On July 22, 1998, this court affirmed the district court's decision to deny Leonard Chavin a discharge in bankruptcy. In re Chavin, 150 F.3d 726, 729 (7th Cir.1998). In our opinion, we noted an appearance of deliberate fraud in Chavin's bankruptcy filings. Accordingly, we referred the case to the Department of Justice for further investigation. Id. Eventually, a grand jury returned a fifteen-count superseding indictment charging Chavin and his attorney, Martin Litwin, with tax and bankruptcy fraud. Following a lengthy trial, a jury found them guilty on eleven of the fifteen counts, and the district court sentenced Chavin to 37 months in prison and Litwin to 33 months in prison. On appeal, they raise multiple issues, including various claims related to the district court's application of the federal sentencing guidelines. We reject all of Chavin's and Litwin's arguments and affirm the district court's disposition.

Leonard Chavin, a businessman and investor, owned a block of commercial real estate on South Halsted in Chicago ("the Halsted property"). He also controlled a corporation known as SCV Corporation ("SCV"), which held as its only asset a clothing store that did business under the name of Howard's Style Shop ("HSS").1 The business struggled, and over a period of years Chavin personally lent SCV in excess of $900,000. In 1992, Chavin sold part of the Halsted property for $975,000, generating a substantial capital gain and greatly increasing his tax burden.2 Chavin and his attorney, Litwin, devised a plan to offset the Halsted-property gain with the SCV-debt loss.

They discovered that the tax code allows an individual to take a deduction for worthless, unrecoverable debts, known as a "bad debt loss." 26 U.S.C. § 166(d) (2002). The only problem was that they had to somehow make the SCV debt worthless to Chavin. They did so by manufacturing a sham sale of SCV's only asset, the HSS clothing store, to Litwin's cousin, Michael Glickman. Glickman paid nothing out of his own pocket for HSS. He was told that the sale was nothing more than a "paper transaction"—meaning he would have no connection with the clothing store except that on paper he would be listed as the owner. Following the "sale," Chavin continued to control HSS and nothing about the business changed. And because SCV no longer had any assets of record, Chavin could claim that the debt SCV owed him was entirely worthless and thereby claim it as a bad debt loss. On his 1992 tax return, Chavin reported a capital gain from the sale of the Halsted property of $328,000. Against this, Chavin deducted $900,000 as a bad debt loss based on the SCV debt. In 1992, Chavin could only take the loss to the extent that it offset the capital gain from the Halsted-property sale plus $3000—in total $331,000; thus, he carried over the remaining portions of the loss on his returns for subsequent years.

After being alerted to possible fraud in Chavin's financial dealings, the government investigated and a federal grand jury returned a fifteen-count superseding indictment. The first six counts of the indictment related to the tax dealings just described. Count 1 charged Chavin and Litwin with conspiring to defraud the Internal Revenue Service ("IRS") through the creation of a fraudulent $900,000 bad debt loss in violation of 18 U.S.C. § 371. Count 2 charged Chavin with violating 26 U.S.C. § 7206(1) by making false statements on his 1992 tax return, in that he understated the sale price of the Halsted property and in that he claimed a $900,000 bad debt loss to which he knew he was not entitled. Counts 3 and 5 charged Chavin with making false statements on his 1993 and 1994 tax returns, respectively, in that he reported a false carryover loss from his 1992 return in violation of 26 U.S.C. § 7206(1). Similarly, Counts 4 and 6 charged Litwin with violating 26 U.S.C. § 7206(2) for aiding and assisting Chavin in the preparation and presentation of his 1993 and 1994 fraudulent tax returns. Following a trial, the jury convicted Chavin and Litwin on all of these tax counts.

The remaining charges in the superseding indictment, Counts 7 through 15, related to a bankruptcy fraud that developed when Chavin's creditors forced him into bankruptcy on December 30, 1994. As part of the bankruptcy proceeding, Chavin was required to file schedules disclosing all of his assets, income, and expenses. On the schedules he filed, he failed to disclose certain assets and income, including the interest he had in HSS. Also, as part of the bankruptcy proceeding, Glickman was deposed about his purported ownership of HSS. Litwin counseled Glickman on his responses so as to make it appear that Glickman owned HSS and that Chavin had no interest in the company.

