United States v. Harris , 597 F.3d 242 ( 2010 )

597 F.3d 242 (2010)

UNITED STATES of America, Plaintiff-Appellee,
v.
Andrea Renee HARRIS, Defendant-Appellant.
United States of America, Plaintiff-Appellee,
v.
DeMarquis LaDelle Williams, Defendant-Appellant.

Nos. 08-11121, 08-11151.

United States Court of Appeals, Fifth Circuit.

February 9, 2010.

Delonia Anita Watson, Asst. U.S. Atty. (argued), Fort Worth, TX, for U.S. in both cases.

Kevin Joel Page (argued), Fed. Pub. Def., Dallas, TX, for Harris.

Stephen P. Fahey, Dallas, TX, for U.S. in No. 08-11151.

Peter Christian Smythe (Court-Appointed), Peter Smythe, P.C., Dallas, TX, for Williams.

Before GARWOOD, DAVIS and DENNIS, Circuit Judges.

GARWOOD, Circuit Judge:

Defendant-Appellant Andrea Renee Harris (Harris) pleaded guilty to one count of bank fraud, in violation of 18 U.S.C. § 1344,^[1] on March 18, 2008, and was sentenced in November 2008. On July 15, 2008, in an unrelated case, Defendant-Appellant DeMarquis LaDelle Williams (Williams) pleaded guilty to one count of conspiracy to traffic in or use unauthorized access devices, in violation 18 U.S.C. §§ 371^[2] and 1029(a)(2), and was sentenced on December 3, 2008.^[3] Both Harris and Williams appeal their sentences only, arguing that the respective district courts erred in using the aggregate credit limits of the credit cards compromised by their crimes to calculate the amount of loss under section 2B1.1 of the Sentencing Guidelines.^[4] Williams also argues that the district court erred in finding that there were sixty-three victims of his offense within the meaning of the Guidelines, when only eight of these institutions suffered actual losses.

In September 2009, the two cases were consolidated at the direction of this court for the purpose of oral argument.

For the following reasons, we affirm Harris's sentence in its entirety, and we affirm Williams's sentence in part and vacate and remand it in part.

FACTS AND PROCEEDINGS BELOW

We discuss the facts of Harris's and Williams's cases separately, because they are unrelated except for a legal issue that both share.

Harris's Case

Citibank, N.A., (Citibank) hired Harris as a customer service representative in November 2002. In early March 2003, Harris began accessing Citibank customer accounts without authorization and providing Keasha Turner (Turner) with confidential credit card information. Turner then used this information to make fraudulent charges. Harris also falsely changed information in several customer accounts to indicate that replacement cards had been requested. She then had these replacement cards mailed to addresses where Turner retrieved them. Throughout March 2003, Harris removed "blocks" on the accounts that had been compromised and entered false bank verifications, enabling Turner's fraudulent charges to be processed even after they had been flagged as suspicious.

On March 25, 2003, Citibank's internal fraud investigator confronted Harris about her irregular activities. Harris admitted her role in the fraud and said that her boyfriend had pressured her into helping Turner. She stated that she had not profited from the use of the fraudulent credit cards or from giving the information to Turner. Harris also stated that she had never made any of the fraudulent charges herself. She said that her boyfriend had told her that Turner planned to use the credit cards to purchase gift cards.

Citibank later discovered that another one of its employees, Christianna Wright (Wright) was also providing Turner with customer account information. However, neither Harris nor Wright knew about the other's involvement with Turner.

Harris compromised eight accounts before being caught, of which six sustained a total of $11,812.41 in fraudulent charges.^[5] The eight accounts that were compromised had an aggregate credit limit of $89,770.00. Most of the fraudulent charges made on the cards added up to less than half of their respective limits. However, one account's credit limit was exceeded. That account had a credit limit of $500.00, and about $690.00 in charges were made. Most of the fraudulent charges made on each account were made on the same day, but there was one occasion on which successful charges were made to the same account on more than one day.

