OPINION OF THE COURT
COWEN, Circuit Judge.This ease presents the issue of whether the bank fraud sentencing guidelines, which we earlier interpreted as requiring a sentencing court to calculate the amount of the victim’s loss as it exists at the time of sentencing rather than at the time of the commission of the offense, permit a defendant convicted of kiting checks to significantly reduce his sentence by paying back all or a portion of the amount he absconded with during the commission of the offense. In the particular case *111of a defendant who has violated the bank fraud statute through the act of kiting checks, we conclude that, in the absence of overriding facts dictating other treatment, the sentencing court must ordinarily calculate the amount of the loss as it exists at the time the crime was detected, rather than as it exists at the later time of sentencing. Because the district court in this case sentenced the defendant by calculating the amount of the actual loss which resulted from the crime as it existed at the time of sentencing, we will vacate the sentence and remand for resen-tencing.
I.
The facts giving rise to Franklin Shaffer’s conviction and sentence are relatively simple and largely undisputed. Shaffer formerly led several engineering and construction firms which were engaged in large construction projects in central Pennsylvania. During the summer of 1988, collections of accounts receivable fell significantly behind, resulting in a sizable cash shortfall for the companies. After the companies’ line of credit was canceled, Shaffer attempted to keep his businesses afloat by kiting checks for large sums of money between his personal bank accounts and the various business accounts.1 At the time he was writing cheeks, Shaffer did not have sufficient funds in the accounts to cover the check amounts.
In September, 1988, a bank officer reported the matter to federal authorities. After an investigation, Shaffer was charged with executing and attempting to execute a cheek kiting scheme from July through September, 1988. By the time the FBI investigated the matter, Shaffer had sufficient funds in all the accounts to cover all the checks. For this reason, the United States Attorney for the Middle District of Pennsylvania recommended Shaffer as a possible candidate for pre-trial diversion. By order dated August 21, 1991, the district court placed Shaffer on pre-trial diversion for twelve months, ordered him to pay the victim banks the interest each would have earned on the money Shaffer had borrowed through the check kiting scheme, and ordered him to perform community service.
While on pre-trial diversion, Shaffer again began kiting checks because of cash flow; shortfalls in his various construction firms. Bank officials notified the government of suspicious activity in Shaffer’s accounts on August 19, 1992. A motion was filed with the district court requesting that Shaffer be removed from pre-trial diversion. The motion was granted and the FBI undertook an investigation. Unlike the first time his check kiting was discovered, Shaffer was not able to cover all the checks he had written. Four of the five victim banks used by Shaffer in the eheck kiting scheme reported gross losses at the time of detection as follows: Fulton Bank — $40,371.46; Commerce Bank — $18,-020.88; Dauphin Deposit Bank — $206,636.60; and CCNB — $197,280.66. The total loss was determined to be $462,309.60.
Shaffer was charged with two counts of bank fraud, in violation of 18 U.S.C. § 1344, for the two separate cheek kiting incidents. After plea bargain negotiations, Shaffer pleaded guilty to both counts on January 13, 1993. As part of the plea agreement, Shaffer *112agreed to make restitution to the victim banks in an amount to be determined by the district court at a pre-sentencing hearing. At his arraignment, Shaffer requested a sentencing delay in order to allow him to get his business affairs in order and to give him sufficient time to attempt to make restitution to the victim banks.
The United States Probation Office prepared a presentence report pursuant to the United States Sentencing Guidelines (“U.S.S.G.”) which assessed a total offense level of 17 against Shaffer. The offense level was determined in the following manner: (1) a base level of 6 was assessed pursuant to U.S.S.G. §' 2Fl.l(a); (2) an increase of 9 levels was added under U.S.S.G. § 2Fl.l(b)(l)(J) since the total amount of the loss to the victim banks exceeded $350,000 but was less than $500,000; and (3) an additional 2 level increase was made pursuant to U.S.S.G. § 2Fl.l(b)(2)(B) because the crime involved a scheme to defraud more than one victim. The pre-sentence report concluded that the applicable guideline sentence range was from 24 to 30 months.