The relevant portion of Count 7 charged Litwin with violating 18 U.S.C. § 1623 for aiding Glickman to commit perjury in the bankruptcy proceedings.3 Litwin was found guilty on this count. Counts 8 through 13 charged Chavin with concealing various assets from the bankruptcy trustee and his creditors in violation of 18 U.S.C. § 152(1). Of these counts, the jury convicted Chavin on Counts 10 and 11, which both related to life insurance polices. Chavin was convicted on Count 14, which charged him with failing to disclose assets on his bankruptcy filings in violation of 18 U.S.C. § 152(3). Finally, the jury convicted Chavin on Count 15, which charged him with converting an asset in bankruptcy in violation of 18 U.S.C. § 152(7).

Following a sentencing hearing, the district court determined that the appropriate sentencing level under the U.S. Sentencing Guidelines was 19 for both defendants, which corresponds to a sentencing range between 30 and 37 months. The district court sentenced Chavin to 37 months imprisonment and fined him $10,000 and sentenced Litwin to 33 months imprisonment and fined him $50,000.

Chavin and Litwin argued as part of their defense to the tax-fraud charges that they relied in good faith on the advice of Chavin's accountant Kessler and therefore did not "willfully" perpetrate the tax fraud. Defendants tendered a theory-of-defense jury instruction that covered this point. The district court, however, refused to give the instruction to the jury, stating that the theory of defense was sufficiently covered by other instructions; specifically, by the pattern instruction on good faith and a further instruction that made clear that defendants must have acted "willfully" within the precise definition of that term.

We review a district court's refusal to give a theory-of-defense instruction de novo. United States v. Meyer, 157 F.3d 1067, 1074 (7th Cir.1998). To be entitled to a theory of defense instruction, a defendant must satisfy a four-part test by showing (1) that the proposed instruction is a correct statement of law; (2) that the evidence in the case supports the theory of defense; (3) that the theory of defense is not already part of the charge; and (4) that failure to include the proposed instruction would deny the defendant a fair trial. Id.

Here, defendants have, at the very least, failed to satisfy the third element. We have recognized in prior tax-evasion cases that a "`good faith reliance' defense is essentially a claim that the [defendant] did not act `willfully.'" United States v. Brimberry, 961 F.2d 1286, 1291 (7th Cir. 1992). Consequently, when, as here, instructions are given that require the jury to find that a defendant acted "willfully" and those instructions define "willfulness" and "good faith" as mutually exclusive, then a further "good faith reliance" theory-of-defense instruction would be unnecessarily redundant. Id.; United States v. Kelley, 864 F.2d 569, 573 (7th Cir.1989). Therefore, we find that the instructions given by the district court adequately covered the defendants' theory of defense.

2. Jury Instruction on Chavin's Signing of His Tax Return

Chavin further argues on appeal that the instruction the district court gave about the effect of his signature on his 1992 tax return could likely have led the jury to believe that his signature alone overrode his "reliance on a tax professional" defense. Chavin offered an alternative instruction based on language from United States v. White, 879 F.2d 1509, 1511 (7th Cir.1989). Ultimately, the district court rejected this instruction and instead used an instruction based on other language from the White opinion—language that was later cited with approval in United States v. Ellis, 50 F.3d 419, 425 (7th Cir. 1995). Both instructions were accurate statements of the law and did not substantially differ from one another. In this situation, we review a district court's decision on the precise wording of the instruction for abuse of discretion. Riemer v. Ill. Dep't of Transp., 148 F.3d 800, 804 (7th Cir.1998). The instruction given plainly stated that the determination of whether Chavin acted knowingly should be based on the "totality of the circumstances, including but not limited to the bare fact of the defendant's signature on various documents." (Tr. 2787) (emphasis added). We find that it was not an abuse of discretion for the district court to conclude that this language would not mislead the jury into thinking that Chavin's signature alone defeated his "reliance on a tax professional" defense.

3. Restriction of Direct Examination of Defense Expert Sharps

At trial, Chavin sought to have defense expert Jason Sharps testify during direct examination that Chavin's accountant, Kessler, had made various "mistakes" on Chavin's tax return. The district court permitted inquiry into the ways that Sharps would have prepared the return differently; however, the court did not allow Sharps to testify that where he and Kessler differed, Kessler had made a "mistake." We review a district court's evidentiary rulings for abuse of discretion. United States v. Smith, 230 F.3d 300, 307 (7th Cir.2000). Chavin argues that the testimony that Kessler had made mistakes was relevant to the issue of whether Chavin relied on his accountant to prepare correct income tax returns and to the calculation of tax loss. We do not perceive how this evidence could be relevant to either of these issues, and Chavin provides no explanation as to why it is relevant. Therefore, we find that the district court did not abuse its discretion in refusing to allow this evidence.