On April 24, 2007, Harris was charged with one count of conspiring to commit bank fraud and one count of bank fraud. She waived indictment and pleaded guilty to bank fraud without a plea agreement. The conspiracy count was dismissed on the motion of the United States.

Harris's Pre-Sentence Report (PSR) recommended that she be held accountable for $89,770.00, the aggregate credit limit of the eight accounts she had compromised, rather than the $11,812.41 in actual losses she had inflicted. This recommendation was made based on our holding in United States v. Sowels, 998 F.2d 249 (5th Cir. 1993), and language from the official commentary to the Sentencing Guidelines. USSG § 2B1.1 comment note 3(A)(i) (Nov. 2007) (providing that the loss inflicted by a defendant convicted of fraud is to be calculated as the greater of actual or intended loss). Harris objected to this loss calculation, arguing that Sowels did not support the use of the aggregate credit limit in her case.

Harris renewed her objection at sentencing, but the district court overruled it and adopted the PSR. Based on the district court's calculations, the Sentencing Guidelines' recommended range for Harris's offense was fifteen to twenty-one months of imprisonment. The district court sentenced her to eighteen months and three years of supervised release. She was also ordered to pay a $100.00 special assessment and restitution in the amount of $11,812.41. If Harris's objection to the use of the aggregate credit limit had been sustained, and the loss she had inflicted had been calculated on the basis of the $11,812.41 in actual losses that she had inflicted, her recommended range would have been ten to sixteen months of imprisonment, and she would have been eligible for a split sentence.

Harris timely filed a notice of appeal on November 24, 2008.

Williams's Case

Williams worked as a tollbooth operator at Dallas-Fort Worth International Airport (DFW). In late June or early July 2007, Curtis Davis (Davis) drove up to Williams's tollbooth, identified himself as "D," and spoke to Williams about "making some extra money." Williams indicated that he was interested, so Davis gave him a credit card skimmer and told him to pass his customers' credit cards through it before giving the cards back to them. Davis told Williams that the skimmer recorded and stored the information found on the magnetic strips of the cards.

Williams began regularly passing his customer's credit cards through the skimmer. In late July 2007, Davis returned and gave Williams $2,000.00 in exchange for the skimmer. On August 3, 2007, Davis returned the skimmer to Williams, so he could resume recording his customers' information. Williams was arrested on August 7, 2007, after investigators determined that twenty-two different credit card fraud cases under investigation had charges to his tollbooth in common. He had the skimmer in his possession when he was arrested, and he admitted to having skimmed over 500 credit cards. It was later determined that the exact number was 547. He cooperated with the investigators and identified Davis.

On March 5, 2008, Williams was charged with one count of conspiracy to traffic and use unauthorized access devices. He pleaded guilty to the indictment on July 15, 2008. His PSR recommended that he be held accountable for an intended loss of $2,649,287.25. This figure was calculated by adding $2,545,287.25, the aggregate credit limit for the 339 credit cards of which the credit limits were known, to $104,000.00, which represented a $500.00 loss for each of the 208 cards for which the credit limits were not known.^[6] The PSR also recommended a four-level enhancement based on its finding that Williams's fraud had involved more than fifty victims. USSG § 2B1.1(b)(2)(B). Following the PSR's reasoning, the range prescribed by the Sentencing Guidelines for Williams's conduct would normally have been seventy-eight to ninety-seven months. However, Williams had pleaded guilty to conspiracy, which carried a maximum sentence of sixty months' imprisonment. 18 U.S.C. § 371. Therefore, the Sentencing Guidelines recommended sentence was 60 months (USSG § 5G1.1(a)), and the PSR recommended a sentence of sixty months' imprisonment.

Williams objected to the PSR on two grounds. First, he argued that the aggregate credit limit should not have been used to calculate his intended loss. Instead, he argued that he should only be held responsible for a loss of $157,138.22. This figure was calculated by adding the sum of the actual loss inflicted on the accounts with known credit limits, $53,138.22, to the $104,000.00 total for the 208 cards for which the limits were unknown.^[7] Second, he argued that only eight of the sixty-three banks whose information had been compromised were "victims" within the meaning of the Sentencing Guidelines, because only eight of the banks had reported an actual loss.

