacks, or "commissions," were not disclosed to the investing public (id. at 27) and the price of Global Net's stock did rise as the brokers promoted the stock (see id. at 27-28). Gushlak then, over three days in the beginning of March 2000, sold around 1.1 million shares from his personal holdings in Global Net at increasingly higher prices. (Id. at 28.) According to Gushlak, the entire scheme lasted about two years—from January 1999 until December 2000. (Id. at 27).
Gushlak waived his right to prosecution by indictment and allowed the Government to proceed against him solely by information. (See Waiver of Indictment (Docket Entry # 4).)
On November 18, 2010, the court sentenced Gushlak principally to seventy-two months in prison and a fine of twenty-five million dollars.
In addition to imposing this sentence, the court was required to order Gushlak to pay restitution to the victims of his fraud. It did not immediately issue such an order, however, because at the time of Gushlak's initial sentencing hearing the proper amount of restitution was still unknown. Cf. 18 U.S.C. § 3664(d)(5) (allowing the court 90 days from the date of sentencing to determine victims' losses if the such losses "are not ascertainable by the date that is 10 days prior to sentencing). Instead, the court directed the Government to submit evidence of the victims' loss and a supporting memorandum of law by December 20, 2010. (See Sentencing Tr. (Docket Entry # 32) at 116.) The issue was to be fully briefed by January 17, 2011. (Id.)
Gushlak pled guilty to an "offense against property" for the purposes of the MVRA, see United States v. Reifler, 446 F.3d 65, 121 (2d Cir. 2006) (citing 18 U.S.C. § 3663A(c)(1)(A)(ii)), and therefore, because the crime harmed "identifiable victims," see 18 U.S.C. § 3663A(c)(B), restitution is generally mandatory, see id. § 3663A(a). The court is empowered to deny restitution, however, if it finds that "the number of identifiable victims is so large as to make restitution impracticable" or if the issues involved are so complex that "the need to provide restitution to any victim is outweighed by the burden on the sentencing process." Id. § 3663A(c)(3). Neither of those factors is present in this case.
For the purposes of restitution, courts in this Circuit estimate the amount of financial loss suffered by victims of a criminal securities fraud in essentially the same way that they calculate damages in a civil securities fraud case. See United States v. Rutkoske, 506 F.3d 170, 179, 180 (2d Cir. 2007). That is to say, they require evidence of "loss causation"—proof that at least part of the decline in a given security's price was caused by the disclosure or cessation of fraud. See generally id. at 178-180. The idea behind loss causation is that demand for a security—which of course drives its price—is usually a function of more than just the fraud. The demand also reflects information about the legitimate value of the underlying company. Thus, when a stock's price falls after a securities fraud is revealed or terminates, it is necessary to try to control for other factors that might have contributed to the decline.
In civil litigation, loss causation is of course more than only a measure of damages; it is itself an element of a claim under Rule 10b-5. See Dura Pharmaceuticals v. Broudo, 544 U.S. 336, 341-42 (2005).
A very simple hypothetical involving the same fraud on two different stocks illustrates the point. Imagine two companies. One company, Company A, has no assets and no expected earnings, now or at any point in the future. Its non-fraud stock price is $0. The other company, Company B, has productive assets and is a going concern. The net present value of all of its future earnings per share is $10, and so its non-fraud stock price is $10. Now, imagine a fraudster illegally "pumps" the stocks of both companies in a manner that causes the price of each to rise by $5; so that Company A's stock trades for $5 and Company B's stock trades for $15. An investor buys both stocks at fraud-inflated prices and holds them continuously. Eventually, the fraudster "dumps" all of his own holdings in the companies and stops pumping. On the same day, however, a fire destroys all of Company B's assets and Company B announces that it will not rebuild. The prices of both stocks fall to $0. While the fraudster should clearly be made to reimburse the investor for the full amount of his loss on Company A's stock, $5, it would be unfair to force him to pay for the investor's entire loss on Company B's stock—two-thirds of which was caused by the fire.
Imagine that the price effects of the fraud and its cessation are instantaneous.
In reality, the process of disentangling a stock's underlying value from its fraud-inflated price is not as tidy. Indeed, any attempt to do so will be inherently speculative because it is impossible to know or weigh the myriad factors investors consider—both consciously and unconsciously—when deciding to buy or sell a security.
