BOGGS, J., delivered the opinion of the court, in which NORRIS, J., joined. NELSON, J. (pp. 1300-02), delivered a separate opinion concurring in part and dissenting in part.
BOGGS, Circuit Judge.
Jerry E. Lucas, Jr., convicted of misapplying bank funds through a fraudulent loan scheme in violation of 18 U.S.C. § 656, appeals from his sentence. At issue is the “amount of loss” caused by Lucas’s criminal behavior within the meaning of the Sentencing Guidelines. In sentencing Lucas, the district court failed to make findings on an important factual matter and followed a questionable case from the Seventh Circuit, United States v. Dion, 32 F.3d 1147 (7th Cir.1994). We reject the holding of Dion and also limit the applicability of dicta in United States v. Wright, 60 F.3d 240 (6th Cir.1995), to amount of loss cases like Lucas’s. As a result, we remand this case for resolution of the unclear factual matter and. for application of the Guidelines in accord with this opinion.
I
Lúeas was employed as the Vice President of Corporate Banking at First Tennessee Bank (“Bank”) in Knoxville, Tennessee from June 1988 until June 1992. While employed by the Bank, Lucas and an associate, Stan Mitchell, decided to purchase a flea market known as Fleas Unlimited. The purchase price of Fleas Unlimited was $1.7 million, requiring a $300,000 down payment. Mitchell and Lucas were able to raise $250,000, leaving them $50,000 short. Lucas told Mitchell that he had a plan for obtaining the $50,000 they needed.
In March 1992, Lucas prepared and executed an unsecured $45,000 loan from the Bank to Stan Mitchell’s brother, Steve Mitchell. Without Steve Mitchell’s knowledge, Lucas falsely indicated on the loan documents that Steve Mitchell was an “antique dealer” and that the purpose of the loan was to “provide capital to purchase inventory for [his] antiques business.” In reality, Mitchell had a net worth of only $82,300 and earned $43,500 annually working for a [1293]*1293trucking company. Steve Mitchell gave the proceeds of the loan to his brother and Lucas, who ultimately used this money to purchase Fleas Unlimited. Lucas maintains that he gave a promissory note to Steve Mitchell in exchange for the $45,000 loan.
It is unclear from the record when the loan scam was discovered by the Bank. The Pre-sentence Report (“PSI”) subsequently prepared in Lucas’s case does state that the Bank discovered the scam in October or November 1992, but the supporting information given in the PSI shows only that the Bank inquired of Steve Mitchell at that time why interest payments on the loan were past due. Mitchell told a Bank official that the address for the loan should be changed to that of Fleas Unlimited because Fleas Unlimited was going to be making the interest payments on the loan. It is .difficult to conclude from this information alone that the Bank had actually come to realize at this time that it had been the victim of a fraud perpetrated by Lucas, however.
In any event, the loan was paid in full on January 15, 1993. The FBI’s investigation began on January 5,1993, but the PSI states that “it is unlikely that Lucas was aware of this or that he repaid the loan in response to the FBI investigation.” Of course, because it is unclear when the Bank discovered the fraud, it is also unclear, however, whether Lucas repaid the money because he was aware that the Bank had uncovered his seam.
Lucas was indicted on August 25, 1995 on one count of the misapplication of bank funds, in violation of 18 U.S.C. § 656. Lucas pleaded guilty that same day, pursuant to a Fed.R.Crim.P. 11(e)(1)(C) agreement. He was sentenced on November 15, 1995, after the district court adopted -the findings of the PSI and increased Lucas’s base offense level by seven to account for the amount of loss Lucas inflicted on the Bank, pursuant to USSG § 2B1.1. Lucas objected that he should not have been sentenced under § 2B1.1. He argued that the proper Guideline provision to apply to determine the amount of loss was USSG § 2F1.1, which he claimed would result in a finding that he had inflicted no loss upon the Bank, since, in some circumstances, § 2F1.1 permits deductions from the amount of loss in fraudulent loan cases for amounts repaid to the victim of the fraud. .And, as we have noted, Lucas has repaid the Bank in full. In rejecting Lucas’s argument, the district court expressed its serious disagreement with the Sixth Circuit’s opinion in Wright, a case that takes a rather liberal view of amounts later recovered by a victim that can be set off to reduce a defendant’s amount of loss for purposes of the Guidelines. In the words of the district court at Lucas’s sentencing hearing, “that Wright case may be heard en banc before it’s all over, but it depends on how much you all push it.” (Our court had in fact denied rehearing en banc in Wright only one week before Lucas’s sentencing hearing.) The court continued, “You can’t reconcile Wright with [United States v. Buckner, 9 F.3d 452 (6th Cir.1993)] and I don’t think anybody can.” Wright had reversed a decision by the same district judge. The district court decided not to require Lucas to begin serving his sentence until after this appeal had been adjudicated.