Before trial, defendant Litwin, an attorney specializing in tax, real estate, and bankruptcy law, filed a motion to serve as co-counsel in his defense. The district court denied the motion, and Litwin raises the issue again on appeal.

There is no question that a defendant has a constitutional right to conduct his own defense. But seeking to serve as co-counsel in a "hybrid representation" is a separate question. United States v. Oakey, 853 F.2d 551, 553 (7th Cir.1988). We recognize, along with all other circuits that have considered the question, that there is no Sixth Amendment right to hybrid representation; rather, whether a defendant may act as co-counsel along with his own attorney, is a matter within the discretion of the district court. United States v. Tutino, 883 F.2d 1125, 1141 (2d Cir.1989) ("The decision to grant or deny `hybrid representation' lies solely within the discretion of the trial court."); United States v. Norris, 780 F.2d 1207, 1211 (5th Cir. 1986) ("[A] defendant does not have the right to a hybrid representation, in which he conducts a portion of the trial and counsel conducts the balance."); United States v. Halbert, 640 F.2d 1000, 1009 (9th Cir.1981) ("A criminal defendant does not have an absolute right to both self-representation and the assistance of counsel."); see also United States v. Singleton, 107 F.3d 1091, 1101 n. 7 (4th Cir.1997) ("The cases reiterating the principle that courts are not required to allow defendants to split the responsibilities of the representation with an attorney are myriad.").

This court has held that hybrid representation is "disfavored." United States v. Kosmel, 272 F.3d 501, 506 (7th Cir.2001); Cain v. Peters, 972 F.2d 748, 750 (7th Cir.1992); United States v. Oakey, 853 F.2d 551, 553 (7th Cir.1988). The primary reason we have expressed concern about hybrid representation is because "it allows a defendant to address the jury, in his capacity as counsel, without being cross-examined, in his capacity as a defendant." Oakey, 853 F.2d at 553. Litwin contends that his assurances, given before trial, that he would testify if permitted to serve as co-counsel neutralized the potential problems with hybrid representation. This sort of pre-trial promise, however, does not fully allay the concerns. If, after the trial began, he were to go back on his promise and decide not to testify, there would be no effective way for the trial court then to compel him to take the stand. Consequently, his promise is without effect. Further, the fact that Litwin is an attorney with experience in the matters involved in this case does not in any way ease concerns that he could make unsworn statements to the jury without being subject to cross-examination. Therefore, we find that the district court properly refused to allow Litwin to serve as co-counsel in his defense.

5. Sufficiency of Evidence: Glickman's Testimony

Litwin challenges two aspects of the verdict against him on grounds of insufficient evidence; the first is whether the testimony of his cousin was sufficient to convict him. We will reverse based on insufficiency of evidence only if, viewing the evidence in the light most favorable to the government, the record contains no evidence from which the jury could find guilt beyond a reasonable doubt. United States v. Combs, 222 F.3d 353, 362 (7th Cir.2000).

Litwin argued in a motion for a judgment of acquittal that the testimony of his cousin, Michael Glickman, was inherently incredible. The district court denied the motion and we affirm. As we have stated in previous cases, "it is not the role of the appellate court to question the jury's credibility determinations." United States v. Stott, 245 F.3d 890, 898 (7th Cir.2001); see also United States v. Durham, 211 F.3d 437, 445 (7th Cir.2000); United States v. Alcantar, 83 F.3d 185, 189 (7th Cir. 1996). Therefore, we will reverse a jury verdict only if the testimony was incredible as a matter of law. Stott, 245 F.3d at 898. Litwin points primarily to evidence that Glickman had previously lied to or cheated various people, including his own family members. We do not find that this evidence, or any of the other evidence that Litwin has put before us, meets his substantial burden of showing that Glickman's testimony was incredible as a matter of law.