Changes in stock prices are attributable to two broad categories of factors, or "risks": systemic risk, which is driven by factors that ought to affect all companies in the overall financial market ("Market"); and idiosyncratic—or "unique"—risk, which is driven by factors that affect less than the full Market. See generally Brealey, Myers, Allen, Principles of Corporate Finance 188 (9th ed. 2008). Idiosyncratic risk may be thought of as itself having two components: industry-specific idiosyncratic risk; and firm-specific idiosyncratic risk. In the example above, companies A and B were affected by purely firm-specific idiosyncratic risk—fraud and fire. While the hypothetical is probably an accurate representation of the factors affecting Company A's stock price, of which fraud would dominate because Company A was not actually engaged in any business; it is a great over-simplification of the factors affecting Company B's stock price, which would be affected by factors as diverse as the global price of oil (systemic risk), demand for its products (industry-specific idiosyncratic risk), the fire, and the fraud (both types of firm-specific idiosyncratic risk).
Over the long run, it may be possible to learn on average how much the movement of a specific stock price is related to changes in the Market—a process that involves, inter alia, determining the stock's "beta," see infra—and then to make generalized predictions and statements. But there is no model or counterfactual that can perfectly break down a stock's historical price into market and idiosyncratic components. What is more, even if such a feat were possible, it would be impossible to know exactly which idiosyncratic factors caused the stock to move more or less than would be predicted given the Market—How much movement can be attributed to the fraud? How much to the fire?
This uncertainty is why courts are allowed to make a "reasonable estimate" of investor loss based on the information available when calculating restitution in a stock fraud case. See United States v. Germonsen, 139 F.3d 120, 129, 130 (2d Cir. 1998). The same is true with respect to calculating loss for the purposes of determining a sentencing range, see Rutkoske, 506 F.3d at 178 (quoting U.S.S.G. § 2F1.1 cmt. n.9), and for damages calculations in civil cases, cf. Boyce v. Soundview Tech. Group, Inc., 464 F.3d 376, 387 (2d Cir. 2004) (stating that, in a breach of contract case, valuation of a security on a date when there was no accurate market price was "necessarily an approximation" (quoting Silverman v. Comm'r, 38 F.2d 927 (2d Cir. 1976))).
Thus, in proving victim loss for purposes of an order of restitution, the Government does not need to show precisely how much money investors lost because of the fraud. Instead it only needs to: (1) prove—by a preponderance of the evidence, see 18 U.S.C. § 3664(e)—that at least some of the investor victims' economic losses were caused by the fraud; and (2) demonstrate—again, by preponderance of the evidence, see id.—a reasonable estimate of the amount of the loss attributable to the fraud.
In past sentencing submissions, the Government failed to meet this standard. This failure was mostly because it did not control for other factors that could have theoretically explained all of change in Global Net's stock price during and immediately following Gushlak's conspiracy.
In its first attempt to prove victim loss, the Government requested an order of restitution in the amount of $20,468,876.29. (Restitution Letter (Docket Entry # 21) at 1.) It based this number on the amount of victim loss calculated by the United States Department of Probation in Gushlak's pre-sentence report, which, in turn, was based on the victims' trading records. (Id.)
As the Government did not provide these records or explain how it or Probation derived loss from the records, the court ordered the Government to provide, inter alia, copies of the relevant victim trading records and an explanation of the methodology used to calculate loss from such records. (See Order Regarding Restitution (Docket Entry # 26) at 4-5.)
The Government responded with copies of the trading records, and explained how it arrived at the twenty-million-dollar-plus loss calculation. (See Reply as to Restitution Letter (Docket Entry # 29).) The method was straightforward: simply calculate each victim's aggregate trading loss in Global Net stock. (Id. at 3.) That is, the Government subtracted any proceeds a victim may have earned from eventually selling the stock from the price that the victim paid to buy the stock in the first place. (Id.) For those victims that continued to hold Global Net stock past November 1, 2000 (apparently the last day for which trading records were available), the Government appeared to accord them zero dollars in proceeds, equating their loss to the full purchase price.