A sentencing hearing was held on July 16, 1993. By this time, Shaffer had negotiated settlement agreements with three of the four victim banks. Pursuant to these agreements, Fulton Bank had accepted a settlement of $20,000 in “full satisfaction” of its loss of $40,371.46; Commerce Bank had accepted $10,500 in “full satisfaction” of its loss of $18,020.88; and Dauphin Deposit Bank agreed to accept the conveyance of a parcel of real estate in Shreveport, Louisiana held by Shaffer in his retirement account, secured by a judgment against one of Shaffer’s business corporations, and a promissory note in the sum of $84,000 from Shaffer in “full satisfaction” for its loss of $206,636.60. No agreement was reached between Shaffer and the fourth victim bank, CCNB.
At the sentencing hearing, Shaffer objected to the 9 level increase pursuant to U.S.S.G. § 2Fl.l(b)(l)(J). He argued that no increase was warranted because he actually intended no loss to the victim banks at the time of the commission of the offense. Based on the evidence presented, the district court agreed that Shaffer at all times intended to repay the amounts borrowed during the check kiting scheme through the collection of accounts receivable, and made a factual finding that Shaffer actually intended no permanent loss to the victim banks. Nevertheless, the district court disagreed that the loss was zero for all victim banks or the three banks which had entered into settlement agreements with Shaffer. The district court concluded that the “actual loss” at the time of sentencing was the total loss of $462,309.60 less the amounts the three victim banks had agreed to accept in lieu of their initial losses pursuant to the settlement agreements.2
Since this reduced the loss for sentencing purposes to $347,809.60, the district court enhanced Shaffer’s base level only by 8 levels pursuant to U.S.S.G. § 2Fl.l(b)(l)(I). In so doing, the district court rejected the government’s position that “actual loss” in a check kiting bank fraud case is the initial loss of the victim banks at the time the fraud is detected, which should not be reduced by any subsequent settlement payments in the nature of restitution that the defendant makes. The district court further granted Shaffer a 2 level base level reduction pursuant to U.S.S.G. § 3El.l(a) for acceptance of responsibility. Premised on a base level of 14, the district court sentenced Shaffer to an eighteen month term of imprisonment, three years of supervised release, and ordered him to make restitution in full to CCNB and as agreed with the other victim banks.3
*113II.
Shaffer filed a notice of appeal challenging his sentence on the theory that the district court overstated the amount of the victims’ loss pursuant to U.S.S.G. § 2F1.1(b)(1). The government also filed a notice of appeal to challenge Shaffer’s sentence taking the contrary view that the district court understated the amount of the victims’ actual loss. We have appellate jurisdiction pursuant to 28 U.S.C. § 1291. These appeals involve a legal interpretation of the appropriate calculation of “loss” under U.S.S.G. § 2Fl.l(b)(l), over which we have plenary review. United States v. Badaracco, 954 F.2d 928, 936 (3d Cir.1992).
III.
The question on appeal is whether to calculate the total amount of the victims’ loss for purposes of sentencing a check Mter as it exists at the time the offense is detected or at the time of sentencing. While the language of U.S.S.G. § 2Fl.l(b) itself, together with its commentary, does not directly address whether the loss calculation should be as it exists when the fraud is detected or at sentencing, the commentary does state that “the loss need not be determined with precision,” U.S.S.G. § 2F1.1 comment, (n. 8). Furthermore, the commentary states that “if an intended loss that the defendant was attempting to inflict can be determined, this figure will be used if it is greater than the actual loss.” Id. (n. 7).
