David Munoz, Bennie McGregor, and Donald Thomson appeal their convictions and sentences for mail fraud and aiding and abetting under 18 U.S.C. §§ 1341 & 2 (2000). After a month-long trial a jury found all three defendants guilty of mail fraud for participating in a fraudulent “Ponzi” investment scheme. We have jurisdiction and affirm their convictions but vacate the sentences of Munoz and McGre-gor and remand for resentencing. We affirm Thomson’s conviction and sentence.

The three defendants press numerous claims of error on appeal. Two major questions predominate. First, we must decide whether for the purposes of calculating loss under U.S.S.G. § 2F1.1 the district court erred in refusing to offset the intended loss by (1) payments made to investors as an alleged return on their investment and (2) the actual value of the investment after the illicit company reorganized in bankruptcy and began doing business as a legitimate enterprise. We hold that the district court properly calculated the loss for sentencing purposes by using the intended loss standard, without offsetting the loss by the value recovered by the victims.

Second, we must decide whether the district court erred at sentencing by using the preponderance of the evidence standard in determining the relevant conduct of Munoz and McGregor. The relevant conduct determination resulted in a fourteen-level rather than a five-level upward adjustment for each defendant. Because the law has changed since Munoz and McGregor were sentenced, we vacate their sentences and remand for resentencing. The district court should reexamine the relevant conduct evidence using the clear and convincing evidence standard in considering the appropriate sentence for these two defendants.

I. FACTUAL BACKGROUND

To understand the issues better, we set forth the relevant facts in some detail.

In 1986, Jean Claude Leroyer formed a corporation called Metro Display Advertising, Inc. (MDA) to sell bus stop shelters as investments to the public. From 1986 to 1991, MDA ostensibly sold bus stop shelters to investors through its sales agents for $10,000 each, then leased the shelters back from the investors for a fee of $200 per month, less a $30 maintenance fee. As part of the sales contract, MDA agreed to repurchase the shelters from the investors after five years for $10,000. MDA was to solicit advertisers to place ads on the shel ters in order to generate the necessary revenue to make the lease payments.

Over the five years that MDA was in business, it sold approximately 4,600 bus stop shelters to 1,442 investors, but installed no more than 2,600 shelters. MDA’s advertising revenues were insufficient to cover the lease payments and overhead, so MDA used the capital investments from new investors to cover those expenses. In short, the shelter investment was a Ponzi scheme.

In December 1991, MDA agreed with the Securities and Exchange Commission, after a ten-month investigation, to stop selling the bus stop shelter investment. The investment scheme inevitably collapsed. MDA no longer made lease payments to investors and promptly filed for bankruptcy, declaring over $100 million in debt and less than $1 million in assets.

The company subsequently reorganized and was taken over by an investor. The investor, with the help of other victim investors, rebuilt the company into a legitimate, profit-generating business. In 1998 it was sold to a bona fide purchaser and $37 million in proceeds from that sale were deposited in escrow as restitution for the victim investors.

The three defendants, Munoz, McGre-gor, and Thomson, were independent sales agents of MDA. They made several fraudulent and misleading statements to perpetuate the scheme. For example, they and their agents told investors that MDA was a financially sound company with adequate advertising revenues to cover its lease payments. Moreover, they concealed from investors the fact that the SEC was investigating MDA for potential securities violations beginning in March of 1991.

To generate sales they used a sales brochure given to them by MDA. The sales brochure contained a breakdown of the costs of the shelter accounting for the entire $10,000 investment. The brochure itemized $9,000 in materials, labor, and permits for the construction of the shelter, and $500 each for “marketing fees” and “consulting fees.” This led investors to believe that sales agents were only receiving $1,000 in commissions. The sales agents, however, were receiving $2,500 total in commissions from each $10,000 shelter sale.

The defendants also used opinion letters prepared by attorneys, which stated that the sales contracts were not securities. The attorneys based their opinions on the representations of Leroyer that the investments were not advertised to the public and that purchasers would be only sophisticated investors who were financially qualified in advance to purchase. The defendants did not, however, financially pre-qualify the investors as required.

Munoz, McGregor, and Thomson were indicted on ten counts of mail fraud stemming from the mailing of lease payments to ten investors on December 15, 1991. The jury convicted Munoz and McGregor on two counts each, only for those mailings to the investors to whom they or their companies directly sold the shelters. Thomson was convicted on all ten counts.

A. Defendant David Munoz.

Munoz sold shelters from 1989 to 1991, first as a sub-agent of Thomson, then as a direct agent of MDA through his company, IBT Financial. IBT sold approximately 883 shelters during its operation for a total of $8.7 million in investor loss. The jury convicted Munoz on two counts for the December 15, 1991, mailings of lease payments to two investors who purchased shelters from IBT. The sales were made in October and November of 1991.

Using the “intended loss” standard under U.S.S.G. § 2F1.1, cmt. n. 8, the probation officer in the presentence report (PSR) calculated the loss caused by Munoz’s conduct at $8.7 million, the total amount of investment money generated by IBT. This resulted in a fourteen-level upward adjustment, which, when combined with a two-level upward adjustment for “more than minimal planning,” put Munoz’s offense level at 22 with a corresponding guideline range of 41 to 51 months.

Munoz objected to the loss calculation on two grounds. He argued first that he should only be held accountable for the loss to the two investors named in the indictment, not for all of the investors to whom IBT sold shelters. He also argued that the loss should be offset by the amount of money recovered by the two investors through the lease payments and the 1998 sale of MDA, placing the total loss amount at $62,890.37. Offsetting the loss by investor recovery returns would have resulted in a five-level upward adjustment to an offense level of 13, with a corresponding guideline range of 12 to 18 months.

The district court agreed with the recommendation in the PSR and found that the intended loss formula was the appropriate standard. The court further concluded that the relevant conduct for which Munoz was accountable included all sales by IBT to investors, not merely the sales to the two investors named in the indictment. Finally, the court found that the value of the investments at the time of the transactions was negligible. Thus, the court would not offset the loss by the amount of the lease payments and the revenue the victims realized by selling MDA six years after MDA declared bankruptcy. The court sentenced Munoz to 41 months’ imprisonment and three years of supervised release. He was ordered to pay restitution in the amount of $97,200.

B. Defendant Bennie McGregor.

McGregor sold shelters from 1989 to 1991 through his companies, McGregor Financial Group, Inc., and McGregor Investment Group, Inc. (MFG and MIG). MFG and MIG sold approximately 550 shelters during their operation for a total of $5,505,000 in investor loss.