II
“[The] meaning of ‘loss’. ... is a question of law that is subject to de novo review.” United States v. Wolfe, 71 F.3d 611, 616 (6th Cir.1995) (quoting United States v. Holiusa, 13 F.3d 1043, 1045 (7th Cir.1994)).
A. § 2F1.1 v. § 2B1.1
The district court accepted the government’s argument that the Seventh Circuit’s Dion decision should be followed as persuasive authority. Dion refused to apply the principles of § 2Fl.l’s commentary to the amount of loss calculation for a defendant convicted of the same crime Lucas pled guilty to, 18 U.S.C. § 656. The defendant in that case, a loan officer, authorized for his own benefit loans in the names of a fictitious person and two individuals who were not customers of the bank. After the bank became suspicious and the defendant felt “the heat,” in the words of the Dion court, he repaid about 40% of the money he had fraudulently obtained. Dion, 32 F.3d at 1148. Looking to the illustrations of the kinds of conduct to which § 2F1.1 was intended by [1294]*1294the Sentencing Commission to apply, the Dion court noted:
These illustrations are distinguishable from Dion’s conduct in this case. The critical aspect of the illustrations offered by § 2F1.1 is the quality of the fraudulent information contained in the bank’s customers’ loan applications — that certain items of information must be falsified in order to obtain the loan. On the other hand, the substance of Dion’s crime is that he converted the funds for his own use. Dion prepared the loan documentation simply to prevent his superiors from discovering his crime in the course of the periodic review of his loans. The quality of the information in Dion’s documentation was immaterial; he needed no approval from anyone to obtain the funds. The bank did not loan Dion this money, he stole it. And he was able to make the transfer of funds look like a loan instead of a theft (albeit with a built-in repayment plan) solely because of his position within the bank. Bolstering the position that § 2F1.1 does not apply to this situation is the absence of 18 U.S.C. § 656 from that section’s list of relevant statutory provisions. From all of this, it is clear that the district court properly selected § 2B1.1 as the guideline section most applicable to Dion’s crime.
Id. at 1149. Judge Cudahy dissented because:
The essential difference between a fraudulent loan and theft or embezzlement seems to me to be whether there was throughout an intent to re-pay the money or, on the contrary, an intent to permanently misappropriate it. Here Dion was regularly making payments on the loans he had illegally arranged; he intended to repay them and under Guideline § 2F1.1 would receive credit for the amount repaid. The majority, on the other hand, seems to think that the difference between §§ 2B1.1 and 2F1.1 is whether the misapplication of funds was an inside or outside job.... I think this is incorrect.
The increased culpability of the loan officer in comparison with, for example, a customer is covered by the enhancement for abuse of a position of trust.
Id. at 1150.