Obviously, unless Global Net's stock was utterly worthless but for the fraud, such an approach could not be squared with the Second Circuit's requirement that the Government prove loss causation. It assumed that Global Net was like Company A in the hypothetical above; that the price of its stock was entirely determined by a single firm-specific idiosyncratic factor—the fraud. Because the Government had not proved by preponderance of the evidence that Global Net was worthless, the court ordered the Government to submit further evidence that would either: (1) show that Global Net's stock was in fact worthless but for the fraud; or (2) somehow account for non-fraud-related factors that could have contributed to the decline in Global Net's stock price. (See 1st Mem & Order (Docket Entry # 35) at 12-14.)
The Government chose the latter option. To account for non-fraud-related factors, the Government attempted to control for systemic risk. It calculated the Market's rate of return during the relevant time period—negative twenty percent—and discounted the victims' aggregate trading losses accordingly. (2d Restitution Letter (Docket Entry # 42) at 2-3.) The story the Government was trying tell was this: Between March 1, 2000 (the first day reflected in the Government's trading records) and November 1, 2000, Global Net's stock price declined seventy-two percent. During the same period, the Market declined by twenty percent. Therefore, the argument went, twenty percent the victims' trading loss on Global Net stock was attributable to market factors, and the residual would have to be attributable to the fraud. (See id.) This revised methodology had a major impact on the Government's total loss calculation, which fell to $8,950,032.05. (Id. at 3.)
It arrived at the Market's rate of return by averaging the respective rates of return of two major equity markets, the NASDAQ and the S&P 500, and a third so-called "small-cap" market, the Russell 2000. (2d Restitution Letter (Docket Entry # 42) at 2-3.)
The Government's revised loss calculation was also a product of having abandoned the assumption that Global Net's stock was worthless on November 2, 2000. To calculate the trading loss for those victims that did not sell their stock during the relevant trading period, the Government used the stock's November 1 closing price of $7.34 instead of zero. (Id. at 2.) The problem with this approach was that Gushlak admitted in his plea allocution that the fraud lasted at least through December 2000, giving rise to the possibility that the November 1 closing price was still tainted by fraud.
The Government's methodology was flawed in two major ways.
First, the Government assumed, without any evidence, that Global Net's "beta" was 1.0. Beta describes an individual stock's sensitivity to changes in the Market. See Brealey, Myers, Allen, supra at 193. A beta of 1.0 shows that a stock's price will move in tandem with the market—as with the Government's assumption, that a twenty percent decline in the Market will be accompanied by a twenty percent decline in the price of the stock. But some stocks are more or less sensitive to changes in the Market, and, theoretically, some might be inversely related to it. Thus, the price of a stock with a beta of, for example, 0.5, will only move half as much as the Market moves; and the price of a stock with a beta of -1.0 will move in the same amount as the Market, but in the opposite direction. Without attempting to establish Global Net's beta, the Government could not produce a reasonable estimate of the victims' loss because there was no reason to believe that the twenty-percent of the stock price's decline was attributable to systemic factors. Supposing—unlikely though it might be—that Global Net's beta was 3.6, the full seventy-two percent decline in Global Net's stock price could be attributed to the twenty-percent decline in the Market.
Second, even if it the Government was correct to assume that the company's beta was 1.0, the Government's methodology still relied on the assumption that the residual change in Global Net's stock price—the remaining fifty-two percent—was caused entirely by the fraud. That is, that Global Net's idiosyncratic risk was wholly firm-specific and, at that, driven by a single factor—the fraud. This is not a plausible assumption given the differences that exist between the companies that make up much of the Market and a small telecommunications firm like Global Net.
Thus, because Government's revised methodology might greatly over or underestimate the effect of Gushlak's fraud on Global Net's stock price, the court concluded that the revised loss figure of around nine million dollars was not a reasonable estimate of the actual loss. (See 2d Mem. & Order (Docket Entry # 58) at 9.) Offering the Government one last chance to meet its burden, the court suggested that the Government produce more evidence concerning the specifics of Gushlak's fraud, and that it attempt again to link Global Net's rate of return to the Market. (See id. at 14-15.)