The jury convicted McGregor on two counts for the December 15, 1991, mailings of lease payments to two investors who purchased shelters from MIG and MFG. The sales to those investors were also made in October and November of 1991.

The PSR calculated the loss caused by McGregor’s conduct at $5.5 million based upon the total amount of investment money generated by MFG and MIG. McGregor challenged the loss calculation at sentencing on the same grounds as did Munoz. The district court again followed the recommendation of the PSR, leading to a fourteen-level upward adjustment rather than the five-level upward adjustment argued for by McGregor. The court granted McGregor a two-level downward departure based on his level of culpability and his history, leading to an offense level of 20, with a corresponding guideline range of 33 to 41 months. The court sentenced him to 36 months’ imprisonment.

C. Defendant Donald Thomson.

Thomson sold shelters from 1986 to 1991 through his company, Financial and Accounting Consultants (FAC). The jury convicted Thomson on all ten counts in the indictment, including the four counts pertaining to the mailings to Munoz and McGregor’s customers. The jury apparently found that Thomson played a more extensive role in the investment scheme than did his co-defendants.

At sentencing, the district court followed the recommendation of the PSR and used the standard of “intended loss,” calculating the loss caused by Thomson’s conduct at $46,497,800 for the total sales of MDA. The court refused Thomson’s request to offset the loss by the amount the investors were expected to recover from escrow on closing of the sale of MDA.

The district court also rejected Thomson’s argument that he should only be held accountable for the sales made by his own company rather than all MDA sales. Thomson argued that the loss calculation should have been $17,115,000, a fourteen-level upward adjustment. Instead, Thomson received a seventeen-level upward adjustment. The court also enhanced his sentence by three levels for playing a managerial/supervisory role and by two levels for obstruction of justice. The court granted Thomson a one-level downward departure based on a combination of other factors, placing his total offense level at 29, with a corresponding guideline range of 87 to 108 months. The court sentenced him to 92 months and ordered him to pay restitution in the amount of $138,798.

II. DISCUSSION

A. Calculation of Loss.

Munoz, McGregor, and Thomson contend that the district court erred by sentencing them according to intended rather than actual loss. They argue that the total loss should be offset by the amounts recovered by the victims through the lease payments made to the victims over the course of the scheme and the reorganization and sale of MDA in 1998. See U.S.S.G. § 2F1.1, cmt. n. 8(a). We affirm the district court’s use of the intended loss standard without offsetting the loss by the amounts ultimately recovered by the victim investors.

In calculating the amount of loss attributable to a defendant for sentencing purposes in a fraud case, the district court should use the amount of loss that the defendant attempted to inflict, provided such a figure can be determined with reasonable certainty and is greater than the actual loss inflicted. U.S.S.G. § 2F1.1, cmt. n. 8. The district court’s application of the Sentencing Guidelines to the facts of a case is reviewed for abuse of discretion. United States v. Leon-Reyes, 177 F.3d 816, 824 (9th Cir.1999). Its factual findings in the sentencing phase are reviewed for clear error. United States v. Barnes, 125 F.3d 1287, 1290 (9th Cir.1997).

Several circuits have grappled with the issue of whether, in a Ponzi scheme case, the loss calculation should include the total investment money generated by the schemers, or should be offset by any value received by the defrauded parties. The Second, Fourth, Fifth, and Seventh Circuits have held that the loss calculation in a Ponzi scheme should not be offset by the amount of the victims’ recovery. See United States v. Carrozzella, 105 F.3d 796, 805 (2d Cir.1997); United States v. Loayza, 107 F.3d 257, 265 (4th Cir.1997); United States v. Deavours, 219 F.3d 400 (5th Cir.2000); United States v. Lauer, 148 F.3d 766, 768 (7th Cir.1998); but see United States v. Holiusa, 13 F.3d 1043, 1046-47 (7th Cir.1994) (holding that intended loss in a Ponzi scheme case did not include amounts ultimately returned to investors).

These courts reason that the gravity of the crime should be measured by the entire sum of money that the schemers put at risk through the misappropriation regardless of whether some victims were fortunate enough to recover part of their loss. Lauer, 148 F.3d at 768 ; cf. United States v. Janusz, 135 F.3d 1319, 1324 (10th Cir.1998) (adopting this line of reasoning in a fraud case where the financial consultant misappropriated clients’ funds, most of which were recovered by the clients when the consultant’s accounts were frozen). Because the schemers typically return money to investors to perpetuate the fraud and ensnare new investors, and not to mitigate damages to the current investors, these courts reason that they should be held accountable for all of the funds that are misappropriated. See Carrozzella, 105 F.3d at 805 .

The Eleventh Circuit is the only circuit that has not adopted the “risk” theory of loss calculation. See United States v. Orton, 73 F.3d 331, 334 (11th Cir.1996). In Orton , however, the court was not directly presented with the issue of whether the risk theory is an appropriate loss calcula tion theory. The court did state that “risk of harm” was not an appropriate measure of loss calculation, but the parties in Orton were actually debating about two types of “net loss” calculation theories, one focusing on the actual loss to each individual victim, the other focusing on the actual gains made by the defendant. Id.

We believe that the “risk” theory of loss calculation is the sensible approach in this case. The goal in sentencing is to deter criminals from engaging in illegal behavior, such as making fraudulent or misleading statements, that deliberately leads unwitting investors to put their money at risk. A Ponzi scheme, in which new investor funds are used to pay returns to prior investors, creates a situation where the business will inevitably collapse at the expense of the investors. If it does not collapse, it is usually by luck alone. Thus, whether a Ponzi scheme produces some value for the investors is irrelevant to calculating the intended loss.

In Lauer , the Seventh Circuit likened this situation to a person who embezzles money from his employer, planning to gamble with it and put the embezzled money back before he is discovered. Lauer, 148 F.3d at 768 . If he is caught prior to placing any bets, and the company recovers all of its money, the embezzler should still be held accountable for the money he stole. Id. Likewise, if the theft is discovered after the embezzler gambled the money, won, and replaced the embezzled money, he should still be held accountable for the amount embezzled.

The defendants contend that the shelter investments produced tangible goods-the shelters-that have independent value as fixed point outdoor advertising sites. Thus, according to the defendants, the proper loss calculation should be the difference between the amount the investors paid for the shelters ($10,000) and the total amounts they recovered once MDA was sold and the $37 million escrow fund was distributed to the victim investors.