At sentencing, Williams renewed his objections. The district court overruled them, finding, in accordance with the PSR, that the amount of loss he had inflicted yielded a recommended range of seventy-eight to ninety-seven months, but that this had to be reduced to sixty months in Williams's case due to the fact that he had pleaded guilty to conspiracy. Accordingly, the district court sentenced him to sixty months' imprisonment with three years' supervised release. He was also ordered to pay $53,138.22 in restitution.

If the district court had assessed the loss inflicted by Williams at $157,138.22 but continued to find that there were more than fifty victims, his guideline recommended range would have been thirty-three to forty-one months' imprisonment. If the district court had assessed the loss at $157,138.22 and reduced the number of victims to eight, his recommended range would have been twenty-one to twenty seven months. And if the district court had reduced the number of victims to eight, but continued to assess the loss at $2,649,287.25, then Williams's recommended range would have been fifty-one to sixty months.

Williams timely filed a notice of appeal on December 8, 2008.

DISCUSSION

On appeal, both Harris and Williams argue that the district courts erred by using the aggregate credit limits of the compromised cards to calculate their intended loss. Additionally, Williams argues that his district court erred in finding that his crime victimized more than fifty individuals. The Government concedes that it was error for the district court to find that Williams victimized more than fifty individuals, but argues that this error was harmless and therefore does not require remand. We address the issue shared by both cases first.

I. Intended Loss Calculations

Under the Sentencing Guidelines, the offense level for offenses involving fraud is increased based on the amount of loss inflicted by the defendant. See USSG § 2B1.1(b)(1). The higher the offense level is, the longer the defendant's recommended sentence will be. The commentary to the Guidelines provides that "loss is the greater of actual loss or intended loss." USSG § 2B1.1, comment n.3(A). "Actual loss" is the reasonably foreseeable pecuniary harm that resulted from the offense. Id. at n. 3(A)(i). "Intended loss" is "the pecuniary harm that was intended to result from the offense." Id. at n. 3(A)(ii). A special rule provides that, in cases involving an "unauthorized access device," a term that includes credit cards, "loss includes any unauthorized charges made with the ... unauthorized access device and shall not be less than $500 per access device." Id. at n. 3(F)(I). The district court "need only make a reasonable estimate of the loss" based upon the evidence. Id. at n. 3(C).

Harris and Williams both argue that the district courts erred in determining the amount of loss they intended to inflict by adding together the credit limits of the cards they compromised. As a threshold matter, they argue that this issue should be reviewed de novo. The Government contends that this issue must be reviewed for clear error and that our precedent indicates that the district courts did not err in using the aggregate credit limits to determine the defendants' intended loss.

We begin by addressing the proper standard of review. We then examine the circumstances under which a district court may use the aggregate credit limit of stolen cards to determine a defendant's intended loss. Finally, we examine the facts of Harris's and Williams's cases to determine whether or not the district courts' use of the aggregate limits was error. We address each case separately, since their different facts led the defendants to make somewhat different arguments before this court.

A. Standard of Review

A district court's sentencing decision is reviewed for abuse of discretion. Gall v. United States, 552 U.S. 38, 128 S. Ct. 586, 597, 169 L. Ed. 2d 445 (2007); United States v. Cisneros-Gutierrez, 517 F.3d 751, 764 (5th Cir.2008). However, the "district court's interpretation or application of the Sentencing Guidelines is reviewed de novo, and its factual findings... are reviewed for clear error. There is no clear error if the district court's finding is plausible in light of the record as a whole." Cisneros-Gutierrez, 517 F.3d at 764 (internal quotation marks and citation omitted).^[8]

Harris's and Williams's cases illustrate how it can sometimes be difficult to distinguish an application of the Sentencing Guidelines from a finding of fact. The defendants argue that the district courts' decision to use the aggregate credit limits of the compromised cards involved an application of the Guidelines that should be reviewed de novo. The Government contends that we must review for clear error, arguing that the intended loss calculations were themselves findings of fact, in that each was essentially a finding that the defendant in question had intended to inflict a certain amount of loss.

