This sentencing appeal primarily concerns a loss calculation under the provision of the Sentencing Guidelines governing an “intended loss” for fraud offenses. See U.S.S.G. § 2F1.1 (1997). The appeal also presents a challenge to an enhancement for offenses committed while released on bail. See id. § 2J1.7. Gary Confredo appeals from the June 29, 2006, amended judgment of the District Court for the Southern District of New York (Leonard B. Sand, District Judge) sentencing him to imprisonment for 205 months following his plea of guilty to various offenses involving fraudulent loan applications. We remand for reconsideration of the intended loss amount.
Background
Criminal conduct and guilty plea. Doing business through an entity called Granite Financial Services, Confredo and his associates coordinated the submission of more than 200 fraudulent loan applications to New York City area banks, including approximately 100 applications seeking in excess of $21 million from Citibank, N.A., on behalf of hundreds of small businesses who were his customers. Confre-do’s customers knew that the loan applications were fraudulent.
To carry out the scheme, Confredo exploited his educational training in finance and his experience as a former loan officer for a bank; he knew what banks looked for when deciding whether to extend a business loan; and he drafted or procured the drafting of the fictitious loan applications, tax returns, financial statements, and other supporting documents accordingly. On paper, the loan applicants were well-established and profitable businesses; in reality, many of them were not even going concerns but merely vehicles concocted by Confredo and his associates for the sole purpose of securing loans for their customers.
The majority of Confredo’s customers were not credit-worthy, and would not have obtained loans without the false information supplied to the banks by Confredo and his associates. Defense counsel alleged at a proceeding before Judge Sand in May 2006, without dispute from the Government, that in the majority of instances, individuals and/or institutions with good credit co-signed the loan applications. However, there is no indication that any of the approximately $12 million in loans that were ultimately granted were secured by bona fide assets pledged as collateral.
For the services provided by Confredo and his co-conspirators, Granite Financial Services received a fee that typically amounted to between ten and fifteen per cent of the loan amount. Customers paid a portion of the fee up front; if the bank denied a customer’s loan application, Con-fredo or his associates would sometimes *146 return the customer’s payment and sometimes retain it. The presentence report (“PSR”) estimates that, after payments to his associates and staff, Confredo’s personal share of the proceeds from the scheme was “approximately $2,276,467.”
While on bail following his arrest, Con-fredo purported to give truthful information during proffer sessions with the Government, but the Government became suspicious and enlisted a cooperator to meet with Confredo and record their discussion. During that meeting, Confredo told the cooperator that he was still arranging fraudulent loan deals while on bail, had been lying during the proffer sessions, and had been involved in loan sharking.
In light of his post-arrest criminal conduct and his discussion with the cooperator, the Government presented additional evidence to the grand jury, which returned a superseding indictment (the “indictment”) in October 1998, certain counts of which related to offenses Confredo had committed while released on bail.
In February 1999, pursuant to a plea agreement, Confredo pled guilty to one count of bank fraud (18 U.S.C. § 1344), two counts of false statements on a loan application (18 U.S.C. § 1014), one count of false statement to a federal law enforcement officer (18 U.S.C. § 1001), and one count of witness tampering (18 U.S.C. § 1512(b)(3)). The last four offenses were committed after Confredo’s initial arrest and release in November 1997. The plea agreement includes no stipulations as to amount of loss or the applicable sentencing range under the Sentencing Guidelines.
The loss calculation. The PSR calculated the loss amount to be the “total amount requested in [the] various loan applications” involved in Confredo’s scheme, which it estimated to be $24.2 million. Citibank was the target of the majority of the fraudulent applications (more than 100), and it loaned approximately $11 million to Confredo’s customers. The PSR noted that the actual loss to the banks was “extremely difficult to ascertain due to the amount of loans involved, the continual loan payments received by the banks from the customers of the loans and the negotiations of settlement agreements between the banks and these customers.” But the Probation Office did obtain loss statements from a few banks. One Citibank official reported payments of about $2.5 million, and hence an expected actual loss of $8.5 million; but another Citibank official reported a total loss of $9.5 million. The combined actual loss reported by the other banks from whom the Probation Office obtained statements was slightly higher than $1 million. Confredo did not file any objections to the PSR.