The value of the shelter investment, however, was in the advertising revenues that MDA promised to generate and the resulting lease payments, representing a 20% return on the investment. The value was not in the actual physical structures, as evidenced by the fact that none of the investors at the time of purchase refused the leaseback provision, instead opting to generate advertising revenues themselves. Moreover, MDA sold 4,600 shelters, but constructed only 2,600. Thus, the shelter investment was essentially a security, and its value depended on the value of the company as a whole.

When the scheme collapsed and MDA declared bankruptcy the investment had no value. The company declared debts of over $100 million and assets of only $1 million. Between the bankruptcy in 1992 and the sale of MDA in 1998, the investors received nothing for their investments. The investments eventually acquired value only because a victim investor took over the company and, with the help of other victim investors, turned it into a legitimate business by constructing more shelters and selling sufficient advertising to make the lease payments. For these reasons, we hold that the district court properly calculated the loss attributable to each defendant based on intended loss, without offsetting the calculation by amounts ultimately recovered by the victim investors.

B. Determining Relevant Conduct.

In determining relevant conduct for sentencing purposes in a fraud case, a district court may consider fraudulent conduct by the defendant other than that for which evidence was offered at trial. See United States v. Amlani, 111 F.3d 705, 719 (9th Cir.1997). The factual findings underlying the relevant conduct enhancement generally must be supported by a preponderance of the evidence. See United States v. Collins, 109 F.3d 1413, 1420 (9th Cir.1997). In the time since the defendants were sentenced, however, this Court has held that a sentencing factor that has an extremely disproportionate effect on the sentence may require a district court to find that factor by clear and.convincing evidence, rather than by a preponderance of the evidence. See, e.g., United States v. Mezas de Jesus, 217 F.3d 638, 643 (9th Cir.2000); United States v. Hopper, 177 F.3d 824, 829 (9th Cir.1999).

Munoz and McGregor each received a fourteen-level rather than a five-level upward adjustment on the basis of uncharged relevant conduct-the sale of all the shelters by themselves and their companies-despite being charged with and convicted of only two fraudulent sales each. This upward adjustment effectively increased their individual sentencing ranges from 12-18 months to 41-51 months. They each contend that the nine-level difference in the upward adjustment was sufficiently disproportionate to require the district court to apply the clear and convincing evidence standard to the factual findings.

In Mezas de Jesus , this Court found that a nine-level upward adjustment based on an uncharged kidnaping that resulted in a sentence increase from 21-27 months to 57-71 months was sufficiently disproportionate to trigger the clear and convincing evidence standard. Mezas de Jesus, 217 F.3d at 643 .

In United States v. Romero-Rendon, 220 F.3d 1159 (9th Cir.2000), this Court held that a sixteen-level upward adjustment for a prior conviction that increased the defendant’s sentencing range from 4-10 months to 41-51 months triggered the clear and convincing evidence standard. Id. at 1161 . We further held that the clear and convincing evidence standard was met in Romero-Rendon on the basis of the facts outlined in the PSR. Id. at 1163 . Romero-Rendon , however, is different from this case. There, the defendant did not challenge the factual allegations contained in the PSR upon which the enhancement was based. Id.

Munoz and McGregor’s PSRs listed all of the individual sales by IBT and MFG/ MIG. Munoz and McGregor did not challenge the PSR conclusion that they and their companies made those sales. But they did dispute the findings that they had knowledge of the fraudulent scheme at the time each of those sales was made. They argue that the jury only found that they had knowledge of the fraud at the time of the sales to the two customers to whom they were convicted of making fraudulent sales. According to Munoz and McGregor, the district court therefore should have included as relevant conduct only the two sales upon which each defendant was convicted.

Munoz and McGregor are incorrect in their assertion that the district court could not include relevant conduct other than the two sales for which they were convicted. The district court is entitled to take into account all relevant conduct, charged and uncharged, provided that the relevant conduct findings are supported by sufficient evidence. The caselaw is clear on this point. See, e.g., Amlani, 111 F.3d at 719 .

We hold, however, that in this case the district court should have relied at sentencing only on those factual findings supported by clear and convincing evidence. The upward adjustments in Munoz and McGregor’s cases are sufficiently disproportionate to justify a remand to the district court to determine whether the evidence is clear and convincing that Munoz and McGregor knowingly and intentionally engaged in all of the uncharged conduct upon which the upward adjustments were based. We state no opinion on what the district court’s determination should be under this heightened standard of proof. That is within the province of the sentencing court to decide.

C. Munoz’s Remaining Claims.

1. Attorney-Client Privilege.

Attorney Alvin Sherron was a key witness for the government in the case against Munoz. Munoz had hired Sherron to assist him in unrelated matters prior to his involvement with MDA. He referred Sherron to MDA to prepare one of the opinion letters used to attract investors. Sherron testified, over Munoz’s objection, that Munoz stated to Sherron in late 1990 that MDA was inadequately funded and was in financial trouble. We affirm the district court’s decision to admit the testimony.

The party asserting attorney-client privilege has the burden of establishing all of the elements of the privilege. See United States v. Plache, 918 F.2d 1875 , 1379 n. 1 (9th Cir.1990). We review de novo whether the party has met these requirements including whether he establishes an attorney-client relationship. United States v. Bauer, 132 F.3d 504, 507 (9th Cir.1997).

Munoz argues that he had an attorney-client relationship with Sherron both as an individual and jointly with MDA. As evidence, he points to the following facts: (1) he referred Sherron to MDA based on his prior relationship with Sherron; (2) he furnished information to Sherron about MDA to assist Sherron in preparing the opinion letter; (3) Sherron consulted Munoz more often than he consulted Leroyer; and (4) Munoz believed at the time he made the damaging statements that Sherron was his attorney.

We agree with the district court that this evidence is insufficient to establish an attorney-client relationship. Sherron addressed all correspondence to Leroyer and MDA without copying Munoz. Munoz offered no evidence that he consulted with Sherron for personal legal advice about his own involvement in the investment scheme. In addition, Sherron issued a sworn declaration and testified at the pretrial hearing that he was not acting as Munoz’s attorney in this transaction.