With all due respect to the Seventh Circuit, we believe that although Dion ultimately reached the correct result by holding that no setoff of the amounts repaid should be permitted, it erred in holding that § 2F1.1 was wholly inapplicable to the facts of that case. Section 2B1.1, comment, (n. 2), specifically states, “Where the offense involved making a fraudulent loan or credit card application, or other unlawful conduct involving a loan or credit card, the loss is to be determined under the principles set forth in the Commentary to § 2F1.1 (Fraud and Deceit).” (Emphasis supplied.) Both Dion’s and Lucas’s conduct unquestionably “involved making a fraudulent loan ... application.” Even if their illegal conduct did not involve making a loan application, however, it most certainly was “other unlawful conduct involving a loan.” Because the Sentencing Commission’s commentary on its Guidelines is, in administrative law terms, an interpretation by an agency of its own regulations, it is binding on the federal courts unless it is unconstitutional, violates the terms of a federal statute, or is plainly inconsistent with the terms of the Guideline being interpreted. Stinson v. United States, 508 U.S. 36, 42-47, 113 S.Ct. 1913, 1918, 123 L.Ed.2d 598 (1993); USSG § 1B1.7 (“Failure to follow such commentary could constitute an incorrect application of the guidelines, subjecting the sentence to possible reversal on appeal”).
Moreover, none of the Stinson exceptions could have justified the result in Dion. There was no hint in the case of an argument that § 2Bl.l’s directive to apply the principles of § 2Fl.l’s commentary in fraudulently obtained loan eases was unconstitutional or in conflict with a federal statute. There is also no inconsistency between the terms of § 2B1.1 and the directive in its commentary to apply the principles of § 2F1.1 in fraudulent loan cases. As Judge Cudahy explained in dissent, the distinction the Dion court drew between inside and outside jobs in fraudulent loan cases is not mandated by the terms of § 2B1.1 and in fact seems inconsistent with the drawing of such a distinction in [1295]*1295§ 3B1.3, which allows sentencing adjustments to be made for abuse of positions of trust. In addition to the distinction between inside and outside jobs, the Dion court relied upon two other rationales, both of which are flawed. First, the court argued that Dion had stolen the money, not borrowed it. This argument obliterates the distinction between fraud and theft drawn in the Guidelines. Fraud is obviously a subset of theft, but this is irrelevant for the purposes of the distinction drawn in the Guidelines between fraud theft and non-fraud theft.
Second, the Dim court noted that the commentary to § 2F1.1 does not include 18 U.S.C. § 656 in the list of statutory provisions to which it applies. This argument ignores the fact that § 2B1.1, comment, (n. 2), specifically directs sentencing courts to apply the principles of § 2Fl.l’s commentary in fraudulent loan cases. Moreover, § 2F1.1 is listed as an applicable Guideline in USSG App. A, entitled “Statutory Index,” which the Sentencing Commission notes as “specifying the guideline section or sections ordinarily applicable to a statute of conviction.”
The government offers no arguments to defend the district court’s refusal to apply the principles of § 2F1.1 in this ease other than the argument that this circuit should adopt Dion. A review of the sentencing transcript shows that the government and the district court may have accepted Dion so readily because of a misunderstanding of what Application Note 2 to § 2B1.1 actually requires. It does not require, as the district court assumed, that § 2F1.1 be applied to the exclusion of § 2B1.1. It merely requires that the principles of § 2Fl.1’s commentary be applied to determine the amount of loss. Once the amount of loss is determined with reference to § 2F1.1’s. commentary, however, the base offense level should be determined by the amount of loss matrix in § 2B1.1, which progresses differently than that in § 2F1.1. This is the reason that § -2F1.1 is listed as an applicable Guideline for § 656 offenses only in an appendix to the Guidelines and not in the commentary accompanying § 2B1.1 itself. Section 2F1.1’s table applies to fraud of a generic kind, while § 2B1.1’s table covers “Theft, Embezzlement, Receipt of Stolen Property, and Property Destruction.” Some of the crimes covered by § 2B1.1, of course, can be accomplished by fraud. So § 2B1.1 applies only to certain types of fraud. The Commission has determined that the “amount of loss” inflicted by all categories of frauds should be determined-in an identical fashion, but that the particular kinds of fraud covered by § 2B1.1 should be dealt with more harshly. Therefore, we conclude that § 2F1.1 is useable in connection with § 656 offenses only to determine the amount to be “plugged into” § 2B1.1’s table.1
B. Application of the Principles of § 2F1.1’s Commentary
Section 2F1.1, comment, (n. 7(b)), entitled “Fraudulent Loan Application and Contract Procurement Cases,” squarely applies to this case and of necessity must be quoted at length:
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 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 [1296]*1296lending 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.