The first sentencing. Using the 1997 Guidelines, applicable to Confredo’s offense conduct, Judge Sand began with a base offense level of 6, see § 2Fl.l(a) 1 , added 12 levels not challenged on this appeal, added 16 levels for an intended loss of more than $20 million but less than $40 million, see § 2Fl.l(b)(l)(Q), and added 3 more levels because four counts of Confredo’s conviction involved offenses he committed while on release after his arrest, see § 2J1.7. The adjusted offense level of 37 in Criminal History Category I yielded a sentence range of 210 to 262 months. 2
*147 With respect to the loss enhancement, the Government argued, and the probation officer agreed, that Judge Sand should determine the enhancement based on the intended loss attributable to Confredo’s conduct, which they contended was represented by the combined face value of the loan applications, $24.2 million. Confredo conceded that the actual loss caused by his scheme was in excess of $10 million, but less than $20 million. 3 Judge Sand ruled that intended, rather than actual, loss was the proper measure of the amount of loss under U.S.S.G. § 2Fl.l(b)(l)(Q), and that the amount of the intended loss was the total of all the loan applications. Judge Sand also adopted the PSR’s restitution recommendation, ordering restitution in the amount of $9,338,479.81.
Although Confredo did not dispute the amount of intended loss for purposes of determining the appropriate loss enhancement, defense counsel did discuss the amount of loss at the first sentencing in the context of Confredo’s request for a downward departure. Among the grounds advanced to justify a departure was a claim that the amount of loss calculation overstated the true amount of loss for various reasons, including Confredo’s alleged “intention ... that the loans would be paid for the most part.”
The three-level enhancement for offenses committed while on release was not discussed at the first sentencing, but is challenged on appeal under
Apprendi v. New Jersey,
Judge Sand denied Confredo’s request for a downward departure and orally announced a sentence of 262 months, which was the top of the applicable Guidelines range. Judge Sand also entered 262 months as the total term of imprisonment on the judgment form. However, Judge Sand’s distribution of the sentence among Confredo’s five counts of conviction yielded a total of only 230 months, as reflected in the summary order disposing of Confredo’s prior appeal.
See United States v. Confredo,
First appeal. Confredo appealed, raising numerous claims. As to two of the issues raised on the first appeal — the confusion about whether the total sentence was 230 or 262 months, and the absence of a section 3147 apportionment — the Government conceded that a remand for re-sentencing was necessary. As additional grounds for resentencing, Confredo presented the two claims that are the subject of this appeal: (1) the District Court’s loss calculation was erroneous; and (2) the section 2J1.7 enhancement was unlawful under Apprendi, decided after the first sentencing, because the indictment did not charge that he had committed offenses while on release.
On the first appeal, we remanded for resentencing on the first two grounds just discussed.
See Confredo I,
Confredo’s conviction in the Eastern District of New York. The United States Attorney’s Office for the Eastern District of New York prosecuted Confredo for laundering some of the proceeds generated by his fraudulent loan application scheme. Confredo pled guilty, and, in March 2001, the late Judge Jacob Mishler imposed a 57-month sentence. Judge Mishler ordered this sentence to run concurrently with whatever sentence Judge Sand would impose on remand in this case.
Resentencing. In June 2006, Judge Sand resentenced Confredo. 4 At the resentenc-ing hearing, the only disputed issues as to the applicable sentencing range were the loss amount calculation and the section 2J1.7 enhancement. As to loss amount, the Government argued that Judge Sand should follow the Probation Office’s recommendation of a 16-level enhancement for a loss amount of between $20 and $40 million. The Government’s position was based entirely on the loan applications; it maintained that because Confredo’s customers had applied for a combined value of more than $20 million in loans, the “intended loss” was more than $20 million. Con-fredo argued for a 15-level enhancement on the theory that his intended loss was between $10 and $20 million; he contended that he did not intend a loss in excess of $20 million to occur because he expected the banks to reject some of the loan applications and expected at least some of his customers to pay back all or some of their loans. Confredo cited evidence that many applications were denied and some loan repayments had occurred, leaving the actual loss between $10 million and $20 million, and in fact closer to $10 million.