Most importantly, only MDA was paying Sherron’s fees and only MDA signed a retainer agreement with Sherron. Although an attorney-client relationship may exist even though a third party is paying the legal fees, see Dole v. Milanas, 889 F.2d 885 (9th Cir.1989), Munoz offered scant evidence to refute the proof that Sherron was only working for MDA in this transaction. We therefore affirm the district court’s decision that Munoz did not establish an individual attorney-client relationship with Sherron.

Munoz further asserts that he held the privilege jointly with MDA. For the same reasons that he did not establish an attorney-client relationship in an individual capacity, we find that he did not establish the relationship in a joint capacity. See Sky Valley Ltd. Partnership v. ATX Sky Valley, Ltd., 150 F.R.D. 648 (N.D.Cal. 1993) (holding that an attorney-client relationship for the purposes of joint-client privilege can only be formed by express or implied contract). Although Munoz supplied Sherron with information regarding the transaction, he did not sign a retainer agreement with Sherron, and did not seek Sherron’s advice in an individual capacity. In short, Munoz offered no evidence of an express or implied contract between himself and Sherron. For these reasons we find that the district court did not err in admitting the testimony of Sherron.

2. Closing Argument.

At the close of the trial, Munoz’s attorney reserved two hours for closing argu ment. He concluded the argument on a Friday afternoon after using only one and one-half hours. The court then recessed for the weekend after the government finished its rebuttal argument. The following Monday, Munoz’s attorney asked the court to re-open closing arguments so that he could use the remaining thirty minutes to address some issues that he felt he had left open. The court rejected his motion after a hearing and it found that Munoz had not made an adequate showing of prejudice.

A district court has wide discretion in limiting time for closing arguments. Barnard v. United States, 342 F.2d 309, 321 (9th Cir.1965). Provided a defendant has adequate time to make all legally tenable arguments supported by the facts of the ease, the district court will not be reversed for limiting closing arguments. United States v. Gaines, 690 F.2d 849, 858 (11th Cir.1982).

Munoz’s attorney did not identify any specific issues that he had failed to address in his closing argument. The district court reviewed the transcript of his closing argument and found that he had covered all of the obvious arguments on his client’s behalf. The court then decided that granting his request would give Munoz an unfair advantage because the government had already completed its closing arguments. We hold that the district court did not abuse its discretion in denying Munoz’s request to reopen closing arguments.

D. McGregor’s Remaining Claims.

1. Sufficiency of the Evidence.

McGregor argues that the evidence was insufficient to establish that he knowingly made false statements in furtherance of the fraudulent scheme. There is sufficient evidence to support a conviction if, viewing the evidence in the light most favorable to the prosecution, a rational jury could have found the elements of the crime beyond a reasonable doubt. Jackson v. Virginia, 443 U.S. 307, 319 , 99 S.Ct. 2781 , 61 L.Ed.2d 560 (1979). When sufficiency of the evidence is preserved by making a motion for acquittal after the close of the evidence, we review the district court’s denial of the motion de novo. United States v. Tucker, 133 F.3d 1208, 1214 (9th Cir.1998).

Under 18 U.S.C. § 1341 the elements of mail fraud are: (1) proof of a scheme or artifice to defraud, and (2) using or causing the use of the mails in order to further the fraudulent scheme. See United States v. Serang, 156 F.3d 910, 914 (9th Cir.1998). The government must demonstrate specific intent to defraud but intent may be inferred from the defendant’s statements and conduct. See United States v. Beecroft, 608 F.2d 753, 757 (9th Cir.1979).

The government points to four categories of evidence that demonstrate McGregor knew the scheme was fraudulent at the time of the December 15, 1991, mailings. First, McGregor made statements to Paul Noe, one of his salesmen, as early as April 1991 that MDA was using new investment money to make the lease payments. He told another salesman, Peter Castillo, in July 1991 that he thought MDA was operating a Ponzi scheme.

Second, in mid-1991, McGregor left MDA to form his own company, Shelter Concepts Advertising Network. McGre-gor operated Shelter Concepts as a Ponzi scheme, using new investor money to make the lease payments to old investors. Despite making several sales, he never constructed a shelter and never solicited companies to place advertisements on shelters. Castillo testified that, to his knowledge, no one solicited advertisers to place ads on shelters and Shelter Concepts had only three or four sites on which to build shelters. Yet Shelter Concepts continued to make lease payments until the fall of 1991, when McGregor folded those investments back into MDA because he was unable to continue making the lease payments.

Third, he used the opinion letters drafted by the attorneys knowing that the factual premises on which the opinions were based were false. The opinion letters were issued, in part, on the assumption that all investors would be qualified financially in advance to invest in the scheme and that the investment would not be advertised to the general public.

When the SEC began investigating MDA, MDA told its sales agents to pre-qualify investors by requiring them to fill out forms that requested information about the investors’ assets and investment experience. According to testimony by investors, McGregor never pre-qualified them to invest. McGregor had the investors sign the signature page of the forms without showing them the pages containing the investor requirements. He then attached the investor requirement pages afterwards to make it look as if the investors had been pre-qualified. Several investors testified that they would not have passed the financial requirements set out in the forms. Moreover, in order to get around the prohibition on general advertisement of the shelter investment, McGre-gor told his salesmen to attempt to sell the shelter investment by placing general advertisements for living wills, then making sales pitches for the shelter investment to those who responded to the ads.

Finally, McGregor used the sales brochure to lead investors to believe that the salesmen were only making $1,000 in commissions for each sale when they were actually making $2,500 in total commissions.

From all of this evidence a rational jury could find that McGregor had the specific intent to defraud and that all of the elements of mail fraud were established by proof beyond a reasonable doubt. We therefore conclude that the evidence was sufficient to support McGregor’s conviction.

2. Jury Instructions.

At the conclusion of the trial, McGregor requested that the jury be instructed that if an allegedly false statement made by McGregor could reasonably be interpreted to be true, the jury must find that McGre-gor did not make a false statement. The district court rejected this proposed instruction, reasoning that the good faith instruction adequately covered this aspect of McGregor’s defense theory.

McGregor also requested that the jury be given an advice of counsel instruction. The district court also rejected that request.

In reviewing jury instructions, the relevant inquiry is whether the instructions as a whole are adequate to guide the jury’s deliberation. See United States v. Frega, 179 F.3d 798 , 807 n. 16 (9th Cir. 1999). We review de novo a trial court’s denial of a proposed jury instruction when the issue is whether the instructions given adequately presented the defendant’s theory of the case. See United States v. Knapp, 120 F.3d 928, 930 (9th Cir.1997). We affirm the district court’s decision to refuse McGregor’s proffered jury instructions.

a. Reasonable Interpretation Instruction.