Adhering closely to the text of this provision will be sufficient in most cases, but resort to the case law on amount of loss will, of course, sometimes be necessary. Against the factual backdrop of this case, we first analyze the text of Application Note 7(b) and then examine the published decisions by our court that have construed this bit of Guideline commentary.
The clear terms of Application Note 7(b) to § 2F1.1 define amount of loss as “actual loss.” This term is in turn defined, by way of example, as “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.” Fastening upon the word “discovered,” Lucas advances a mens rea-like theory of Application Note 7(b) by arguing that the amount of loss in this case should be zero, since he paid back the loan before he knew an FBI investigation of his fraud had begun.
First, Application Note 7(b) does not make knowledge of an investigation by a defendant in a fraudulent loan case the temporal standard for defining the cut-off date before which amounts repaid can be deducted from the amount of loss for sentencing purposes. The temporal standard employed in Application Note 7(b) is the “time the offense is discovered.” Black’s Law Dictionary defines the cognate word “discovery” as follows: “In a general sense, the ascertainment of that which was previously unknown; the disclosure or coming to light of what was previously, hidden; the acquisition of notice or knowledge of given acts or facts; as in regard to the ‘discovery’ of fraud affecting the running of the statute of limitations, or the granting of a new trial for newly ‘discovered’ evidence.” Black’s Law Dictionary 418 (5th ed. 1979). Application Note 7(b) focuses on discovery of the offense. It is obvious that Lucas’s fraud did not suddenly become known to him on the day he became aware of an FBI investigation. As the perpetrator of the fraud, he was aware of it from the time of its inception. In writing Application Note 7(b) the Commission could have used the phrase “at the time the defrauder discovers .... ” instead of the phrase “at the time the offense is discovered_” If the former phrase had been chosen, we could infer that the Commission was concerned with the time when the defrauder discovered that his offense had been discovered by the authorities or by the victim of the fraud. The Commission’s choice of the latter phrase shows that it was focusing on discovery of the fraud by the victim or by the authorities, whichever comes first, because this is the time that an offense is discovered. This temporal definition of “discovery” recognizes that fraud differs from theft precisely because it is more difficult for the victims to know that they have been victimized.
Second, while the PSI indicated that it was unlikely that Lucas knew that an FBI investigation had begun at the time he repaid the loan, there is no discussion in the PSI or by the district court of what Lucas knew about the Bank’s knowledge of his fraud. It must be remembered that Lucas paid the loan back only after Steve Mitchell was contacted by a curious loan officer at the Bank inquiring about overdue interest payments. This information obviously could have been relayed by Steve Mitchell to Lucas. Indeed, it seems too coincidental that interest payments on the loan were overdue as of October or November of 1992, but that Lucas fully repaid the loan by January 1993, only a short time later. If Lucas did in fact pay back the loan because he knew his fraud had been discovered by the Bank, then even Lucas’s interpretation of Application Note 7(b) could support a refusal by the district court to deduct any of the amounts Lucas had repaid to the Bank.
We reject Lucas’s mens rea theory of Application Note 7(b) and hold that the concept of when an “offense is discovered” relates to discovery by the victim or by the proper authorities, whichever comes first. After discovery, so defined, has occurred, amounts later repaid on a fraudulently obtained loan cannot be set off from the amount [1297]*1297of loss under Application Note 7(b).2 We will remand to the district court to apply this standard to the facts of Lucas’s case and to make explicit findings about: (1) when the Bank discovered Lucas’s fraudulent loan; and (2) when Lucas became aware of the Bank’s discovery. Of course, only the first finding is required under our interpretation of Application Note 7(b), but in fairness to Lucas, who may wish to initiate further review of our decision, we think it wise to solidify the facts upon which his theory depends. We explain in the remainder of our opinion how our decision in this case is consistent with the previous cases our circuit has decided in this area.