We confronted a similar situation in United States v. Klein, 543 F.3d 206 (5th Cir.2008). In Klein, the defendant was a doctor who committed insurance fraud by several means, one of which was to submit claims for the in-office administration of injections he had instructed his patients to self-administer at home. Id. at 208. Another of his fraudulent practices was to charge retail prices for medications he had bought wholesale. Id. In calculating the losses he had inflicted, the district court simply added together the face values of the fraudulent bills he had submitted. Id. at 209. It did not subtract the legitimate value of the injections that were self-administered or the legitimate wholesale value of the medications he had sold from the face values of the fraudulent charges. Id. On appeal, the Government argued for clear error review of the district court's method of calculating loss, while the defendant argued that de novo review was appropriate. We agreed with the defendant:

"The government is correct that the finding as to the amount of loss is a factual finding, and we cannot reassess the evidence but owe the finding deference. We can, however, consider de novo how the court calculated the loss, because that is an application of the guidelines, which is a question of law. United States v. Saacks, 131 F.3d 540, 542-43 (5th Cir.1997). In fact, before assessing the court's loss estimate, we ``first determine[ ] whether the trial court's method of calculating the amount of loss was legally acceptable.' {United States v.} Olis, 429 F.3d {540,} 545 {(5th Cir.2005)} (citing Saacks, 131 F.3d at 542-43; United States v. Krenning, 93 F.3d 1257, 1269 (5th Cir.1996))." Klein, 543 F.3d at 214 (italicization and brackets in original; material in braces added).

This language and the cases from which it was derived require us to review a district court's method of determining the amount of loss de novo.^[9]See Klein, 543 F.3d at 214; Olis, 429 F.3d at 545; Saacks, 131 F.3d at 542-43.

Adding the credit limits of compromised cards together to estimate intended loss is a method of determining the amount of loss. Therefore, we hold that the district courts' use of this method in the defendants' cases must be reviewed de novo.

B. The Method of Calculating Intended Loss

The method used by the district courts in this case was not to determine the defendants' intended loss based on direct inferences from the facts, but rather to apply Sowels as a sort of bright-line rule that allowed them to factually estimate the defendants' intended loss as being equal to the aggregate credit limits of the stolen cards. We have affirmed methods of determining intended loss similar to the one used by the district courts numerous times in the past. United States v. Oates, 122 F.3d 222, 225 (5th Cir.1997) ("This Court has long adhered to the view, supported by the relevant application note, that the amount of loss for the purpose of determining a base offense level in U.S.S.G. § 2F1.1(b)(1) is the dollar amount placed at risk by a defendant's fraudulent scheme or artifice."^[10]) (citing United States v. Brown, 7 F.3d 1155, 1159 (5th Cir.1993); Sowels, 998 F.2d at 251; United States v. Wimbish, 980 F.2d 312, 315-16 (5th Cir. 1992), abrogated in part by Stinson v. United States, 508 U.S. 36, 113 S. Ct. 1913, 1918-19, 123 L. Ed. 2d 598 (1993);^[11]United States v. Hooten, 933 F.2d 293, 298 (5th Cir.1991)). However, an examination of our opinions reveals that it is only proper for sentencing courts to use these methods under certain factual circumstances.

1. The Rule of Wimbish

In United States v. Wimbish, the defendant forged a series of stolen checks and deposited them, requesting cash back from each deposit. Wimbish, 980 F.2d at 313. At his sentencing, the district court calculated the amount of intended loss by adding together the face values of the checks he had forged. Id. The defendant objected to this method and appealed his sentence on the ground that he had known the deposits would be detected as forgeries and voided. Id. Thus, he argued, his intent was only to steal the amount of money that he had requested as cash back when he made the deposits. Id. He concluded that this meant the district court should have found that his intended loss was the same as the actual loss he had inflicted: the amount of money that he had requested as cash back from the fraudulent deposits. Id.