On appeal McGregor argues that the rejection of his proffered “reasonable interpretation” jury instruction violated his right to a fair trial. His specific complaint relates to the allegation that he misled investors regarding the amount of commissions the salesmen were making. The jury was instructed that it must unanimously agree on the type of fraudulent behavior upon which it was convicting. But the jury did not render a special verdict disclosing the behavior that they had agreed was fraudulent. Because the prosecution offered evidence of several different types of fraudulent activity, McGregor argues that the jury may have convicted him based solely on the sales brochure that misled investors to believe the salesmen were only making $1,000 in commissions.

He further argues that there is a reasonable interpretation of the sales brochure that makes it true-that is, that the salesmen were only receiving $1,000 in commissions from the revenues generated by the individual shelter sales and that the remaining $1,500 in commissions were coming from some other revenue source. Thus, he contends, the district court’s refusal to give the reasonable interpretation jury instruction may have led the jurors to convict based on an incomplete or misleading statement of the applicable law.

As support for his argument, McGregor cites to several cases involving 18 U.S.C. § 1001 . See United States v. Dale, 991 F.2d 819 (D.C.Cir.1993); United States v. Race, 632 F.2d 1114 (4th Cir.1980); United States v. Anderson, 579 F.2d 455, 460 (8th Cir.1978). Section 1001 criminalizes the making of false statements to governmental agencies. In the cases cited by McGre-gor, the defendants were required to report certain items to government agencies at each agency’s request. The defendants argued that the agencies’ requests were susceptible to more than one interpretation. These courts held that the prosecution must prove beyond a reasonable doubt that the requests were subject to no reasonable interpretation that would make the defendants’ responses to the agencies’ requests factually true. Cf. United States v. Migliaccio, 34 F.3d 1517 (10th Cir.1994) (holding that the reasoning in § 1001 cases applies equally well in mail fraud cases in a reporting case where the defendants’ statements to the agencies were allegedly susceptible to more than one interpretation).

This case- is distinguishable from the cases cited by McGregor, however, all of which involved prosecutions where the defendant was convicted of making false statements. Under the mail fraud statute the government is not required to prove any particular false statement was made. See United States v. Halbert, 640 F.2d 1000, 1007 (9th Cir.1981). Rather, there are alternative routes to a mail fraud conviction, one being proof of a scheme or artifice to defraud, which may or may not involve any specific false statements. Id.; see also Lustiger v. United States, 386 F.2d 132, 136 (9th Cir.1967) (affirming mail fraud conviction of a real estate salesman who mailed purchasers brochures that, while not literally false, misled purchasers to believe that the real estate had access to water).

McGregor argues that the government prosecuted him on a false statement theory of mail fraud. One of the jury instructions-that the jury must unanimously agree on which false statement it believed the defendant made in order to convict-tends to support this conclusion. The jury instructions, however, defined false statements as including statements of half-truths and concealment of material facts. Moreover, McGregor was aware of the prosecution’s theory of the case since the indictment charged the defendants with a scheme to defraud, and the specific instances of fraudulent behavior listed as evidence of the scheme are more reasonably characterized as deliberately misleading than as literally false statements. We think on this record that the jury’s verdict must have been based on a theory of scheming to defraud, not on a theory of making specific false statements.

Finally, McGregor’s defense theory was that he was innocent because he had no knowledge that the statements he was making were untrue; in other words, he had no intent to defraud. The jury was given a “good faith” instruction that stated, in part: “A person who acts ... on a belief or opinion honestly held is not punishable under this statute merely because the belief or opinion turns out to be inaccurate, incorrect, or wrong. An honest mistake in judgment or an honest error in management does not rise to the level of criminal conduct.” A good faith instruction is adequate to cover a defense theory that the defendant did not act with fraudulent intent. See United States v. Lawrence, 189 F.3d 838, 843 (9th Cir.1999).

The jury instructions, taken as a whole, covered both the prosecution’s theory of the case as a scheme to defraud, and McGregor’s theory that he had no knowledge of the fraud and thus no intent to defraud. We therefore affirm the district court’s decision to reject McGregor’s reasonable interpretation jury instruction.

b. Advice of Counsel Instruction;

McGregor also requested an advice of counsel instruction, arguing that he honestly believed the opinion letters written by the attorneys were accurate and that he did not understand the importance of not advertising the shelters to the general public and. of financially pre-qualifying potential investors. The district court refused the suggested jury instruction after finding that McGregor’s theory of the case in this area was adequately covered by the good faith instruction.

To qualify for an advice of counsel instruction, a defendant must demonstrate that he fully disclosed to his attorney all material facts and relied in good faith on the attorney’s recommended course of conduct. United States v. Ibarra-Alcarez, 830 F.2d 968, 973 (9th Cir.1987).

McGregor offered no evidence that he consulted any attorney, including either of the two who drafted the opinion letters, on the legal status of the shelter investments. Moreover, the two attorneys who drafted the opinion letters were not given all of the material facts by McGregor or any other defendant. Finally, McGregor did not follow the attorneys’ advice to financially pre-qualify investors and to refrain from general advertising. Thus, we find that the district court did not err in refusing the proffered advice of counsel instruction.

E. Thomson’s Remaining Claims.

1. Sufficiency of the Evidence.

Thomson argues that there was insufficient evidence to establish that he had a specific intent to defraud and that he knowingly made false statements in furtherance of the fraudulent scheme. He contends that several pieces of evidence indicate a lack of intent to defraud. For example: (1) he personally lost money on the scheme; (2) he had only minimal contact with MDA and Leroyer; (3) many MDA employees who worked closely with Leroyer did not know of the fraud; (4) none of his customers ever complained about the investments; and (5) he and his family invested heavily in the scheme, which would be irrational if he knew that it was on the verge of collapsing. Thomson further asserts that the evidence was insufficient to establish that he aided and abetted Munoz and McGregor, which resulted in convictions on the four counts pertaining solely to Munoz and McGre-gor’s customers. We conclude that the evidence was sufficient for a rational jury to conclude beyond a reasonable doubt that Thomson committed and aided and abetted mail fraud.

a. Mail Fraud.