6. Sufficiency of Evidence: Conviction on Counts 4 and 6

Counts 4 and 6 of the indictment charged Litwin with assisting Chavin in the presentation of fraudulent tax returns for 1993 and 1994—the years in which Chavin carried over the fraudulent bad debt loss. Litwin contends that the guilty verdict on these counts must be reversed because he did not aid Chavin in the preparation of these returns, nor, he claims, was it foreseeable to him that Chavin would make false statements on these returns. We disagree. Litwin knew that Chavin would have a large bad-debt loss in 1992 because he had helped Chavin create the loss. And based on Litwin's experience in tax matters, it was reasonable for the jury to conclude that he would have known that the entire loss could not have been deducted in 1992—that some of it would have to be carried over to subsequent years. Therefore, we find that there was sufficient evidence to support his conviction on Counts 4 and 6.

Chavin and Litwin also challenge various aspects of their sentences. We review a sentencing court's factual determinations for clear error and interpretations of the guidelines de novo. United States v. Owolabi, 69 F.3d 156, 162 (7th Cir.1995).

Chavin challenges the district court's refusal to group the tax and bankruptcy counts together as closely related under § 3D1.2(d) of the sentencing guidelines. Had the district court grouped these counts, Chavin would have received a lower sentence. Guideline § 3D1.2 provides:

All counts involving substantially the same harm shall be grouped together into a single Group. Counts involve substantially the same harm within the meaning of this rule:

...

(d) When the offense level is determined largely on the basis of the total amount of harm or loss, the quantity of a substance involved, or some other measure of aggregate harm, or if the offense behavior is ongoing or continuous in nature and the offense guideline is written to cover such behavior.

Offenses covered by the following guidelines are to be grouped under this subsection:

[Several guideline provisions then are listed, including § 2F1.1 and § 2T1.1, the offenses for which Chavin was convicted.]

U.S. Sentencing Guidelines Manual ("U.S.S.G.") § 3D1.2 (2000).

Chavin first contends that since he was convicted of offenses listed in the "to be grouped" category the district court had no choice but to group the offenses because listed offenses must be grouped automatically. Because such a conclusion flies in the face of the overriding purpose of § 3D1.2, we disagree, and as we note below, we are not alone in this determination. See, e.g., United States v. Williams, 154 F.3d 655, 657 (6th Cir.1998) ("The bulk of the courts to have considered the proper construction of subsection (d) have concluded that there is no automatic grouping of counts simply because those counts are on the `are to be grouped' list.").

In United States v. Wilson, we noted that the primary goal of § 3D1.2 "is to combine offenses involving closely related counts." 98 F.3d 281, 282 (7th Cir.1996) (quotations omitted). In cases, like this one, where two different offense guidelines are at issue, automatic grouping would often lead to the grouping of entirely unrelated counts. See United States v. Harper, 972 F.2d 321, 322 (11th Cir.1992) ("In some circumstances, automatic grouping detracts from the main purpose of section 3D1.2; to combine offenses involving closely related counts."). In light of the purpose of § 3D1.2, we seriously doubt that its drafters would have desired such an outcome. In fact, the commentary to § 3D1.2 makes it clear that they did not: "[c]ounts involving offenses to which different offense guidelines apply are grouped together under subsection (d) if the offenses are of the same general type and otherwise meet the criteria for grouping under this subsection." U.S.S.G. § 3D1.2 cmt. n. 6 (emphasis added). Consequently, at least in cases where different offense guidelines apply—as is the case here, where the offenses are under § 2F1.1 and § 2T1.1—grouping is not automatic; rather the sentencing court is required to make an additional inquiry into whether the offenses are "of the same general type."

In support of his argument for automatic grouping, Chavin points to cases in the Second and Ninth Circuits that hold that grouping is required when offenses are listed on subsection (d)'s "to be grouped" list. We see an important distinction, however, between those cases and the case before us. In both of those cases, the defendants sought to have charges grouped that were covered by the same offense guideline. United States v. Gelzer, 50 F.3d 1133, 1144-45 (2d Cir.1995) (counts that the court held were required to be grouped were possession of a firearm with an obliterated serial number and possession of a firearm by a felon, both offenses to which § 2K2.1 applied); United States v. Buenrostro-Torres, 24 F.3d 1173, 1176 (9th Cir.1994) ("If the offenses at issue are covered by one of the listed guidelines we do not analyze the facts of the particular case. Here, the offenses ... are covered by § 2F1.1.") (emphasis added). Consequently, the courts in those cases did not need to consider the commentary language about situations where the offenses at issue are covered by different offense guidelines.4