On October 24, 2011, the Government responded with the instant submission. Its memorandum contains supporting affidavits that delved into greater detail about the workings of Gushlak's fraud than ever before (Sentencing Mem. (Docket Entry # 70)) and the submission includes a report by a new expert witness (DeRosa Expert Report (Docket Entry # 71)) that uses much different and more sophisticated methods to estimate the victims' total loss. The evidence—sworn affidavits from Gushlak's coconspirators and a federal agent—tells a convincing story of how Gushlak pumped up the price of Global Net stock and then let the price collapse after he had sold most of his personal holdings. If credited, it shows that cessation of the fraud rather than revelation of the fraud caused the price of the stock to decline. The expert report is even more persuasive. It shows through relatively simple econometric techniques that, during the period of the fraud, the rate of return on Global Net stock was not correlated with the rates of return of competing companies or that of a specialized telecommunications market index, but that after the fraud, Global Net's rate of return was statistically significantly related to the market index and several of its competitors. This gives rise to a strong inference that the fraud had a major effect on Global Net's stock price—that at least some of the victims' trading losses were caused by the fraud. The expert report then provides a sound method for estimating exactly how much of the loss can be attributed to firm-specific idiosyncratic factors, such as the fraud, and how much can be attributed to systemic and industry-specific idiosyncratic factors one would expect to affect a company like Global Net. In total, the Government estimates that the victims sustained $17,492,817.45 in losses. Given the evidence available, this is a reasonable estimate.
The court held a Fatico hearing on February 14, 2012 and April 13, 2012 to allow the parties to examine and cross-examine their respective experts.
The Government's submissions are summarized in more detail below.
Gushlak's coconspirators aver (see generally Aff. of Salvatore Romano (Ex. 1 to Docket Entry # 70); Aff. of Howard Appel (Ex. 3 to Docket Entry # 70).) the following facts. At some point in 1999 Gushlak acquired a controlling interest in Global Net, then a small privately held telecommunications company. He later "backdoor" listed Global Net by reverse merging it into a publicly listed shell corporation called Rich Earth, which he had acquired for a few hundred thousand dollars. Once this process was complete, Gushlak controlled nearly one-hundred percent of the publicly listed merged entity. Gushlak then agreed with corrupt brokers, of whom Ramano was one, to "pump and dump" the stock. The brokers used aggressive sales techniques, including false statements, to generate interest in Global Net. When demand was high in early March 2000, Gushlak sold around one-million of his shares in three days through the brokers, paying the brokers about seventeen percent of the proceeds as a kickback. As time went on, the brokers continued to push the stock through illicit means and Gushlak continued to liquidate his holdings. The brokers made false claims about Global Net's business prospects, bought Global Net stock on behalf of unsuspecting customers without their consent, and, at times, refused to sell Global Net stock on their customers' behalf when requested to do so. With only Gushlak selling and numerous defrauded investors buying, the price of Global Net stock increased, and reached its peak in June 2000. By then, Gushlak had presumably sold most of his stock in Global Net, and, without the prospect of further kickbacks, the brokers were easing their pressure tactics. One of the two corrupt brokers, Montrose Capital, even stopped participating in the scheme altogether. Savvy investors quickly appreciated what was happening, and began short-selling Global Net. The price collapsed.
The reverse merged company originally remained known as Rich Earth—the source of its listing—but Gushlak changed the name back to Global Net a short time later.
The coconspirators claim that Gushlack made millions of dollars on this scheme, and it is easy to see how. He acquired control of Global Net for an unknown, but minimal, amount of money. He then acquired control of Rich Earth for between $325,000 and $350,000, and merged the two companies—most likely squeezing-out Global Net's other investors (if there were any). At that point, he relied on the false statements and other pressure techniques of corrupt brokers to hugely inflate demand for the newly merged company. Then, once demand was high enough, he sold his holdings at an inflated price. As his coconspirator, Salvatore Romano states, "these sale represented pure profit for Gushlak."