As he had at the first sentencing, Judge Sand agreed with the Government’s loss calculation. As a threshold matter, Judge Sand ruled that Confredo had “waived” his loss amount argument by not raising it at the first sentencing, but then rejected Confredo’s argument on the merits. In summary, Judge Sand found the expectation-of-repayment point unpersuasive because
[Confredo] undertook no obligation himself to pay off the loans he secured. His “cut” was paid when the loans were secured; it did not matter whether they were paid back. Confredo is not a borrower who used fraud to obtain a loan which he then paid back in full. See generally United States v. Schneider,930 F.2d 555 [ (7th Cir.1991) ]. Confredo secured loans for dozens of entities and he retained no control whether those loans were paid off, nor did his remuneration change depending on whether the borrowers paid any of the funds back.
Consequently, Judge Sand again ruled that in this case the “intended loss” was the combined face value of all the loans for which Confredo’s customers had applied, which the PSR estimated to be $24.2 million. He stated that he had rejected all of Confredo’s arguments, including the statutory and Apprendi objections to the section 2J1.7 enhancement, and was staying with the original 210-262 month Guidelines range. Confredo’s counsel asked the *149 Judge to impose a sentence at the low end of the range, taking into account various mitigating factors, including the fact that several million dollars of loans provided to Confredo’s customers by Citibank had been repaid, reducing the bank’s losses.
Judge Sand then orally delivered a formal opinion explaining his reason for giving a non-Guideline sentence, consistent with what he understood his obligations to be under
United States v. Rattoballi,
Discussion
I. Amount of Loss
Availability of claim.
As noted, Judge Sand ruled at the resentencing that Confredo had “waived” (more accurately, forfeited) his challenge to the loss amount calculation by failing to raise it at the first sentencing. It is true that Confredo filed no objections to the PSR, and his counsel appeared to concede at the first sentencing both that the intended loss controlled and that the amount of this loss was more than $20 million. But it is also true that Confredo sought a departure in part on the ground thaf he intended some of the loans to be repaid, which in fact they were. Even if the loss claim was not properly asserted in the District Court, we will entertain the claim under all the circumstances on the somewhat relaxed application of plain error review that we and other courts have on occasion deemed appropriate for unpreserved sentencing errors.
See United States v. Simmons,
Merits. Confredo challenges the aggregation of all the loan applications to arrive at the intended loss amount for two reasons: (1) based on his experience as a loan officer, he did not expect that all of the loans for which he prepared applications would be granted, let alone at the full amounts sought, and (2) he expected some of the loans to be repaid by his customers. Conceding an intended loss of more than $10 million but less than $20 million, he contends that the loss enhancement should have been 15 levels, see § 2F1.1(b)(l)(P), instead of 16. Although the dispute concerns only one level of enhancement, the difference between the resulting máximums of the two arguably applicable ranges is 27 months.
In considering Confredo’s claim, we encounter some uncertainty as to the standard of review, arising from uncertainty as to whether Judge Sand’s 16-level enhancement was based on a finding of fact, which we would review only for clear error,
see United States v. Rubenstein,
If Judge Sand found as a matter of fact that Confredo intended to cause a loss equal to the face amount of all of the loans, such a finding would likely be affirmed on review only for clear error. Similarly, if Judge Sand concluded that Confredo had failed to present evidence putting his intent in issue, we would likely affirm a conclusion that the aggregate amount of the loans was the intended loss. However, the rationale for Judge Sand’s decision appears to be based, at least in part, on his view that a presenter of fraudulent loan applications will be deemed as a matter of law to have intended a loss equal to the aggregate amount of the loans whenever the presenter is not the borrower. Whether that is a correct interpretation of the fraud guideline requires consideration of the somewhat varied case law on the subject.
Prior to 1991, we had stated that the proper measure of intended loss was the total value of the loan obtained or sought, without regard to whether the defendant had intended to repay the lender.
See United States v. Brack,
In 1991, the Sentencing Commission amended Application Note 7. See U.S.S.G.App. C, Amend. No. 393, at 222-23 (1991). The Note had instructed the sentencing court to determine loss based on “a probable or intended loss.” U.S.S.G. § 2F1.1, comment, (n.7) (1990). The amended version deletes the word “probable,” but preserves the general rule, applicable in both fraud and theft cases, that the sentencing court should use the loss amount that the defendant intended if that amount exceeds the actual loss and can be determined. See U.S.S.G. § 2F1.1, comment, (n.7) (1991). It also acknowledges that “[f]requently loss in a fraud case will be the same as in a theft case.” Id. But the amended version recognizes that there might be types of fraud where an analogy to theft would not be appropriate; in such cases “additional factors are to be considered in determining the loss or intended loss.” Id.