Thomson told numerous investors that he had access to MDA’s books and records and that he monitored them regularly. He represented MDA as a financially sound company based on his review of the records. According to government exhibits, however, the records demonstrated that MDA was not performing well financially, showing losses of $2.9 million in 1989, $5.9 million in 1990, and $10.2 million in 1991. If Thomson did monitor MDA’s books, he must have known that his representations that MDA was a financially healthy company were false.

The government presented evidence that Thomson was involved in preparing misleading financial statements that showed MDA operating at a profit by combining advertising revenues with the revenues generated from the sale of the shelters. The government introduced a photocopy of an old MDA financial statement that separated the advertising revenues from the shelter sales revenues. It contained handwritten notes identified as Thomson’s suggesting that the two columns be combined in order to lead investors to believe that MDA was generating a profit. Thomson gave copies of the new, misleading financial statements to Munoz and McGregor to assist them in soliciting investors.

Several investors testified that Thomson informed them that a major accounting firm had agreed to perform an audit of MDA in the fall of 1991. In fact, Thomson knew that the firm had declined to perform the audit. When pressed by the investors for the results of the audits, Thomson showed them copies of the misleading financial statements he had prepared and attempted to pass them off as preliminary audit findings of the accounting firm.

Thomson also misrepresented the sales commissions received by him and his salesmen. Several investors testified that they were misled by the sales brochure into believing that the salesmen were only receiving $1,000 in commissions. One investor testified that Thomson directly told her that the commissions were only $500. The investors further testified that had they known that the salesmen were receiving $2,500 in commissions they would have reconsidered their investments. Even Thomson’s brother testified that he believed Thomson was only making 10% to 15% in commissions for each sale. Moreover, Thomson testified in a prior trial that the marketing and consulting fees listed in the brochure went to him, but in this trial testified that those fees did not go to him.

Thomson knew about the SEC investigation in March 1991, as did Munoz and McGregor. Yet, he continued selling the shelter investment until December 1991 without informing investors of the ongoing investigation.

Thomson also knew that the factual premises upon which the securities opinion letters were based were false. He knew that at least one of his sales agents was soliciting shelter sales through newspaper advertisements. Moreover, Thomson, like McGregor, falsified forms in order to make it look as if he had financially pre-qualified the investors to whom he sold when, in fact, he had not pre-qualified them.

Investors testified that Thomson represented to them that he had received IRS approval to use the shelter investment as an individual retirement account (IRA). At trial, he attempted to introduce into evidence a letter ostensibly sent to him from the IRS indicating that the IRA had been approved. The government demonstrated, on cross-examination of Thomson, that the letter did not belong to FAC, Thomson’s company. Rather, the letter was sent to a different company expressing approval for an unrelated pension plan. The name and employer identification number of that company had been masked, but the serial number was inadvertently left on the letter, clearly indicating that the letter belonged to the other company, not FAC. Other than this fraudulent letter purporting to grant IRS approval for FAC’s IRA, Thomson offered no evidence that he either sought or obtained IRS approval for the IRA.

Finally, several investors testified that Thomson told them he was a CPA, when, in fact, he was not. Thomson also falsely represented himself as a former IRS agent. He argues on appeal that these statements did nothing to further the fraudulent scheme. He might have made them, however, to instill confidence in prospective investors that they were making a sound, legitimate investment.

From all of this evidence, a rational jury could conclude beyond a reasonable doubt that Thomson intended to defraud the investors. Sufficient admissible evidence satisfied the elements of mail fraud.

b. Aiding and Abetting.

Numerous investors testified that Thomson informed them that he assisted Leroyer in designing and marketing the fraudulent investment scheme. Moreover, Leroyer’s assistant testified that Thomson met with Leroyer behind closed doors two to three times a week to discuss the company. Thomson also gave to Munoz and McGregor the misleading financial statements that he had prepared to assist them in selling the investment. Finally, the government offered testimony from Munoz’s brother that Thomson recruited Munoz to sell shelters for MDA, that Munoz worked as a sub-agent for Thomson from 1989 to 1990 before striking out on his own, and that Thomson received a substantial portion of Munoz’s commissions for each sale Munoz made.

From this evidence the jury could have rationally concluded beyond a reasonable doubt that Thomson had aided and abetted Munoz’s and McGregor’s fraudulent behavior.

2. Admission of Evidence of Investor Recovery.

Before trial, the government moved in limine to exclude all reference to the sale of MDA. At trial, Thomson attempted to cross-examine three victim investors on the.amount of money they actually lost on the investment, arguing that the evidence was relevant to rebut the implication that the investors had lost all of their investment money. The district court initially sustained the government’s objections to the line of questioning, but eventually allowed Thomson to introduce the evidence through the cross-examination of one government witness because the government had opened the door to the evidence on direct examination.

A trial court’s limitation of a defendant’s cross-examination of a witness is reviewed for abuse of discretion. See United States v. Bensimon, 172 F.3d 1121, 1128 (9th Cir.1999). We review de novo whether the limitation on cross-examination violated the defendant’s Sixth Amendment confrontation rights. Id. This Court makes a two-part inquiry to determine whether a defendant’s right to confrontation was violated. Wood v. Alaska, 957 F.2d 1544, 1549-50 (9th Cir.1992). First, the evidence must be relevant since a defendant does not have a constitutional right to present irrelevant evidence. Id. at 1550 . If the evidence is relevant, the court will then inquire “whether other legitimate interests outweigh the defendant’s interest in presenting the evidence.” Id. The trial court violates the defendant’s right to confrontation if it abuses its discretion in excluding the evidence. Id.

We hold that the district court did not abuse its discretion in excluding this testimony. The evidence was marginally relevant at best. At the time of trial the investors had not yet recovered any money from the sale of MDA, and the defendants offered no proof that they would, in fact, recover any of the escrowed funds. The sale of MDA took place well after reorganization in bankruptcy when the defendants were no longer operating the company. Because it was not relevant, Thomson’s Sixth Amendment rights were not violated.

Moreover, even if the evidence were relevant, Thomson was eventually allowed on cross-examination to introduce the evidence of the sale of MDA and the existence of the escrow fund. Any further testimony would have been cumulative.

3. Admission of Evidence of Investigation of MDA.

Eugene Zech, an attorney for MDA, testified at trial that he had sent a letter to all MDA salesmen in May 1990 requesting that they suspend selling the investment. On direct examination, the court allowed Zech to discuss a conversation that he had with Leroyer regarding an inquiry into the investment scheme by the State of Michigan. The court allowed the testimony over Thomson’s hearsay objection, for the limited purpose of explaining what caused Zech to write the letter.