The Government's expert report supports the theory that fraud had a major effect on Global Net's stock price during 2000. The Government's expert, an economic consultant named David DeRosa, gained access to the daily historical closing prices of Global Net from February 29, 2000 until March 21, 2002. This appears to represent the entire time that Global Net was publically traded. Using the fraud time period established at Gushlak's plea allocution, DeRosa divided the data into two sets: one set, from February 29, 2000 through December 31, 2000, comprising the "manipulation period;" and another set, from January 1, 2001 through March 21, 2002, comprising the "post-manipulation period." (See DeRosa Expert Report at 5.) DeRosa also acquired historical closing price information for the same two periods of time for five competing companies ("Competitors") and the NASDAQ Telecommunications Index ("Index"), which is a composite of stocks of companies similar to Global Net. (See id. at 6-8.) He then ran two regressions. In the first regression he analyzed changes in the rate of return of Global Net's stock as against changes in the rate of return for the Index and the Competitors in the post-manipulation period. (See id. at 12-14.) In the second regression, he analyzed and compared these changes in the manipulation period. (See id.) The contrast between the results is stark. During the manipulation period, price changes of Global Net's stock were unrelated to price changes in the Index or among the Competitors. (See id.) During the post-manipulation period changes in the price of Global Net's stock were closely associated with changes in the Index.(See id.) In other words, during those months when Gushak was defrauding investors, the price of Global Net's stock rose and fell totally independently from an index of similar companies—companies that one would expect to have similar betas and the same industry-specific idiosyncratic risk profiles. This result indicates that during the manipulation period, some firm-specific idiosyncratic factor was driving almost all of the changes in Global Net's stock price. Unless Gushlak can point to some benign firm-specific event—such as a change in management, a new big customer, a fire, etc.—the only plausible candidate to be this factor is the fraud.
DeRosa has a PhD in Finance and Economics from the University of Chicago Business School. He operates a finance consultancy in New Canaan, Connecticut, and is an also an adjunct professor at the Columbia Engineering School. (DeRosa Expert Report at 2.)
In the post-manipulation period, changes in Global Net's stock price were also statistically significantly related to changes in the price of three, but not all five of the Competitors.
DeRosa's report not only indicates that the fraud affected Global Net's stock price, it also provides a reasonable estimate of the loss.
Analyzing the post-manipulation period, DeRosa learned that Global Net's rate of return was closely associated with that of the Index. (See id. at 13.) Indeed, the famed "t-statistic" for this relationship was nearly four. (Id.) His analysis also showed that a changes in the Index could explain over half of any change Global Net's stock price. (See id. at 13-14.) Based on the strength of the relationship between the Index's rate of return and Global Net's rate of return, DeRosa determined that changes in the Index's price could serve as a reliable proxy for changes in Global Net's stock price. (See id. at 14.) The tight correlation indicates that the Index had a similar beta as Global Net, and that the two were probably affected by the same industry-specific idiosyncratic risk factors. DeRosa thus used historical price information about the Index to calculate a "fair market", or "but for fraud," price for Global Net's stock during the manipulation period. (See id.)
A t-statistic is used to determine whether there is a statistically significant relationship between the independent and dependent variable. See Brerenson, Levine, Krehbiel, Basic Business Statistics at 515. Generally, with relatively large sample sizes, a t-statistic over two is regarded as establishing a statistically significant relationship. See id. 255-260 (9th ed. 2004).
Here is how he did it. DeRosa started with Global Net's closing price on January 1, 2001 (the first day that Global Net's stock price could be truly deemed un-tainted by fraud) and then referenced the Index's rate of return for the previous month—what percent of its value the Index lost or gained during December. (See id.) He then applied that rate of return retroactively to Global Net's stock, and was able to arrive a fair market price for a month earlier. (See id.) For example, if historical closing prices indicated that the Index lost five percent during December 2000, then DeRosa would take Global Net's January 1, 2001 closing price and solve for of what number the January 1, 2001 price was ninety-five percent. The solution was Global Net's fair market price for December 1, 2000. DeRosa then prorated the change to arrive at daily fair market values. (See id.) The difference between the price that Global Net's stock actually closed on a given day and the fair market price based on the Index can only be attributed to firm-specific idiosyncratic factors, and, absent a showing to the contrary by Gushlak, the only such factor that is plausible is Gushlak's fraud.
Thus, it was easy for DeRosa to calculate the victims' loss. The loss was simply the victims' aggregate trading loss, as accurately calculated in the Government's second restitution submission, less the aggregate fair market loss. (See id. at 16-17.) For instance, on an individual level, if a victim bought a share during the manipulation period for $15 when the fair market price was $3, and sold it during the same period for $6 dollars when the fair market price was $2, then the victim would be owed $8 in restitution—[(15 - 6) - (3-2)]. As noted above, when DeRosa preformed these calculations for all of the victims, he arrived at a total loss figure of $17,492,817.45. (Id. at 19.)