Among such cases are “Fraudulent Loan Application and Contract Procurement Cases,” which are treated in subsection (b) of Note 7. The 1997 version of Note 7(b), applicable here, provides the general rule for loss calculation in fraudulent loan application cases:
In fraudulent loan application cases and contract procurement cases, the loss is the actual loss to the victim (or if the loss has not yet come about, the expected loss). For example, if a defendant *151 fraudulently obtains a loan by misrepresenting the value of his assets, the loss is the amount of the loan not repaid at ■the time the offense is discovered, reduced by the amount the lending institution has recovered (or can expect to recover) from any assets pledged to secure the loan. However, where the intended loss is greater than the actual loss, the intended loss is to be used.
U.S.S.G. § 2F1.1, comment. (n.7(b)) (1997) (emphasis added). Thus, amended Note 7(b) gives the defendant credit for objective facts — payments prior to discovery of fraud and assets pledged to secure the loan — that might alter a loss calculation if based solely on face amounts of loan applications.
Although Note 7(b) sensibly takes into account differences between theft offenses and fraudulent loan applications, it does not specifically provide a method for determining intended loss in a case like the present one, where (a) no collateral is involved, and (b) the defendant is not a borrower but a preparer of fraudulent loan applications who claims to have expected that a number of loans would be denied and at least some portion of the loans granted would be repaid.
In
Brach,
a case governed by the pre-amendment version of section 2F1.1, we upheld the use of the face value of a fraudulently obtained loan as the loss amount, even though the defendant, who was the borrower, had in fact repaid the entire loan, and we assumed that he had intended to repay the loan when he applied for it.
See Brach,
Brach
was rejected by several courts, notably the Third Circuit in a thoughtful opinion by the late Judge Becker.
See United States v. Kopp,
Judge Becker’s opinion disagreed with
Brach
and sided with a Seventh Circuit opinion authored by Judge Posner,
see id.
at 529, 532-33, which had observed that it was “simple” but “irrational” to treat all frauds as equivalent to thefts, preferring an approach that took account of whether the defendant actually intended to pocket the face value of the amount he had fraudulently procured,
see United States v. Schneider,
After
Kopp,
the Third Circuit has consistently held that “[ijntended loss refers to the defendant’s subjective expectation, not to the risk of loss to which he may have exposed his victims.”
United States v. Yeaman,
Since the Commission amended Note 7, we have left open the possibility that a defendant is free at sentencing to present evidence of his intent regarding the issue of loss. In
United States v. Ravelo,
In
United States v. Singh,
We conclude that, after adoption of amended Note 7, the defendant should have an opportunity to persuade the sentencing judge that the loss he intended was less than the face amount of the loans. A defendant who applied for, or caused someone else to apply for, a $1 million loan, fully expecting at least $250,000 to be repaid, intended a loss of no more than $750,000 (although, if no repayment is made, he would be subject to punishment for an actual loss of $1 million). Similarly, a defendant who applied for, or caused others to apply for, ten $1 million loans, expecting at least three to be rejected, intended a loss of no more than $7 million (although, if all were accepted and none was repaid, he would be subject to punishment for an actual loss of $10 million).
We will therefore remand to afford Judge Sand an opportunity to reconsider the intended loss in accordance with this opinion. Either on the present record, or after receipt of additional evidence in the Judge’s discretion, the Judge should determine the extent, if any, to which Confredo has proven a subjective intent to cause a loss of less than the aggregate amount of the loans, in which event the applicable loss calculation should be based only on the intended loss, unless the actual loss is *153 higher. As with all loss calculations, absolute precision is not required. See § 2F1.1 comment, (n.8).
II. The Section 2J1.7 Enhancement
The 1997 version of section 2J1.7, 6 which was applied here to enhance Confredo’s offense level by three levels, provided:
If an enhancement under 18 U.S.C. § 3147 applies, add 3 levels to the offense level for the offense committed while on release as if this section were a specific offense characteristic contained in the offense guideline for the offense committed while on release.