C. Case Law Interpreting Application Note 7(b)
Our court has crafted two judicial exceptions to the plain text of Application Note 7(b)’s discussion of the term “amount of loss.” First, in United States v. Chichy, 1 F.3d 1501 (6th Cir.), cert. denied, 510 U.S. 1019, 114 S.Ct. 620, 126 L.Ed.2d 584 (1993), the court created an exception for fraudulent loans where the proceeds flow not to the defendant, but to third parties. The defendants in Chichy had been part of a scheme to procure fraudulent Federal Housing Administration (FHA) loans for others, with a total face value of $1,563,000. At the time of sentencing, only one of these loans, in the amount of $57,150, was in default. The district court estimated the amount of future defaults, when added to this single actual default, to fall in the range of $70,000 to $120,000. Id. at 1503-04. Holding that this was a conservative estimate and that a reasonable methodology had been used, Chichy affirmed the district court’s sentencing of the defendants. Id. at 1508-11. “Actual loss” as defined in Application Note 7(b) would require the face value of all of the loans, minus only the value of any assets pledged to secure the loan that the banks issuing the loans could expect to recover and any repayment amounts made before discovery, to be included in the amount of loss calculation. Chichy, however, also allowed the district court to exclude from the amount of loss calculation payments that the borrowers might be expected to make in the future. Chichy is not a case, however, where, even if their fraud was never discovered, the defrauders involved in that case could ever hope to net any amount approaching the face value of the fraudulently obtained loans. The Chichy court is not clear about the point, but some of the defrauders in that case, who were real estate agents, must have been profiting only by collecting commissions on the sale of homes pin-chased with the fraudulently obtained FHA loans. In this case, however, it is possible that Lucas could have attempted to abscond with all of the $45,000 he obtained through the loan. The key difference, in other words, between this ease and Chichy is that Lucas was the true borrower here. In .the Chichy scheme, the defendants were not the true borrowers.
The second judicial exception to the plain text of Application Note 7(b) was crafted in a line of cases beginning with Wright, which the district court accurately recognized is a troubling case because of certain dicta that it contains. The government also specifically requests that we restrict Wright’s dicta. In the course of analyzing Wright and the cases that follow in its wake, we accept the invitation of the district court and the government by limiting Wright to its facts.
In Wright, the district court faithfully employed a literal reading of Application Note 7(b), but was reversed by our court. The defendant had fraudulently obtained two loans for himself and one loan for a business associate. Although the bank discovered that fraudulent representations had been made in relation to the collateral securing the first loan, it managed to recover all but $59,-750 of the total amount of that loan by selling the collateral. It then exercised its contractual right of set off against the defendant’s cheeking account to recover this entire amount. The bank attempted a similar procedure with regard to the second loan. However, after the setoff was applied to the first loan, no funds remained in the defendant’s [1298]*1298checking account to offset the net loss to the bank on the second loan. As a result, the bank was left with a loss of $13,000 on the second loan. The third loan, to the defendant’s business associate, was in the amount of $100,000. While it was also initially secured by fraudulent collateral that was worthless, when the bank discovered the fraud, the debtor on the loan secured it to the bank’s satisfaction. Following a literal interpretation of Application Note 7(b), the district court held that the amounts of loss, respectively, were $59,750, $13,000, and $100,000. The district court refused to reduce the amount of loss on the first loan by the amounts recovered from the defendant’s checking account because this contractual right was neither a “pledge” nor collateral (see the text of Application Note 7(b), quoted above). It is a bit unclear, but the district court’s finding of a loss of $13,000 on the second loan does not seem to have been challenged in Wright. In relation to the third loan, the district court found the amount of loss to be the face value of the loan because at the time of discovery no amounts had been repaid to the bank and because the collateral in that ease was a completely worthless note owed by a fictitious person. Wright, 60 F.3d at 241-43.
The court of appeals reversed and remanded for resentencing. With regard to the first loan, it held that the amounts the bank had recovered from the defendant’s checking account should be set off, even though the contractual right involved was not a “pledge” in a technical legal sense because, “[t]he example in the Commentary is faulty, but the general words ‘actual loss,’ ‘intended loss,’ and ‘expected loss’ are clear enough.” Id. at 241. We think there is no necessary reason to conclude that the example in Application Note 7(b) is inconsistent in any way with these “general words.” However, we are bound to follow Wright’s clear holding that amounts recovered by a bank pursuant to such a contractual set-off right can be used to reduce the amount of loss under § 2F1.1.