The Government's instant submission contains compelling evidence that: (1) Gushlak's fraud affected the price of the Global Net's stock during the manipulation period; (2) the effect was an initial jump in the stock's price followed by a precipitous decline as the fraud unwound; and (3) a reasonable estimate of the victims' loss as a result of these price changes is around seventeen and a half million dollars. In the face of this evidence, Gushlak must present equally strong countervailing evidence to prevent the Government from meeting its burden of proof.
He does not come close to doing so. Gushlak's opposition, like the Government's case, has two components: A narrative of the fraud; and an expert report. The narrative is mostly belied by Gushlak's guilty plea; and, where it is not, it is unsupported by exhibits. The expert report fails to provide a better alternative methodology for measuring victim loss.
Gushlak also sought to adduce testimony from an additional expert on the eve of the Fatico hearing on restitution. (See Gov't Mot. to Exclude Expert Witness (Docket Entry # 93).) Because this expert, Robert W. Lowry, had not prepared a report in anticipation of the hearing as required by previous court order, the court excluded his testimony from the first day of the hearing. (See Order granting Mot. to Exclude dated Feb. 13, 2012.) It did, however, grant a continuance so as to allow the Lowry to submit an expert report and testify at a later date. (See Minute Entry dated Feb. 14, 2012).
The court has now reviewed Lowry's report (Docket Entry # 99) and considered his testimony, and finds them utterly unrelated the issues raised at this stage in the sentencing proceeding. Whereas the Government must prove loss, Lowry's opinion is directed almost entirely to liability; whereas the Gushlak has pled guilty to a pump and dump scheme involving misstatements and omissions, Lowry's opinion focuses exclusively on deceptive practices—prearranged trades—that are not even alleged in this case.
Sections of Lowry's report indicate that Lowry either did not review the Government's submission or did not understand its most basic parts. His criticism, for instance, that Government did not attempt to establish a time frame for the Gushlak's fraud (Expert Report at 78) is belied not only by the Government's affidavits, but also by Gushlak's own plea allocution. Lowry's allegation that "DeRosa did not consider market evidence relevant to measuring investor losses and declared all [Global Net] losses attributable to manipulative activity" is also wrong. As is noted above, DeRosa calculated a fair market price to account for nonfraudrelated factors that affected Global Net's stock price. It is one thing to criticize the accuracy of DeRosa's calculations, it is quite another to insist that the calculations do not exist when they are prominently displayed for all to see.
Because of its irrelevance and poor quality, the court accords Lowry's expert opinion no evidentiary weight in determining whether the Government has met its burden of proof.
At the heart of Gushlak's narrative is a dramatic rejection of his guilt. Gushlak contends that his only crime was the omission of a single material fact from public disclosure. According to Gushlak's memorandum of law (Docket Entry # 80), the scope of his offense was limited to failing to initially disclose in public filings that he was paying one broker-dealer a commission for sales of Global Net. (See Mem. of Law at 2, 8.) Without evidentiary support, Gushlak maintains that the commissions to this broker were disclosed on May 30, 2000, thereby "revealing" the full extent of the fraud. (See id. at 1, 9.) Relying on the Second Circuit's decision in Rutkoske, Gushlak argues that any effect the fraud had on Global Net's stock price would have become apparent after the this revelation; and that, because, "the market shrugged" in response to the disclosure, (id. at 2), his fraud did not cause any loss (see id. at 2, 10, 20-22).
Gushlak does include citations for this assertion, but they are inconsistent and unsupportive. On the first page of his memorandum he cites to page 5 of Global Net's May 30 "Form DEC 14C" (2d Rosner Decl., Ex. 2 (Docket Entry # 82, Ex. 1).) Later Gushlak cites to page 6 of the Form DEC. (Mem. of Law. at 9.) A review of both pages, however, fails to reveal any such disclosure, and the court found none elsewhere in the document. Although, for reasons discussed below, any disclosure of the kickbacks to one broker would be irrelevant given Gushlak's guilty plea, the irregularities surrounding Gushlak's citations for his key factual assertion prevents the court from crediting this statement.