In fact, when the Second Circuit was faced with the issue of whether to group counts that were covered by different guidelines— § 2F1.1 (fraud) and § 2S1.1 (money laundering), both of which were on the "to be grouped" list—they did not find that automatic grouping was appropriate, stating instead that the two counts "should only be grouped if the counts are also of the same general type." United States v. Napoli, 179 F.3d 1, 10 (2d Cir.1999). Ultimately, the court found that the two offenses involved there should not be grouped. Id. at 10-11. Therefore, the Second Circuit clearly does not view that grouping is always automatic simply because the offenses are listed. The Second Circuit is not alone in this finding. At least four other circuits have reached the same conclusion. See Williams, 154 F.3d at 657 (noting that the courts that have considered the proper construction of subsection (d) have held that there is no automatic grouping); United States v. Walker, 112 F.3d 163, 166-67 (4th Cir.1997) (stating that when considering whether to group mail fraud (§ 2F1.1) and money laundering (§ 2S1.1) "[s]ubsection (d) expressly permits the grouping of offenses under the fraud and money laundering guidelines, but any grouped offenses must be `closely related'"); United States v. Seligsohn, 981 F.2d 1418, 1425 (3d Cir. 1992) ("[A]lthough all of the counts are listed in subsection (d) as appropriate for grouping, that inclusion does not mean that the counts must be grouped. Counts must be of the `same general type' before grouping is appropriate.") (citations omitted); United States v. Harper, 972 F.2d 321, 322 (11th Cir.1992) ("[G]rouping is not automatic, even if all offenses in question are encompassed within this [`to be grouped'] category."). We join these circuits in finding that grouping under subsection (d) is not automatic simply because the offenses at issue are listed on the "to be grouped" list.

Although we reject automatic grouping, the counts in Chavin's case could still be grouped if the offenses are "of the same general type and otherwise meet the criteria for grouping under ... subsection [(d)]." U.S.S.G. § 3D1.2 cmt. n. 6. Chavin's offenses, however, are not "of the same general type." We agree with the district court's reasoning as to why the tax and bankruptcy counts are not of the same general type:

[T]he victims are different, and the time frame and context do not completely overlap. Mr. Chavin's effort to cheat his creditors simply does not involve "substantially the same harm" as does his effort to cheat the government.

(R.173.)

Defendants argue at some length that the district court erred because it based its finding that the offenses were not of the same type partially on the fact that the victim of the bankruptcy fraud was different from the victim of the tax evasion. According to Chavin, while identity of victims is required for application of subsections (a) and (b) of § 3D1.2, it is irrelevant to subsections (c) and (d). While we agree that identity of victims is not required for grouping under subsection (c) or (d), the notion that identity of victims is irrelevant is in conflict with precedent in this circuit and with the guideline commentary. In United States v. Wilson, in considering whether to group two counts under subsection (d), we stated "[w]hether the offenses involve different victims is, as the background commentary notes, `a primary consideration' in the grouping decision." 98 F.3d at 283 (citing U.S.S.G. § 3D1.2 cmt. background). In fact, our decision to group the two offenses involved in Wilson was based largely on the fact that the victim was essentially the same. Id. at 283 (rejecting an argument that mail fraud and money laundering involve different victims in favor of a more intuitive argument that the victims are in reality the same). Consequently, the district court properly considered the fact that the victim of the tax fraud was the United States government and the victims of the bankruptcy fraud were Chavin's creditors as a factor in determining whether these offenses were of the same general type.

Even aside from there being different victims involved, there are other considerations that lead us to the conclusion that Chavin's crimes are not of the same general type. The tax and bankruptcy frauds were not "part of a single continuous course of criminal activity." United States v. Petrillo, 237 F.3d 119, 125 (2d Cir.2000). Indeed, the two offenses are quite separate. The core of Chavin's tax fraud was committed in 1992 when he made the sham sale to Glickman and filed his fraudulent tax return for that year.5 The core of the bankruptcy fraud did not occur until December 1994. Further, the bankruptcy was not even initiated by Chavin, but rather by his creditors, so the idea that it was part of his criminal plan seems implausible. Also, as the government notes in its brief, the harms involved in these offenses were quite different: "the financial loss to bankruptcy creditors is not `of the same general type' as the harm implicated by a tax offense, which is a crime against the polity and which has indirect effects on all taxpayers and on the government's ability to protect and serve the public at large." (Gov. Brief at 37.) Finally, as the district court correctly noted at the sentencing hearing, the only real connection between the bankruptcy fraud and tax fraud appears to be the testimony offered by Glickman in the bankruptcy proceedings. Even construing "of the same general type" broadly, as the application notes suggest we should, we do not find that this minimal connection between the offenses requires grouping in this case. U.S.S.G. § 3D1.2 cmt. n. 6.