We rejected this argument. Id. at 312. We found that it was proper to use the sum of the full face values of the forged checks, because the banks could have failed to detect his fraud in a timely manner. Id. at 315-16. The possibility of such a failure meant that the defendant's scheme placed the entire face value of each stolen check at risk. Id. We held that the defendant's "conscious," "callous indifference" to this risk placed the face values of the stolen checks within the ambit of his intended loss. Id. at 316. Our subsequent cases have interpreted Wimbish as establishing that the full value of property recklessly jeopardized by a defendant's crime may be considered part of his intended loss. E.g., Oates, 122 F.3d at 225.

2. The Rule of Wimbish and Third-Party Transfers

In United States v. Morrow, we examined one way in which property can be recklessly jeopardized by a defendant's crime. 177 F.3d 272, 300-01 (5th Cir.1999) (per curiam). The defendants in Morrow were mobile home salesmen who committed fraud in helping their customers obtain financing. Id. at 285. At sentencing, the district court calculated their intended loss by adding together the face values of the loans they had procured fraudulently. Id. at 300. On appeal, the defendants challenged this calculation. Id.

We began by noting that the repayment of the fraudulent loans was controlled entirely by the defendants' customers, not by the defendants themselves. Id. We found that this lack of control meant that the district court had "accurately characterized the defendants as consciously indifferent or reckless about the repayment of the loans." Id. at 301 (internal quotation marks omitted). Accordingly, we held that the district court had not erred in calculating the defendants' intended loss by adding the face values of the fraudulent loans. Id. Morrow stands for the proposition that one way a defendant can recklessly jeopardize the full value of property during the commission of a crime is to place that property in the hands of a third party whom he does not control. See id. at 300-01. If transferring the property to a third party creates a significant risk that the property will be completely destroyed, then its face value may be used to determine the defendant's intended loss under the rule of Wimbish. See Morrow, 177 F.3d at 300-01; Wimbish, 980 F.2d at 316.

3. The Rule of Wimbish and Incomplete Offenses

In United States v. Sowels, the defendant was a postal employee who stole between fifty and seventy-five letters containing credit cards in November 1991 and 110 letters with cards in January 1992. 998 F.2d at 250. He was arrested before he was able to use the 110 cards he stole in January, but after he had already made $28,540.89 in unauthorized charges on the cards he had stolen in November. Id. At sentencing, the district court calculated his inflicted loss by adding the $28,540.89 in actual charges made in November to the $351,600.00 aggregate limit of the 110 cards that were stolen, but never used, in January. See id. at 252.

On appeal, the defendant challenged the propriety of using the aggregate credit limit of the unused cards stolen in January. Id. at 250-51. We affirmed his sentence. See id. at 251-52. The district court had made an explicit finding of fact that the defendant had intended to use all of the credit available under the cards. Id. at 251. We reviewed this finding for clear error. Id. Even though there was no direct support for the finding in the record, we found that it was not clearly erroneous for two reasons. See id. The first reason was that the finding was supported by the rule of Wimbish, because the defendant had planned to transfer the stolen cards to third parties whom he did not control, meaning his crime would recklessly jeopardize the full value of the cards' credit limits. Id. The second reason was that the offense was incomplete at the time the authorities intervened. Id. We stated that, "[h]ad Sowels completed or withdrawn from his offense before being apprehended, he might have been able to rebut the evidence that he intended to charge the cards to their limit. Given that authorities cut short his plans, however, the district court did not clearly err." Id. See also id. at 252 ("[T]his case is unique because it involves an uncompleted offense. For this reason, the district court faced the difficult task of projecting into the future Sowels's intent as to the extent to which he would use the cards."). Thus, Sowels established first, that the rule of Wimbish can be applied to credit cards, and second, that a defendant cannot easily rebut a factual finding that his intended loss was equal to the full value of property jeopardized by his crime if his fraud was never completed due to the intervention of law enforcement. See id. at 251-52.