Apparently in the alternative, Gushlak disputes DeRosa's estimates of Global Net's fair market value—which at the beginning of the fraud was around $2.50—by insisting that he paid six-million dollars for his stake in the company, buying shares for ten dollars a piece. This also cannot be verified, and it would prove nothing even if it were supported. Gushlak again cites to inapposite portions of the Form DEC 14C and to the declaration of his counsel Brian Rosner, which states, "I have discussed this [Global Net's] history with Mr. Gushlak. He has informed me . . . that he purchased the shares in the first $6,000,000 Private Placement . . .." (2d Rosner Decl. ¶ 29.) Needless to say, self-serving hearsay cannot, without more, be credited by the court. It is not plausible that Gushlak could have paid six-million dollars for a private placement and have no record evidence of the transaction. In any event, even if Gushlak actually did pay that amount, it is, of course, possible that he simply over-paid.
This theory is incredible. Not only is its principal factual assertion unsupported, see supra note 13, but it is totally at odds with the allegations set out in the Government's Information—all of which Gushlak has admitted to be true. Gushlak did not plead guilty to single material omission. He pled guilty to a wide-raging conspiracy that spanned nearly two years, and in which a number of individuals worked continually to pump up the price of Global Net's stock. (See Plea Tr. at 26-28.) The narrative also fails to rebut, or substantively address,the detailed affidavits of two of Gushlak's coconspirators.
Gushlak does attack the assertion that Montrose Capital's repudiation of the conspiracy affected the price of Global Net's stock. (See Mem. of Law at 29-32.) But it is irrelevant whether Montrose Capital continued in the conspiracy because the Government attributes the price collapse to several other factors, such as short-sellers and an overall reduction in "pumping" activities. (See Sentencing Mem. at 4.) Similarly, Gushlak's criticism of the agent's sworn statements about Gushlak's repurchasing Global Net stock to "support" the price (Mem. of Law at 25) is of no bearing because Gushlak's repurchase is not alleged to be an essential part of his scheme.
Gushlak also has no credible answer to the Government's expert report. In his narrative he claims again and again that the spike and collapse of Global Net's stock price is attributable to the "mania" associated with the dot-com bubble. (See, e.g., Mem. of Law at 2.) But this theory fails to explain why Global Net's stock price behaved in a radically different fashion than did the Index or the Competitors's stock, all of whom were subject to same market forces and environmental conditions. It does not even begin to address the question of why this difference would have disappeared once the fraud ended.
Gushlak's expert, David Juran, raises some more legitimate concerns with the Government's expert report, but nothing to seriously undermine its reliability.
Juran criticizes DeRosa's use of the Index as proxy by which to calculate a fair market price for Global Net's stock during the manipulation period, arguing that the Index would not capture firm-specific events that could affect Global Net's stock price. (See Juran Expert Report (Docket Entry # 81) at 3.) This is true, but the onus must be on Gushlak identify events other than the fraud that contributed strongly to changes in Global Net's stock price. As discussed above, Gushlak's alternative explanation is limited to industry-wide events—like the bursting of the dot-com bubble—that would have affected the Index as much as Global Net.
On cross-examination of DeRosa during the Fatico hearing, counsel for Gushlak also posited a theory that investors would have valued Global Net differently than the Competitors, not because of any fundamental difference in the firms' structure, activities, or business plan, but because at some point in time it had a higher ratio of revenue to market capitalization. (See Hr'g Tr. (not docketed) at 74-78.)
This makes no sense. To begin with, as DeRosa testified, revenue is an exceedingly crude and partial device for valuing a stock. (See id. at 7475 ("[I]f that's how you did your analysis of stock's you'd be poor very quickly.") Equity prices are a function of expected future earnings, see Brealey, Myers, Allen, supra at 88, and so any investor using accounting numbers to value a stock would be at least as interested in a firms' costs as he or she would be in its revenues. Gushlak provides no data on costs. Next, and more importantly, any explanation of change in Global Net's stock price must start with a change in the outside world. Gushlak does not point to any change in Global Net's revenue. Instead he provides a snapshot in time arguing that the difference in revenue that existed at that moment meant Global Net was a "more attractive purchase." (Id. at 74.) To the extent that investors were interested in the issue, any difference in revenue between Global Net and the Competitors would already be priced into Global Net's stock at that time; what would move Global Net's stock price would be a change in revenue. Finally, because DeRosa compared rates of return between companies, it would not be enough to merely show that Global Net's revenue was growing (which is not even shown here); Gushlak would also have to show Global Net's revenue's growing relative to the Competitors. Gushlak provides no data on this.