Litwin argues that the district court erred by not grouping the aiding and abetting perjury count with the tax evasion charges against him. We agree with the district court and reject Litwin's claim. Litwin bases his argument on § 3D1.2(c), which provides that counts involve substantially the same harm "[w]hen one of the counts embodies conduct that is treated as a specific offense characteristic in, or other adjustment to, the guidelines applicable to another of the counts." U.S.S.G. § 3D1.2(c). Litwin's sentence on the tax offense was enhanced two levels for sophisticated concealment. Litwin argues that the perjury charge was used as part of the basis for the sophisticated-concealment enhancement of the tax sentence; thus, the perjury count should have been grouped with the tax offense under subsection (c) because it was used as "an adjustment to" the tax sentence.

We begin by noting that the commentary specifically states that subsection (c) was intended to apply only to counts that are "closely related." Id. § 3D1.2 cmt. n. 5. While we recognize that Glickman's perjury did to some extent aid in the perpetuation of the tax conspiracy: the ostensible purpose of the perjury was to keep Chavin's assets hidden from his creditors. Consequently, we doubt that these counts are closely related.6

However, the reason we reject Litwin's argument for grouping is that his situation simply does not fit within the primary purpose of subsection (c), which is to "prevent `double counting' of offense behavior." Id. § 3D1.2 cmt. n. 5. Refusal to group the perjury offense and the tax conspiracy offense did not result in double counting here because, as was brought out at oral argument, Litwin would have received the same enhancement for sophisticated concealment even if the perjury charge was not considered at all. In other words, other evidence besides the perjury offense provided ample support for the sophisticated-concealment enhancement. Indeed, the district court appears to have concluded that the perjury actually did not provide any support to the sophisticated-concealment enhancement because the purpose of the perjury was not to cover up the tax crimes.7 Consequently, it cannot be said that the perjury charge was double counted because it was not counted at all under the sophisticated-concealment enhancement.

3. Tax-Loss Calculation: Unclaimed Deductions

Chavin and Litwin argue that they were improperly sentenced because the tax-loss figure upon which the district court based its sentence was inflated for two reasons: First, they contend that the district court should have reduced the tax-loss figure by the amount of legitimate but unclaimed deductions on Chavin's tax return. Second, they maintain that part of the bad debt loss Chavin took was legitimate and should not have been considered in determining tax loss.

We start with the unclaimed-deductions issue. Chavin and Litwin were sentenced on the tax counts under § 2T1.1 of the guidelines. Assessing the proper sentence under § 2T1.1 requires a determination of the "tax loss" resulting from the offense. Section 2T1.1(c)(1) provides the definition of tax loss applicable to this case: "If the offense involved tax evasion or a fraudulent or false return, statement, or other document, the tax loss is the total amount of loss that was the object of the offense (i.e., the loss that would have resulted had the offense been successfully completed)." U.S.S.G. § 2T1.1(c)(1). Also, notes (A) and (C) under this subsection provide a further explanation of how to calculate tax loss in certain circumstances:

(A) If the offense involved filing a tax return in which gross income was underreported, the tax loss shall be treated as equal to 28% of the unreported gross income ... unless a more accurate determination of the tax loss can be made.

...

(C) If the offense involved improperly claiming a deduction to provide a basis for tax evasion in the future, the tax loss shall be treated as equal to 28% of the amount of the improperly claimed deduction... unless a more accurate determination of the tax loss can be made.

U.S.S.G. § 2T1.1(c)(1)(A) & (C). Chavin and Litwin assert that if the district court had considered all of the legitimate deductions that Chavin mistakenly failed to claim on his 1992 return then the tax-loss figure would have been reduced from $199,000 to $57,352, which would have corresponded to a lower sentencing level.