We have previously held that in the context of procuring a secured bank loan through fraudulent misrepresentation, fraud “loss” pursuant to U.S.S.G. § 2Fl.l(b) “is, in the first instance, the amount of money the victim has actually lost (estimated at the time of sentencing), not the potential loss as measured at the time of the crime.” United States v. Kopp, 951 F.2d 521, 536 (3d Cir.1991). The actual loss calculation, in a case such as Kopp, will reflect the deduction of the value of the collateral, pledged as security for the loan, from the loss sustained by the defrauded lender. We recognized in Kopp that if actual loss as calculated at the time of sentencing understates the amount of loss the defendant intended to inflict, then the “loss” figure should be revised upward to the intended loss figure. Id.
The present ease does not involve an intended loss theory for the calculation of “loss” pursuant to U.S.S.G. § 2Fl.l(b) because the district court made a finding of fact that Shaffer actually intended no permanent loss whatsoever.4 Citing Kopp for authority, Shaffer argues that the district court should have determined that there was no loss to the three victim banks which entered into settlement agreements with Shaffer, which would have resulted in an increase of only 7 base levels pursuant to U.S.S.G. § 2Fl.l(b)(l)(H) since the fourth victim bank, CCNB, had a total loss of just under $200,000. Alternatively, Shaffer argues that the district court was correct in its loss calculation pursuant to U.S.S.G. § 2Fl.l(b) in that the loss was calculated as it existed at the time of sentencing, not as it existed at the time of the detection of the offense, and therefore his sentence should be affirmed. The government argues that this case is distinguishable from Kopp because the crimes of kiting cheeks and the fraudulent procurement of a secured loan, while both bank frauds, are sufficiently distinct to warrant differing treatment under the Guidelines.
In Kopp, a case involving bank fraud where the borrower submitted false information in order to obtain a commercial real estate mortgage, we limited the loss calculation pursuant to U.S.S.G. § 2Fl.l(b) to the higher of “actual loss” calculated at the time of sentencing or the amount of loss the borrower actually intended. 951 F.2d at 527-36. In that case, we held that the actual loss is that which exists at the time of sentencing, *114not at the time the fraudulent act was committed, because the total amount of the loan without a reduction for the actual or estimated value of the collateral pledged to the bank would overstate the victim bank’s actual loss. See id. at 528-30. Thus, we determined that calculation of the victim’s actual loss at the time of sentencing would more accurately reflect the defendant’s culpability for sentencing purposes.
Subsequently, we have indicated that the Kopp holding does not provide a rule that should be followed for all types of bank fraud convictions. For instance, where a bank officer was convicted of bank fraud for using his position for his personal benefit by conditioning loan approval on the borrower using contractors in which he owned an interest, we determined that actual loss was the total amount of the contracts received by the related contractors, rather than the net gain or profit to these companies as calculated or estimated at the time of sentencing. United States v. Badaracco, 954 F.2d 928, 936-38 (3d Cir.1992). Thus, we distinguished Kopp because the type of bank fraud at issue in Badaracco was more similar to an embezzlement crime, where the loss is calculated as gross gain, rather than a secured loan crime where the defendant actually pledges something of value, the collateral, which will reduce the amount of the victim bank’s loss below the face value of the loan. See 954 F.2d at 937-38.
We stated in Badaracco that “[ajlthough section 2F1.1 is applicable to a wide variety of fraud schemes, the sentencing judge is entitled, probably compelled, to evaluate the size of the loss based on the particular offense.” Id. at 937. - We believe that check kiting crimes, because of their particular nature, are crimes where the district court must calcúlate the victim’s actual loss as it exists at the time the offense is detected rather than as it exists at the time of sentencing. Cf. United States v. Katora, 981 F.2d 1398, 1406-07 (3d Cir.1992) (sentencing court not obliged to reduce amount of wire fraud loss by speculative value of personal guaranties given at time of commission of offense). We come to this conclusion for several reasons.