Wright did not comment negatively upon the district court’s method of dealing with the second loan. In terms of the third loan, the court stated, “there was no loss because the defendant, Wright, was not the debtor on the loan but made a false statement about a deed of trust, and the debtor then collateralized the note to the bank’s satisfaction when the bank raised the issue with the defendant.” Id. at 241. Although Wright did not cite Chichy for this point, the Wright court’s holding in this regard is perfectly consistent with the first judicially-crafted exception to the text of Application Note 7(b) in Chichy. In óther words, Wright holds that where the fraud committed by a defendant lies in obtaining a loan for a third party, the amount of loss is only the amount of money currently in default on the loan. Chichy adds the corollary that amounts that are expected to be lost on loan defaults in such cases are also includable in the amount of loss calculation. On the facts of Wright, which involved only one loan fraudulently obtained for a third party instead of the many such loans iri Chichy, the bank was able to satisfactorily collateralize that single loan, indicating a zero loss for the bank in the event of default.
Neither of the two non-textual exceptions to Application Note 7(b)’s definition of actual loss, for loans to third parties or for loans taken out while granting to a bank a contractual right, to set off any deposits at the bank, is applicable to Lucas’s case. While Lucas fraudulently obtained the loan in Steve Mitchell's name, Mitchell immediately turned these funds over to Lucas. Mitchell was the borrower in name only. There is also no evidence in Lucas’s case that the Bank had a contractual right of set-off against Lucas’s account(s) or indeed that Lucas had any accounts with the Bank whatsoever.
Limited to its facts, Wright reaches completely sensible conclusions about the fraudulent loans involved in that case. We reject, however, the following dicta, in Wright that could completely eviscerate the text of Application Note 7(b):
“Loss” should not include amounts that a bank can and does easily recover by foreclosure, setoff, attachment, simple demand for payment, immediate recovery from the actual debtor and other similar legal reme[1299]*1299dies, including the sale of a “pledged” asset covered by the example.
Id. at 242. Wright did not involve attachment, the simple demand for payment, pledges, or “other similar legal remedies” and therefore these portions of Wright are dicta.
Moreover, application of Wright to attachment and similar legal remedies are now barred by Wolfe because these are not remedies in which the agency of the defendant plays a role. See Wolfe, 71 F.3d at 618 (money lost by victims of a Ponzi scheme that is potentially recoverable by a bankruptcy trustee through legal action cannot be deducted from the amount of loss because the recovery of these monies is speculative and not brought about by the agency of the defendant). Wolfe specifically distinguished Wright on this basis: “The difference between Wright and Wolfe’s case is that the recovered funds in Wright stemmed directly from the defendant’s actions, not the actions of others.” Ibid. Recovery of fraudulently obtained money or property by the coercive actions of the state, whether these actions occur years after or immediately after the fraud is discovered, cannot be used to reduce the amount of loss. Whatever mechanism the government uses to assist recovery after a fraud is discovered, the scope of the defendant’s fraud remains unaltered. Such governmental action is unrelated to the defrauder’s conduct and may never occur, whereas the scope of the fraud is limited from the outset in cases where the defrauder himself first acts to constrain the magnitude of the money or goods that he can fraudulently obtain.
Furthermore, application of Wright to simple demands for repayment is too much at odds with the text of Application Note 7(b) to be accepted. Even Lucas should recognize that this dicta would destroy the “hand caught in the cookie jar” rationale, which, in his mens rea-like theory, he claims is central to the amount of loss calculation in fraudulent loan eases. Obviously, a defendant cannot be permitted under the Guidelines to avoid an increase for amount of loss in a fraud scheme simply by being financially capable of repaying the money when discovered. As Judge Batchelder pointed out in her dissent in Wright, “wealth [then] become[s] a determining factor in the calculation of sentences for fraudulent loan application crimes [and] [s]urely this is repugnant to our philosophy of criminal justice.” Wright, 60 F.3d at 244 (Batchelder, J., dissenting).