The issue of whether to allow consideration of legitimate but unclaimed deductions in the determination of tax loss hinges on the definition of "tax loss." On the one hand, Chavin and Litwin argue, in essence, that "tax loss" refers to the actual amount of loss suffered by the government as a result of the tax scheme. If we accept this interpretation, then unclaimed deductions should be considered because the government did not actually lose as much money as the scheme intended since the unclaimed deductions serve to offset the amount that the defendants attempted to evade. On the other hand, the government contends that "tax loss" refers to the amount of loss that the defendant attempted or intended to create through his tax offense. If we accept this interpretation, then unclaimed deductions should not be taken into account because they have no relevance to the amount of loss that the scheme attempted to produce.

It is apparent from the definition of "tax loss" provided in the guidelines that the government has the correct position. The guidelines state that "tax loss is the total amount of loss that was the object of the offense." Id. § 2T1.1(c)(1). We take the phrase "the object of the offense" to mean that the attempted or intended loss, rather than the actual loss to the government, is the proper basis of the tax-loss figure. Here, the object of Chavin's offense was the amount by which he underreported and fraudulently stated his tax liability on his return; reference to other unrelated mistakes on the return such as unclaimed deductions tells us nothing about the amount of loss to the government that his scheme intended to create.

Chavin and Litwin argue that we should accept their interpretation because it is less subjective than the government's interpretation. They contend that trying to determine the loss that the defendant intended to create is simply too speculative. While in principle it is true that trying to assess intentions is often a subjective endeavor, we strongly disagree that the government's view leads to more subjective results. Here, the defendants' intention is embodied in the tax return that was filed with the IRS. We need to look no further than that return to find the tax-loss figure under the government's view. In fact, it is the defendants' position that would insert subjectivity into the calculation because it would require us to recreate a "perfect" tax return, taking into account all the legitimate unclaimed deductions, which would undoubtedly engender a great deal of dispute between the parties over which deductions were legitimate and which were not.

Defendants further argue that the 1993 changes in the commentary to the tax-fraud sections of the guidelines mandate courts to consider unclaimed deductions. Before 1993, § 2T1.1(a) defined "tax loss" as the greater of (1) the total amount of tax that the taxpayer evaded or attempted to evade; and (2) 28% of the amount by which the greater of gross income and taxable income was understated.8 U.S.S.G. §§ 2T1.1(a) & 2T1.3(a)(2) (1992). The commentary to this section provided that the "alternative standard [the 28% rate] may be easier to determine, and should make irrelevant the issue of whether the taxpayer was entitled to offsetting adjustments that he failed to claim." Id. § 2T1.1 cmt. n. 4. When the Guidelines were amended in 1993, this note was deleted. Defendants contend that this deletion means that now courts must consider offsetting adjustments.

Defendants, however, do not acknowledge that the entire definition of tax loss was changed by the 1993 amendments. The government offers an equally plausible reason as to why the commentary about offsetting adjustments was deleted, that takes account of the new "tax loss" definition. The government contends that the language was deleted because the new tax-loss definition specifically excludes consideration of unclaimed deductions on its face by defining tax loss as the "object of the offense." Consequently, the explanatory note was no longer required. We agree with the government that the current definition of tax loss appears to exclude consideration of unclaimed deductions, and we refuse to ignore this definition based on the fact that certain language was deleted from the commentary notes in 1993.

Finally, defendants base their argument in favor of the consideration of unclaimed deductions on the language in § 2T1.1(c)(1)(A) and (C), which states "the tax loss shall be treated as equal to 28% of the amount of the unreported gross income... unless a more accurate determination of the tax loss can be made." U.S.S.G. § 2T1.1(c)(1)(A) & (C) (2000) (emphasis added). Chavin and Litwin contend that this provision plainly requires the sentencing court to recalculate the defendant's tax return considering the unclaimed deductions if such consideration would result in a "more accurate determination" of tax loss. Defendants point out that in dicta the Second Circuit in United States v. Martinez-Rios stated that the phrase "a more accurate determination" allows a defendant to have legitimate but unclaimed deductions considered. 143 F.3d 662, 671 (2d Cir.1998).9 We note that the Tenth Circuit has questioned the Second Circuit's interpretation, stating "[w]e do not interpret this provision as giving taxpayers a second opportunity to claim deductions after having been convicted of tax fraud." United States v. Spencer, 178 F.3d 1365, 1368 (10th Cir.1999). The Tenth Circuit further explained that the phrase "a more accurate determination" simply allows the defendant or the government to argue for a rate that is "more accurate" than the 28% presumptive rate. Id. at 1368. It is our view that the Tenth Circuit interpreted the guideline correctly.10