First, the commentary to U.S.S.G. § 2F1.1 indicates that loss valuation should be made in accordance with the valuation of theft loss pursuant to the commentary to U.S.S.G. § 2B1.1. U.S.S.G. § 2F1.1 comment.'(n. 7). “As in theft cases, loss is the value of the money, property, or services unlawfully taken.” Id. In a check kiting scheme, where the offender writes bad checks to “temporarily ... obtain credit,” Black’s Law Dictionary 238 (6th ed. 1990), the amount of money owed to the victim banks at the time the kite is detected is the value of the money unlawfully taken by the defendant. In effect, the gross amount of the kite at the time of detection, less any other collected funds the defendant has on deposit with the bank at that time and any other offsets that the bank can immediately apply against the overdraft (including immediate repayments), is the loss to the victim bank.
Second, we do not believe that most check kiting frauds are sufficiently analogous to secured loan frauds to require, as we held in Kopp, that the actual loss determination be made as the loss exists at the time of sentencing rather than at the time of detection. Although both of these types of bank fraud involve fraudulently obtained loans, the similarities end there. Secured loan frauds include an aspect that is ordinarily entirely absent from -a check kiting scheme — namely collateral, which while probably insufficient to protect the victim bank completely against risk of loss, usually provides some recovery against the loan amount. By its very nature, the crime of kiting checks ordinarily involves the borrowing of funds without authorization from the bank and without the offender providing any security to protect the bank against risk of loss. This distinction warrants treating perpetrators of check kiting loan frauds in most cases differently from perpetrators of secured loan frauds for sentencing purposes.
Furthermore, we do not believe that calculating check kiting fraud loss at the time of detection is contrary to the commentary contained in the Sentencing Guidelines. The Sentencing Guidelines have been clarified to make explicit the rule we announced in Kopp as applicable to secured loan frauds, but not *115unsecured loan frauds like check kiting. The Guidelines commentary now states in relevant part:
In fraudulent loan application 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 lending institution has recovered (or can expect to recover) from any assets pledged to secure the loan.
U.S.S.G. § 2F1.1 comment, (n. 7(b)). The Sentencing Guidelines apparently limit this wait-and-see approach to calculating actual loss to secured loans because with unsecured loans, like those which sometimes result when check kiting schemes are detected, any recovery is entirely speculative.5 Nevertheless, to the extent the Guidelines commentary indicates that in secured loan frauds the actual loss should be reduced by the amount of money immediately repaid by the offender at the time of discovery, the same is true of check kiting frauds.
Moreover, the weight of authority, while sparse, supports our conclusion that for purposes of sentencing a defendant who has perpetrated a bank fraud by kiting checks, actual loss should be calculated as it exists at the time of detection rather than at the time of sentencing. Likening check kiting frauds more to simple theft than secured loan transactions, the Court of Appeals for the Fifth Circuit adopted the position we adhere to today. United States v. Frydenlund, 990 F.2d 822, 825-26 (5th Cir.), cert. denied, — U.S. —, 114 S.Ct. 192, 126 L.Ed.2d 150 (1998). Other courts of appeals have held that the fortuity of the defendant paying full restitution to the victim banks after the time when the check kiting fraud was detected does not warrant a downward departure on the sentence for acceptance of responsibility. United States v. Carey, 895 F.2d 318, 322-23 (7th Cir.1990); United States v. Bolden, 889 F.2d 1336, 1340-41 (4th Cir.1989).
Finally, we, like the district court, are troubled by the outcome which would result if actual loss is calculated as it exists at the time of sentencing rather than at the time of detection. We agree with the sentiment, stated by the district court at sentencing, that a reduction of sentence because of a last minute payment of restitution would unfairly discriminate in favor of those with greater financial resources.6 We are also concerned that permitting such a reduction in sentence might encourage the use of undue pressure by a defendant to induce the victim bank into settling for payment of only a portion of the amount it has lost. In sum, it is a hallmark of our sentencing scheme that criminal defendants who have committed identical crimes, and who have the exact same culpability, should be treated equally at sentencing pursuant to U.S.S.G. § 2Fl.l(b) even though one has the means through business or personal relations to make restitution to the victim banks after he has been convicted, while the other does not.