4. The Requirement of Actual, Not Constructive, Intent

Both Harris and Williams argue that it is improper for a sentencing court to find that a defendant intended to inflict the total loss of property that he recklessly jeopardized, citing language from some of our opinions that provides that intended loss must be calculated by looking to the defendant's actual, not constructive, intent. E.g., United States v. Sanders, 343 F.3d 511, 527 (5th Cir.2003) ("[O]ur case law requires the government prove by a preponderance of the evidence that the defendant had the subjective intent to cause the loss that is used to calculate his offense level."); United States v. Hill, 42 F.3d 914, 919 (5th Cir.1995) ("When reviewing the calculation of an intended loss, we look to actual, not constructive, intent ...."). We recognize that this language is confusing when contrasted with the rule of Wimbish, which allows intent to be inferred from recklessness for the purpose of calculating intended loss. See Wimbish, 980 F.2d at 315-16. However, this apparent conflict can be resolved by examining the origin of the language requiring "actual, not constructive, intent."

We first mentioned this requirement in United States v. Henderson, 19 F.3d 917, 928 (5th Cir.1994). In Henderson, the defendant was convicted of bank fraud for making loans from one bank, of which he was the president and largest shareholder, to a second bank he had started with a friend, without disclosing to the first bank that he was a co-owner of the second bank. Id. at 919-22. The sentencing court, apparently having confused its actual and intended loss inquiries, refused to consider the possibility that the defendant had intended to inflict no loss. Id. at 927-28. Instead, it found that his intended loss was the face value of the fraudulent loans he had procured, reasoning that "we intend the result of the acts that we take." Id. On appeal, we found that the district court had misapplied the rule of Wimbish, because it had found intent to inflict a loss of the loan's face value based not upon recklessness, but on the fact that an actual loss ultimately had occurred. Id. at 928. We noted that, unlike Wimbish and Sowels, which had involved stolen property, Henderson involved a fraudulent loan. Id. Obtaining a loan fraudulently is different from stealing property outright, because defendants who fraudulently obtain loans often intend to repay them in full. See id. It follows that, "where the defendant intends to repay the loan or replace the property, the intended loss is zero." Id. Accordingly, we vacated the defendant's sentence and ordered the district court to consider his argument that he had intended no loss. Id. at 928-29.

Thus, Henderson only established that courts could not use the fact that a defendant's crime ultimately resulted in a loss of property to find that he constructively intended to inflict any loss. See id. Such a method of calculating intended loss confuses what actually occurred with what the defendant intended to occur. See id. It was in this context that we said, "[t]he Sentencing Guidelines refer to actual intent, not constructive intent." Id. at 928. It is clear from this context and from the way we have discussed Henderson in later opinions that we did not use the phrase "constructive intent" in the sense that it is often used in discussing the mental states of substantive criminal offenses. Compare Henderson, 19 F.3d at 928 ("[T]he intended loss for stolen or fraudulently obtained property is the face value of that property.") and United States v. Hill, 42 F.3d 914, 919 (5th Cir.1995) (stating that "we look to actual, not constructive, intent," but also holding the defendant responsible for the face value of fraudulently-lent securities based on the risk imposed on the victims who borrowed the securities), with Wayne R. LaFave, 1 Substantive Criminal Law § 5.2(e), at 352 (2d ed. 2003) ("[T]he notion of ``constructive intent' has been used by some courts; it is first asserted that intent is required for all crimes, and then it is added that such intent may be inferred from recklessness or negligence.").

We did not prohibit sentencing courts from inferring intent from a defendant's recklessness, for to do so would have been in direct conflict with Wimbish and its progeny. See, e.g., Tedder, 81 F.3d at 551 (inferring intent to cause a loss from the fact that the defendant had helped third party clients whom he did not control to obtain loans, while recklessly disregarding the fact that their poor credit histories and fake social security numbers meant that they were highly unlikely to pay).^[12] Instead, we held that a sentencing court cannot reject a defendant's claim that no loss was intended based purely on the fact that a loss was eventually inflicted.^[13]Henderson, 19 F.3d at 928-29.