Given the definition of tax loss discussed above, we simply cannot read the phrase "a more accurate determination" to allow for consideration of unclaimed deductions. Considering unclaimed deductions is outside the purview of what we are trying to accomplish in tax-loss calculations, which as the Tenth Circuit stated, is to "merely assess[ ] the tax loss resulting from the manner in which the defendant chose to complete his income tax returns." Id. at 1368. It is simply not the role of this court to consider other hypothetical ways that the defendant could have completed his return. United States v. Wu, 81 F.3d 72, 75 (7th Cir.1996).11

4. Tax-Loss Calculation: the $339,000 Loan Payment

Finally, Chavin and Litwin argue that $339,000 of the claimed $900,000 bad debt loss was in fact a legitimate deduction, and should not have been included in the tax-loss figure upon which their sentence was based. In 1992, Chavin made a $339,000 payment to Cole Taylor Bank as a guarantor of a loan the bank had made to SCV. This payment, as reflected in SCV's financial statements, created a $339,000 debt on the part of the company to Chavin, adding to SCV's already substantial indebtedness to him. This $339,000 debt was part of the $900,000 bad debt loss that Chavin claimed on his 1992 tax return. The defendants argued at trial that the $339,000 debt was legitimately deducted. The jury returned a general verdict finding defendants guilty on all tax counts. At sentencing, the defendants argued that while the jury convicted them of having created a fraudulent bad debt loss, the jury did not specify the amount of the loss that was illegitimate. Thus, they maintained that it was still an open question as to whether Chavin was entitled to the $339,000 deduction. The district court rejected the argument, finding that the jury had determined that Chavin improperly took a bad debt loss of $900,000 and was not entitled to any portion of it. Consequently, the district court sentenced defendants based on a $900,000 tax loss figure, noting that to make a finding that the tax-loss figure was lower than $900,000 would contradict the jury's verdict.

On appeal, Chavin again argues that the jury did not decide the issue of whether the $339,000 deduction was legitimate, and that because the verdict did not necessarily resolve the issue, the district court erred by not making its own finding on the issue. We agree with the district court's conclusion that the jury had already preclusively decided the issue and that the sentencing court could not sentence defendants based on an amount different from what the jury decided. Defendants were convicted on all the tax offenses charged— Counts 1 through 6. All of these counts reference $900,000 as the bad-debt-loss figure, and make no mention of or distinction between the various elements that comprised the total debt. Defendants' attempt to cut the total debt into pieces and argue about the individual parts at sentencing was, as the district court found, unavailing.

In any event, the record reveals that Chavin was not entitled to the deduction for which he argued. Chavin argues that the $339,000 was deductible because SCV was insolvent prior to and independent of the sham sale to Glickman and because he had personally guaranteed the loan he had paid on SCV's behalf. These factors alone, however, do not trigger the ability to take a deduction. Section 166(d) of the tax code, the section under which Chavin attempted to take the deduction on his 1992 return, permits individuals to take deductions for nonbusiness debts only when those debts become "worthless." 26 U.S.C. § 166(d)(1)(B) (2000). As this court has noted, "[t]he criterion of worthlessness is interpreted strictly: the deduction is unavailable if even a modest fraction of the debt can be recovered." Buchanan v. United States, 87 F.3d 197, 198 (7th Cir. 1996). At the end of 1992, SCV, though insolvent, still had assets worth $839,000— against liabilities of $1,240,000. Given this amount of assets, it was reasonable for the jury to have found that Chavin had at least a "reasonable prospect of recovering" a "nontrivial" amount of the money SCV owed him. Id. at 200.

As the government points out, the real triggering event that made the debt worthless and thereby "entitled" Chavin to the bad-debt-loss deduction was the sham sale to Glickman, which drained SCV of all its assets and ensured its inability to repay him. Indeed, as even the defense's tax expert noted at trial, Chavin's tax return itself showed that the sale to Glickman, and nothing else, was the triggering event. Because the event that triggered the availability of the deduction was fraudulent, the deduction itself was illegitimate. Therefore, Chavin was not entitled to any portion of the $900,000 bad debt loss, not even the $339,000 portion related to the loan payment.

For the foregoing reasons, we AFFIRM the sentences imposed by the district court on Litwin and Chavin.