Finally, although Wright’s dicta relating to pledges obviously should be followed because pledges are addressed by the text of Application Note 7(b),-Wright’s dicta relating to “immediate recovery from the actual debtor” should be limited to cover the Chichy-iike situation where a criminal defendant in a fraudulent loan case procured the loan for a third party. If this latter piece of dicta were read broadly to apply to any situation where repayment was made immediately, then it contradicts other more tightly and cautiously reasoned circuit precedent. See United States v. Scott, 74 F.3d 107, 111 (6th Cir.1996) (refusing to permit a bank loan officer to deduct repayments made within about one month after discovery of the fraud by the bank). See also United States v. Flowers, 55 F.3d 218 (6th Cir.) (“the relevant point in time for determining the amount of loss in a fraud case is at the time the crime was detected, rather than at sentencing”), cert. denied, — U.S.-, 116 S.Ct. 261, 133 L.Ed.2d 185 (1995), a check kiting case.
Most other precedent in our circuit is entirely consistent with the text of Application Note 7(b). See United States v. Lavoie, 19 F.3d 1102, 1104-05 (6th Cir.1994) (district court erred in a fraudulent loan case by refusing to deduct the value of collateral sold from the amount of loss); United States v. Buckner, 9 F.3d 452, 454 (6th Cir.1993) (district court erred in refusing to ednsider evidence of the amount of repayment made on fraudulent loans that occurred before the time the fraud was discovered).
The result in the very recent case of United States v. Sparks, 88 F.3d 408 (6th Cir.1996), is also consistent with the text of Application Note 7(b) and on its facts is very close to Lucas’s case. In Sparks, our court rejected the argument advanced by a father-son team of bank defrauders that the district court should have reduced the amount of loss on a loan nominally in the name of a third [1300]*1300party, where the third party paid back that loan a year after the fraud was discovered. Under our textual reading of Application Note 7(b), this amount paid back by the third party should not have been deducted because repayment occurred after discovery of the fraud. Neither of the two judicially created exceptions to the text of Application Note 7(b) would have applied in Sparks either. The Chichy exception does not apply even though a third party paid back the loan because the district court found that the loan was only nominally in the name of the third party. (Thus, it closely resembles Lucas’s case, since Mitchell was only the nominal borrower.) And the Wright-Wolfe exception, allowing the deduction of monies whose potential recovery is triggered by the defendants’ own agency before the fraud is discovered, is also not implicated.
While we obviously agree with the result in Sparks, we note that the decision in that case rests partially on the argument that the textual example defining “actual loss” in Application Note 7(b) applies only to misrepresentations about the value of assets made to obtain loans. We can see no policy reason to limit Application Note 7(b)’s definition to the circumstances where a loan is fraudulently obtained only by misrepresentations about the value of assets. The defendant in Lavoie, for instance, lied about his place of employment. It makes little sense to distinguish between those who fraudulently obtain loans by misrepresenting the value of their human capital and those who do the same by misrepresenting the value of their physical capital.3
The touchstone in our court’s amount-of-loss fraudulent loan cases is the scope of the fraud, as designed by the defrauder. The limitations on the scope of the fraud in both third party situations (Chichy) and in preexisting agency situations (Wright) have been properly accounted for in our circuit’s § 2F1.1 jurisprudence. In Lucas’s case, we are not faced with any tricky problems about the scope of the fraud involved. Since he clearly authorized the loan for his own benefit and undertook no actions by his own agency before the fraud was perpetrated to reduce the magnitude of the Bank’s potential loss, we are clearly within the text and spirit of Application Note 7(b) in holding that Lucas should be charged with the full value of the loan if the Bank had discovered his fraud before he repaid that loan. This case upon remand turns only on the need for a factual finding about when the Bank discovered Lucas’s fraud.
Ill
We REMAND this case for an application of the principles in Application Note 7(b) of USSG § 2F1.1 as we have interpreted them, after the district court has made factual findings consistent with this opinion.