For these reasons the court does not credit Gushlak's theory that a difference in the ratio of revenue to market capitalization between Global Net and its competitors explains in any way why those companies' rates of return were uncorrelated during the fraud and closely correlated afterward. Rather, it appears that Gushlak's data are the product of mining—a search for any nonfraudrelated metric, no matter how irrelevant, that would distinguish Global Net from the Competitors during the period of noncorrelation.
Juran next points out that in the post-manipulation period, the Index only explained about fifty-seven percent of the changes in Global Net's stock price, but that in DeRosa's counterfactual the Index explains one-hundred percent of the changes in the fair market price of Global Net's stock. (See id. at 5.) The response to this critique is that victim loss estimates need not be precise; they need only be reasonable. It is impossible to know the exact price that Global Net's stock would have traded for day-by-day but for the fraud, and, if the Index is not fully accurate, there is as much chance that it over-estimates the fair market price as that it under-estimates it.
Obviously, if DeRosa's fair market price estimates were too high, rather than too low as Gushlak and Juran argue, the amount of restitution Guslak owed to the victims would be even greater.
As an alternative to DeRosa's methodology Juran suggests a so-called "life-cycle" analysis of Global Net and some of the Competitors. (See id. 3-5.) This analysis looks for relationships between changes in the firm's stock prices, not by dates, but by the age of the firms as measured from the peak closing price for each company's stock. Not surprisingly, Juran finds a close relationship between Global Net's rate of return and that of the Competitors across the companies' life cycles (see id. at 5), but Juran does not explain why he believes that this methodology is superior. There are several good reasons to think that it is not. For one thing, unlike with DeRosa's model, a life-cycle analysis does not control for various outside events—systemic and industry wide factors—that would affect all of the companies and that are reflected in standard temporal models. The importance of identifying these factors is discussed above. For another, it is unsurprising that, when measured from their peak, rates of return of the companies in Juran's model track each other. Depicted graphically, all will form, to one extent or another, a bell curve, rising up toward the peak and then falling away from it.
Finally, Juran raises some minor technical questions about DeRosa's methodology. He complains that DeRosa did not make the raw data available so that Juran could exactly replicate his regressions (see id. at 2.), and criticizes DeRosa for using monthly rather daily returns for the Index in calculating the fair market price for Global Net (see id. at 5). Finally, he suggests that DeRosa could have used a better proxy than the Index, based on the results of post-manipulation period regressions. (See id.) None of these attacks truly call into question the sufficiency of DeRosa's model.
Indeed, the court finds that the Index was the best proxy. Juran criticizes DeRosa for not using some combination of Competitors and the Index that had a higher "R-squared" than the Index alone. (See Juran Expert Report at 5.) The R-squared measures the proportion of variation of a dependent variable—this case Global Net's rate of return—that can be explained by an independent variable—in this case the rate of return of the Index, the Competitors, or some combination thereof. See Berenson, Levine, Krehbiel, supra at 496. The Index did not have the highest R-squared, but, as noted above, it did have the highest "t-statistic." A high t-statistic indicates that there is a genuine association between two variables. See supra note 11. In the absence of such an association, the correlation reflected in the R-squared could be spurious. (See Hr'g Tr. at 48.) DeRosa was therefore correct to use the Index as a proxy because its high t-statistic shows that the Index's rate of return has the most meaningful relationship to Global Net's rate of return.
--------
In light of Gushlak's guilty plea, the court must credit the allegations contained in the Information. It must further accept any facts that Gushlak admitted were true in his plea allocution; and, because Gushlak has not produced any evidence to rebut the substance of the affidavits attached to the Government's most recent submission, the court fully credits the facts contained there as well.
This leaves the court with a relatively well developed picture of Gushlak's fraud. A fraud that began when Gushlak acquired control of Global Net for small consideration and continued until the time that Gushlak and his possibly also his conspirators had foisted all of their holdings on unsuspecting investors at greatly inflated prices. DeRosa's model shows by preponderance of the evidence that this scheme had an effect on Global Net's stock price, and it provides a reasonable estimate of the extent of that effect—about seventeen and a half million dollars.
Accordingly, the court ORDERS Gushlak to pay restitution for the amount of $17,492,817.45. SO ORDERED.
Dated: Brooklyn, New York
April 20, 2012
_________________________
NICHOLAS G. GARAUFIS
United States District Judge