Thomson argues that the district court erred by admitting Zech’s testimony regarding Leroyer’s statement about the Michigan investigation into MDA. He asserts that this testimony is hearsay that does not fall into an exception to the hearsay rule. We review a district court’s evidentiary rulings for abuse of discretion. See United States v. Fleming, 215 F.3d 930, 938 (9th Cir.2000). We hold that the district court did not abuse its discretion in admitting this testimony.

Rule 801 of the Federal Rules of Evidence prohibits the introduction of a statement made by an out-of-court declarant that is offered at trial to prove the truth of the matter asserted. Fed.R.Evid. 801(c). We conclude, however, that this testimony is not hearsay because it was not offered for its truth.

The district court allowed the government to question Zech only generally about the chain of events causing him to send the letters. Zech never described the nature or the result of the investigation and the government did not address the investigation in its closing argument. Zech’s testimony was not offered for its truth but to explain why Zech wrote the letters to the defendants requesting that they suspend further sales. We therefore hold that the district court did not abuse its discretion in admitting Zech’s testimony over the hearsay objection.

4. Cross-examination of Thomson.

Thomson testified at the trial in his own defense. During direct examination, he mentioned another sale-leaseback investment in which he had been involved that had failed. On cross-examination, the government asked him about the failure of a third sale-leaseback investment called Copymaid that he sold from 1984 to 1986. Thomson objected on the grounds that the evidence was too remote to be relevant, but the objection was overruled.

Rule 608(b) of the Federal Rules of Evidence allows a party to inquire on cross-examination into specific instances of a witness’s conduct provided those instances of conduct concern the witness’s character for truthfulness or untruthfulness. Evidence of prior frauds perpetrated by the witness is generally considered probative of the witness’s truthfulness, see United States v. Gay, 967 F.2d 322, 328 (9th Cir.1992), although such evidence may in some cases be too remote in time to be deemed probative of the witness’s truthfulness. See United States v. Jackson, 882 F.2d 1444, 1447 (9th Cir.1989).

Here, the evidence of the Copymaid investment was relevant to the truthfulness of Thomson’s claim that he was an unwitting participant in the bus stop shelter scheme. The terms of the Copymaid investment were similar to the MDA investment, and Copymaid collapsed after two years in business. Moreover, the evidence was not too remote to be probative of Thomson’s truthfulness. See id. at 1448 (holding that the prosecution could permissibly question a testifying attorney about his disbarment fourteen years earlier for misappropriating client funds). Thus, we hold that the district court did not abuse its discretion in admitting the impeachment evidence.

5. Jury Instructions.

Thomson argues, on the basis of our holdings in United States v. Aguilar, 80 F.3d 329 (9th Cir.1996), and United States v. McAllister, 747 F.2d 1273 (9th Cir.1984), that the district court erred in giving the jury an instruction that the defendant’s reckless indifference to the truth or falsity of a statement would be sufficient to establish specific intent under the mail fraud statute. We review de novo whether a jury instruction misstates the intent element of mail fraud as a matter of law. Gay, 967 F.2d at 326 .

This Court has repeatedly held that reckless indifference to the truth or falsity of a statement satisfies the specific intent requirement in a mail fraud case. See, e.g., id. (affirming a reckless indifference instruction in a mail fraud case on the grounds that it conforms with a long line of Ninth Circuit precedent); United States v. Schaflander, 719 F.2d 1024, 1027 (9th Cir.1983) (same); United States v. Cusino, 694 F.2d 185, 187 (9th Cir.1982) (“Fraudulent intent is shown if a representation is made with reckless indifference to its truth or falsity.”). Aguilar and McAllister concern a different type of intent instruction where a defendant is accused of trying to avoid obtaining actual knowledge of the illegality in order to avoid the potential legal consequences. These cases are therefore inapposite. We conclude that the district court properly included the reckless indifference language in the jury instruction.

6. Sentencing Adjustments.

a. Role in the Offense.

At sentencing, the district judge increased Thomson’s offense level by three levels, finding that he played a managerial or supervisory role in the criminal activity pursuant to U.S.S.G. § 3Bl.1(b). Such an adjustment is permissible if the district court finds that defendant was a “manager or supervisor and the criminal activity involved five or more participants or was otherwise extensive.... ” Id.

Thomson challenges this upward adjustment on the grounds that the district judge failed to identify those participants over whom Thomson exercised control. He relies on United States v. Luca, 183 F.3d 1018 (9th Cir.1999), for this proposition. In Luca , however, the court found no evidence in the record that anyone other than the defendant actually participated in the criminal behavior. Id. at 1024 . The district court there had concluded that it need not identify any other participants, because it found that the criminal activity was otherwise extensive. Id. This Court reversed, holding that the terms “manager” and “supervisor” imply the existence of a person or persons over whom the defendant exercises control and thus, as a matter of law, there must be some evidence of other participants in the criminal activity to justify an upward adjustment of a defendant’s sentence for being a manager/supervisor. Id.

Thomson misinterprets Luca as requiring that the district court specifically list the other participants in its factual findings in order to justify an upward adjustment on the basis of role in the offense. This Court has stated on a number of occasions that the district court need not make any specific findings of fact to support an upward adjustment for role. See, e.g., United States v. Govan, 152 F.3d 1088 (9th Cir.1998); United States v. Lopez-Sandoval, 146 F.3d 712, 716 (9th Cir. 1998); United States v. Ponce, 51 F.3d 820, 826 (9th Cir.1995).

Here, there was ample evidence of other participants in the criminal activity (for example, Thomson’s co-defendants), and there was ample evidence for the district court to conclude that Thomson played a managerial or supervisory role. Thomson repeatedly informed investors that he assisted in creating and marketing the investment. Moreover, he created and provided materials such as the financial statements to Munoz and McGregor to assist them in selling the investment. In fact, Thomson was responsible for recruiting Munoz into the fraudulent scheme. Munoz worked directly under Thomson for a time and gave Thomson a portion of his commissions. Thomson’s company, FAC, employed over twenty salespeople.

Perhaps most importantly, the jury convicted Thomson on all counts, including those relating to Munoz and McGregor’s customers. This indicates that the jury found that Thomson had exercised some measure of control and responsibility over the actions of Munoz and McGregor. Thus, we hold that the district court did not clearly err in giving Thomson an upward adjustment for his role in the fraudulent scheme.

b. Obstruction of Justice.