IV.
We conclude that the district court erred in calculating the victims’ loss pursuant to *116U.S.S.G. § 2Fl.l(b) from this check kiting fraud as it existed at the time of sentencing rather than as it existed at the time of detection. Thus, we will vacate Shaffer’s sentence and remand for resentencing consistent with this opinion.
. The following example illustrates a typical check kiting scheme:
The check kiter opens an account at Bank A with a nominal deposit. He then writes a check on that account for a large sum, such as $50,000. The check kiter then opens an account at Bank B and deposits the $50,000 check from Bank A in that account. At the time of deposit, the check is not supported by sufficient funds in the account at Bank A. However, Bank B, unaware of this fact, gives the check kiter immediate credit on his account at Bank B. During the several-day period that the check on Bank A is being processed for collection from that bank, the check kiter writes a $50,000 check on his account at Bank B and deposits it into his account at Bank A. At the time of the deposit of that check, Bank A gives the check kiter immediate credit on his account there, and on the basis of that grant of credit pays the original $50,000 check when it is presented for collection.
By repeating this scheme, or some variation of it, the check kiter can use the $50,000 credit originally given by Bank B as an interest-free loan for an extended period of time. In effect, the check kiter can take advantage of the several-day period required for the transmittal, processing, and payment of checks from accounts in different banks. Williams v. United States, 458 U.S. 279, 281 n. 1, 102 S.Ct. 3088, 3089-90 n. 1, 73 L.Ed.2d 767 (1982). (quotation marks and citation omitted).
. Thus, the district court agreed to reduce the total initial loss by $114,500, which was composed of the settlement agreement amounts of $10,500 for Commerce Bank, $20,000 for Fulton Bank, and $84,000 for Dauphin Deposit Bank.
. Shaffer ■ filed a post-sentencing letter with the district court seeking a further 1 level base level reduction in his sentence pursuant to U.S.S.G. § 3El.l(b). The district court agreed that a base level reduction of 3 levels was required for Shaffer’s acceptance of responsibility, rather than the 2 base level reduction which was granted during sentencing. Although the district court did not enter an order changing the sentence since it felt that 18 months of incarceration was appropriate and 18 months was still well within the guideline range, the district court did instruct the probation office to adjust the guideline offense level to 13. Neither party contests this adjustment on appeal.
. At oral argument, the government contended first that it was unnecessary to address whether the district court was clearly erroneous as to its factual finding concerning Shaffer’s intent, and alternatively, that the finding was clearly erroneous. Since we conclude that resolution of this case more appropriately turns on whether the district court settled on the appropriate actual loss figure by calculating the loss as it existed at the time of sentencing, rather than as it existed at the time the crime was detected, we need not address whether the district court's finding of fact was clearly erroneous.
. We reject the attempt made by Shaffer in this appeal to liken his crime more to a secured loan fraud than a theft because he was the principal of several businesses which had a large amount of accounts receivable outstanding during the period of the kite. The record reveals that at the time of the second check kite, which resulted in loss to the four victim banks, Shaffer’s various business interests also had a significant outstanding loan from an individual named "Cochrane" which was secured with the business receivables. See App. at 174, 185 (testimony of Shaffer).
. The district court stated the following at sentencing:
As a final note, I should add that the court is troubled by the outcome in this case. The [cjourt recognizes that the prospect of a potentially reduced sentence acts as a carrot to a defendant sparing that individual to make speedier or more adequate restitution than he or she might otherwise do benefiting the victims of his or her crime. However, the [cjourt sees a great danger in allowing an individual to buy his or her way out of jail time.
Unfortunately, this is what a reduction of sentence in a last minute payment of restitution amounts to. Such a policy I think unfairly discriminates in favor of those with greater financial resources. I believe in light of the cases and in light of the way the Third Circuit has spoken, it is the only outcome this [cjourt can conclude is appropriate for this case.
App. at 157-58.