As is clear from its text, section 2J1.7 was designed to implement 18 U.S.C. § 3147, which provides, in relevant part:
A person convicted of an offense committed while released under this chapter shall be sentenced, in addition to the sentence prescribed for the offense to—
(1) a term of imprisonment of not more than ten years if the offense is a felony....
A term of imprisonment imposed under this section shall be consecutive to any other sentence of imprisonment.
18 U.S.C. § 3147;
see United, States v. Stevens,
Confredo did not challenge the section 2J1.7 enhancement at his first sentencing because the conceptual basis for his argument is Apprendi, which was decided later. He raised the issue for the first time in his initial appeal and then at his resentencing, and does so again on this appeal. Although the Government contends that plain error review applies, Judge Sand considered the enhancement on the merits at the resentencing following the remand order that suggested he do so.
Whether or not the
Apprendi
objection was preserved, it is without merit, although the matter requires some discussion. The initial issue is whether
Apprendi
applies to Confredo’s sentence. It is undisputed that the sentence does not exceed the statutory maximum punishment Judge Sand could have imposed, based on Confredo’s guilty plea, even without the three-level section 2J1.7 enhancement. This reason has prompted numerous courts to reject
Apprendi
challenges to section 2J1.7 enhancements in analogous cases.
See United States v. Samuel,
The three level enhancement of section 2J1.7 is to be added “[i]f an enhancement under 18 U.S.C. § 3147 applies.” U.S.S.G. § 2J1.7. Section 3147 is entitled “Penalty
*154
for an
offense
committed while on release,” and several courts have held that it does not create a separate offense but merely provides a sentence enhancement.
See United States v. Jackson,
Because Judge Sand did not impose a specific sentence under section 3147 on any of the counts charging offenses committed while on release, it is arguable that section 3147 has not been “applie[d].” However, Judge Sand recognized that section 2J1.7 represents the Sentencing Commission’s method for implementing section 3147’s requirement of additional consecutive punishment. Understanding why this is so requires consideration of the Commission’s statements in the Background explanation of section 2J1.7 and the Commission’s technique, explained in Application Note 2 to this section, for determining and imposing a sentence for an offense committed while on release.
The Background explanation includes the following
[T]he court is required to impose a consecutive sentence of imprisonment under [section 3147], but there is no requirement as to any minimum term. [Section 2J1.7] is drafted to enable the court to determine and implement a combined “total punishment” consistent with the overall structure of the guidelines, while at the same time complying with the statutory requirement [i.e., consecutiveness].
U.S.S.G. § 2J1.7 comment, (backg’d).
Application Note 2 states:
Under 18 U.S.C. § 3147, a sentence of imprisonment must be imposed in addition to the sentence for the underlying offense, and the sentence of imprisonment imposed under 18 U.S.C. § 3147 must run consecutively to any other sentence of imprisonment. Therefore, the court, in order to comply with the statute, should divide the sentence on the judgment form between the sentence attributable to the underlying offense and the sentence attributable to the enhancement. The court will have to ensure that the “total punishment” (ie., the sentence for the offense committed while on release plus the sentence enhancement under 18 U.S.C. § 3147) is in accord with the guideline range for the offense committed while on release, as adjusted by the enhancement in this section. For example, if the applicable adjusted guideline range is 30-37 months and the court determines “total punishment” of 36 months is appropriate, a sentence of 30 months for the underlying offense plus 6 months under 18 U.S.C. § 3147 would satisfy this requirement.
U.S.S.G. § 2J1.7, comment, (n.2). 7
As the Seventh Circuit has usefully explained, the correct way to perform the task required by Note 2 comprises several steps: (1) determine the applicable sentencing range for the offense committed on release without the section 2J1.7 enhancement, (2) determine the applicable sentencing range with the enhancement, (3) select an appropriate sentence within the en
*155
hanced sentencing range, (4) apportion any part of the sentence that falls within the unenhanced range to the offense committed while on release, (5) apportion the remainder of the sentence to the enhancement, and (6) impose the term apportioned to the enhancement to run consecutively to the term apportioned to the unenhanced sentence.