An interpretation of Henderson's "constructive intent" language that used the term as it is often used in discussing the intent requirements of criminal statutes would produce absurd results. In Tedder, the defendant, who had helped people with poor credit histories obtain a series of fraudulent loans with fake social security numbers, attempted to argue that it always had been his "intent" that the loans be repaid. Tedder, 81 F.3d at 551. He even presented credible evidence that he always instructed his clients to "pay their bills on time." Id. The sentencing court calculated his intended loss as being the face value of the loans he had recklessly obtained for third parties, and he appealed, arguing the record clearly showed that his actual intent had been for his clients repay their loans. Id. We were able to reject this strained argument, because the district court could reasonably infer intended loss from defendant's reckless acts, regardless of his evidence of "actual intent." See id.

A rule that prohibited sentencing courts from inferring intent from a defendant's recklessness would approach effectively creating a defense capable of eviscerating the "intended loss" provisions of the Sentencing Guidelines for any criminal who managed to insulate his crime from the ultimate infliction of loss. It takes no great feat of the imagination to picture pickpockets selling stolen credit cards to third parties, only to claim at sentencing, "Sure, I sold them the cards, but it was none of my business what they planned to do with them," or computer hackers gaining access to bank account numbers on behalf of clients, only later to claim, "I don't know, and I didn't ask, what they planned to do with those numbers." We did not write Henderson to prevent the intended loss provisions of the Sentencing Guidelines from being applied to criminal middlemen who act recklessly or in willful blindness, and we decline to read it in a manner that accomplishes that result now.^[14]

5. The Aggregate Limit of Stolen Credit Cards

Under the rule of Wimbish, a sentencing court may find that a defendant intended to inflict a loss of the face value of property that was recklessly jeopardized by his crime. Wimbish, 980 F.2d at 315-16. We have consistently held that one way a defendant may jeopardize property is by transferring it to a third party whom he does not control. Morrow, 177 F.3d at 300-01. We have also held that credit cards constitute a form of property that can be recklessly jeopardized in this way. Sowels, 998 F.2d at 251; United States v. Mordi, No. 92-1675, 1993 WL 152261, at *2-4 (5th Cir. Apr. 21, 1993) (per curiam). Therefore, in cases where a defendant recklessly jeopardizes the full credit limit of a card by transferring it to a third party whom he does not control, the sentencing court may reasonably find that his intended loss was equal to the limit of that card. See Sowels, 998 F.2d at 251; Mordi, 1993 WL 152261, at *4. Although it is possible for a defendant to rebut the inference that he intended to charge a card to its limit by introducing evidence of a contrary modus operandi, this will be difficult to accomplish, so as to render clearly erroneous a contrary district court finding, if there is no evidence that his offense was complete at the time the defendant was stopped by the authorities. See Sowels, 998 F.2d at 251-52.

C. Harris's Intended Loss

Harris makes three major arguments in support of her assertion that the district court erred in sentencing her based on the aggregate limit of the cards she compromised. The first is that, while sentencing based on the credit limits of compromised cards is supported by Sowels, her case is distinguishable. Her second major argument is that the district court erred by relying on a First Circuit case, United States v. Alli, 444 F.3d 34 (1st Cir.2006), that is not good law in the Fifth Circuit. Her final argument is that, even if Alli is good law in this circuit, its holding is inapplicable to her case, because the record reflects that she controlled the third parties whom she provided with credit card account information. We examine each argument in turn, but find no error in the district court's sentence.