One of the issues at trial was whether Thomson lied to investors when he told them that he had obtained a letter from the IRS qualifying the shelter investment as an IRA. As evidence that he had told investors the truth, Thomson produced a letter purportedly granting IRS approval to FAC to sell the investment as an IRA. The letter contained no company name or address, or employer identification number, but did contain a serial number. On cross-examination the government introduced a second letter written by the IRS to a bank regarding a pension plan. This letter was the same letter as the one Thomson had introduced, but contained the name, address and EIN of the bank. The serial numbers on the two letters were identical, indicating that on the letter produced by Thomson someone had deliberately masked the other identifying information to pass off the letter as belonging to FAC.

Thomson also attempted to prove that he had purchased a shelter investment as late as November 1991 to demonstrate his continuing belief in the financial stability of MDA. He produced a sales contract that purportedly documented the November sale. The government demonstrated that there was no listing of the sale in MDA’s books and that the contract was typed in a different font on a different printer than were all of the other sales contracts. Moreover, the government introduced Thomson’s list of shelter purchases that he claimed as credits at the bankruptcy hearing. The list did not contain the November shelter purchase. Finally, Thomson’s bank account records demonstrated that the check he ostensibly used to purchase the shelter had never been cashed.

Based on these two incidents, the district court granted a two-level upward adjustment for obstruction of justice pursuant to U.S.S.G. § 3C1.1. The court found that Thomson testified falsely and produced fabricated documents to show that he received IRS approval for the shelter investment as an IRA and that he purchased a shelter investment in November 1991.

Thomson argues that the district court erred in assessing the upward adjustment for three reasons. First, he asserts that the district court did not articulate which testimony was false and which evidence was fabricated. Second, he says that the court did not make specific findings that Thomson’s false testimony and production of false evidence was willful and material. Finally, he argues that upwardly adjusting a defendant’s sentence merely because his testimony was inconsistent with the government’s theory of the case would serve to stifle defendants from testifying on their own behalf.

In its March 5, 1999, order the district court did, however, make specific findings identifying the testimony and documents that constituted obstruction of justice and it found that Thomson’s actions in testifying and producing the false evidence were both willful and material. Thus, Thomson’s first two objections to the upward adjustment are without merit.

Thomson’s argument that his testimony was merely inconsistent with the government’s theory of the case is a gross mis-characterization. He testified falsely on the basis of documents he introduced that were deliberately fabricated to make it look as if he had received IRS approval for his IRA and had purchased a shelter as late as November 1991. The district court therefore properly found by a preponderance of the evidence that Thomson had obstructed justice. Accordingly we affirm the upward adjustment.

c. Calculation of Loss.

Thomson raises the issue, as did Munoz and McGregor, of whether the district court erred in calculating loss based on the intended loss standard. For the reasons stated above, we affirm the district court’s decision to use intended loss.

Thomson also contends that the district court erred in holding him responsible for the entire loss caused by MDA. In United States v. Blitz, 151 F.3d 1002 (9th Cir.1998), this Court upheld the district court’s decision to hold telemarketers in a fraudulent telemarketing scheme responsible for the loss caused by the entire scheme, despite the fact that the telemarketers did not design or implement the scheme. Id. at 1013 . The telemarketers worked together to further the scheme, helping each other with sales, teaching each other sales pitches, and providing each other with advice on how to sell the scheme. Id. This Court held that the loss was foreseeable to the telemarketers and they should be held responsible for it. Id.

Thomson argues that Blitz is distinguishable because the telemarketers were not competing with one another for customers, as were Thomson, Munoz, and McGregor. The evidence in this case indicates, however, that the defendants were all working together in the scheme more than they were working in competition with each other. The evidence showed that Thomson was responsible, along with Leroyer, for creating the sale-leaseback scheme and the marketing plan including the misleading sales brochure. In addition, he provided Munoz and McGregor and the other salesmen with financial statements to assist them in making sales.

Perhaps most importantly, Thomson was convicted on all ten counts, including those counts relating to Munoz and McGregor’s customers, indicating that the jury found that Thomson assisted Munoz and McGre-gor in perpetrating the fraud. From the evidence adduced at trial and the jury convictions for those counts relating to Munoz and McGregor’s customers, the district court could reasonably have found that Thomson provided other salesmen with sufficient assistance to justify holding him responsible for the loss that they caused. We therefore hold that the district court did not err in including all loss caused by MDA in its loss calculation for Thomson’s sentencing.

III. CONCLUSION

We commend the district court for the manner in which it conducted this difficult, lengthy trial in a major white collar criminal case with multiple defendants. We find that the district court erred only in using the old preponderance of the evidence standard rather than the new clear and convincing evidence standard in determining the relevant conduct of David Munoz and Bennie McGregor where the resulting offense level was extremely disproportionate. We therefore vacate their sentences and remand to the district court for resentencing. The district court should apply the heightened standard of clear and convincing evidence to the evidence adduced at trial. We affirm the convictions of Munoz, McGregor, and Thomson, as well as Thomson’s sentence, and we affirm the district court’s decisions on all other issues raised on appeal by the defendants.

AFFIRMED in part, VACATED & REMANDED in part.

. Thomson, unlike Munoz and McGregor, did not request an offset at sentencing for the amount of lease payments made to the investors. . In limited circumstances courts have stated that the privilege may apply where a party reasonably but mistakenly believes that an attorney represents him rather than another party. See, e.g., In re Grand Jury Subpoena Duces Tecum, 112 F.3d 910, 923 (8th Cir. 1997) (acknowledging, in dicta, that a party’s reasonable belief about the nature of his relationship with an attorney may be relevant to the determination of whether the privilege applies); United States v. Hart, No.Crim.A.92-219, 1992 WL 348425 , at *l-*2 (E.D.La. Nov. 16, 1992). This Court has never addressed this issue. On appeal Munoz urges us to adopt such a rule but we decline to do so in this case. Munoz, who was represented by a different attorney at trial than on appeal, did not directly make this argument at the district court level and, as a result, the record is insufficient for us to decide the issue. . The Rules provide for several exceptions none of which was applicable here. The government argues that the co-conspirator exception to the hearsay rule should apply. See Fed.R.Evid. 801(d)(2)(E). The district court never made a preliminary finding that this rule should apply, however, so this argument is unavailing. See Bourjaily v. United States, 483 U.S. 171, 175 , 107 S.Ct. 2775 , 97 L.Ed.2d 144 (1987).