See United States v. Wilson,
Judge Sand endeavored to follow Application Note 2 and the Wilson methodology. He recognized, however, that, because of the adjustment to account for the sentence imposed by Judge Mishler — reducing by 57 months what would have been a Guidelines sentence of 262 months — Con-fredo’s sentence would be a non-Guidelines sentence and the Judge’s calculation would be “similar to that performed in Stevens.” Judge Sand’s methodology, detailed in the margin, 8 satisfied the purpose of Note 2 and section 3147 because the enhanced portions of the sentences for all offense-on-release counts run consecutively to all other sentences.
The Government contends that Judge Sand applied only section 2J1.7, and not section 3147. The Government relies on the language of section 2J1.7, which states that the three-level enhancement is to be applied “as if this section were a specific offense characteristic contained in the offense guideline for the offense committed while on release.” The District of Columbia Circuit has accepted that view, at least as long as a defendant is not separately charged with a section 3147 offense,
see Samuel,
Since the section 2J1.7 enhancement is the Commission’s technique for implementing section 3147, use of that enhancement is an application of section 3147. Indeed, the only reason the Commission requires apportionment is to satisfy section 3147’s requirement of consecutiveness. The Commission cannot take a statutory requirement of consecutive punishment and, by calling it an offense characteristic, avoid the reality that a defendant, punished for committing an offense while on release, is exposed to the additional consecutive ten-year maximum provided by section 3147.
Since section 3147 applies (even though not as a separately charged offense),
Ap-prendi
also applies because section 3147 exposes Confredo to a higher maximum,
i.e.,
ten more years, than the highest maximum he could have received on the offense-on-release counts. Under our decision in
Gonzalez,
it does not matter that the added sentence in fact left the total sentence within the maximum for the underlying offenses.
See
*156
Although
Apprendi
applies, its jury fact-finding requirement has not been violated. Confredo sufficiently admitted the fact on which the enhancement rested,
ie.,
that he committed the offenses while on release.
See United States v. Booker,
The only aspect of the
Apprendi
claim that might remain is the absence from the indictment of an allegation that the offenses were committed while on release.
See Apprendi
Confredo’s challenge to the section 2J1.7 enhancement is rejected. 9
Conclusion
The case is remanded for reconsideration of the sentence.
Notes
. All references are to the 1997 Guidelines, unless otherwise noted.
. The Probation office had determined the adjusted offense level to be 39, which included a 2-level sophisticated means enhancement pursuant to U.S.S.G. § 2F1.1(b)(5)(C) (1998). Judge Sand did not apply this enhancement on ex post facto grounds because it was authorized after Confredo's offense.
. Apparently Confredo concedes that the larger $9.5 million loss figure reported by Citibank is accurate, bringing the total actual loss to an amount in excess of $10 million.
. The delay between remand and resentenc-ing was caused by several factors, as Judge Sand recounted orally at resentencing. The principal reasons for the delay appear to have been defense counsel’s requests for extensive discovery related to loss amount, and, once
Blakely v. Washington,
. The amended judgment being appealed was further amended by a judgment entered July 20, 2006, to correct a clerical mistake.
. In 2006, the Commission deleted section 2J1.7 and moved it in substance to Chapter Three of the Guidelines, see U.S.S.G. § 3C1.3 (2007); Supplement to Appendix C, Amendment 684, at 154-58 (2006).
. As we have pointed out, the Commission’s example in the last six lines of Note 2 is incorrect under the Commission’s own recommended procedure. See
Stevens,
. Judge Sand determined a total punishment range at level 37 (the enhanced level), which was 210-262 months, selected the top of the range, 262, as an initial punishment and then subtracted the 57 months of Judge Mishler’s sentence to reach a total punishment of 205 months. Then he spread those 205 months among the five counts of conviction as follows: 100 months on Count 2, 30 months on Count 15, 30 months on Count 16, 30 months on Count 23, and 15 months on Count 28. Then he made an allocation within the sentence for each of the four offense-on-release counts as follows: Count 15, 13 months for the offense and 17 months for the enhancement; Count 16, 13 months for the offense and 17 months for the enhancement; Count 23, 14 months for the offense and 16 months for the enhancement; Count 28, 7 months for the offense and 8 months for the enhancement. Finally, and most significantly, he specified that all five sentences are to run consecutively.
. The Government contends that no
Apprendi
error occurred because the fact of committing the offenses while on release is similar to the prior conviction facts that are exempted from
Apprendi. See Almendarez-Torres v. United States,