1. Sowels and Harris's Intended Loss

Harris argues that the method of assessing intended loss as being equal to the aggregate limit of stolen credit cards, which we approved in Sowels, cannot be applied to her case, because our holding in Sowels was conditioned on the fact that the defendant had not yet used the cards he had stolen at the time he was apprehended. See generally Sowels, 998 F.2d at 251. She contends that, while the method used in Sowels is appropriate in cases where the cards have not been used, the actual charges are better indicators of a defendant's intended loss in cases where the defendant managed to make fraudulent charges before being apprehended. See generally id. She notes that in Sowels, the sentencing court had calculated the loss on cards that had actually been charged by adding the actual charges, rather than the credit limits. Id. at 250. She concludes that it was improper for the district court to hold that she intended to inflict a loss of the aggregate limit of the cards she compromised, because most of the cards she compromised were charged well below their limits.

Harris is correct that the reasoning in Sowels rests in part on the fact that the defendant was not able to establish a pattern of charges due to the intervention of law enforcement. See Sowels, 998 F.2d at 251 ("Had Sowels completed or withdrawn from his offense before being apprehended, he might have been able to rebut the evidence that he intended to charge the cards to their limit."). However, the broader principle behind our result in Sowels was the rule of Wimbish. This is evident from the fact that Sowels relied heavily on our unpublished opinion in United States v. Mordi. See Sowels, 998 F.2d at 251.

In Mordi, the defendant was arrested for using stolen credit cards to purchase merchandise. 1993 WL 152261, at *1. At sentencing, the district court found that he had inflicted an actual loss of $45,368.00. Id. It further found that his intended loss was $90,768.00, because at the time the defendant was arrested, there remained $45,400.00 in unused credit on the accounts the defendant had compromised. Id. The district court used the larger number to determine his sentence, in accordance with the Guidelines. Id. at *3. On appeal, we affirmed this sentence. Id. at *4. Citing Wimbish for support, we held that the defendant recklessly had put his victims at risk for the aggregate amount of the credit cards' limits. Id. We held that he could therefore be held to have intended a loss of the cards' aggregate limit.

Thus, the central rationale behind Sowels is derived from our opinion in Mordi, which itself is based on the rule of Wimbish. As is illustrated by our opinion in Mordi, whether or not a stolen credit card has sustained charges does not dispositively determine whether or not its full credit limit was recklessly jeopardized by the defendant's crime. See Mordi, 1993 WL 152261, at *4. A defendant can more easily establish a modus operandi defense where his offense was completed before he was apprehended by the authorities. See Sowels, 998 F.2d at 251-52. But even in a case involving a completed offense, it may still be reasonable for the sentencing court to find that the defendant intended to inflict a loss of the entire credit limit of a stolen card if he committed his crime in such a way that the entire limit was recklessly jeopardized. See Mordi, 1993 WL 152261, at *4 (holding that a defendant had intended to inflict the loss of the aggregate limit of stolen cards, because his crime had recklessly jeopardized this limit, even though the actual fraudulent charges on the cards did not exceed it).

Therefore, we disagree with Harris that Sowels's application of the rule of Wimbish to a case of credit card theft wholly relied on the fact that the defendant in Sowels had not actually made any fraudulent charges at the time he was apprehended by the authorities.^[15] The fact that Harris's co-conspirators had made charges on the cards she compromised before they were stopped by the authorities merely means that it was easier for her to mount a modus operandi defense to the PSR's finding that she intended to inflict a loss of the entire aggregate limit. It did not guarantee that this defense would succeed.

2. The District Court's Use of United States v. Alli

Harris also argues that the district court erred in relying on the First Circuit case of United States v. Alli. 444 F.3d at 34. The district court cited that case as support for the proposition that, because Harris "gave account information to other people, including her codefendant [sic], Keasha Turner," the aggregate credit limit of the accounts she compromised was the proper measure of her intended loss, since "she had no control over the amount of charges [these] other people made to the victim's [sic] accounts." Harris argues that the district court's reliance on Alli was improper, because the facts of Alli involved no actual loss, and because the First Circuit declined to follow what it viewed as the law of this circuit in reaching its holding.

In Alli, the First Circuit held that a defendant's "``intended loss' includes ``losses that might naturally and probably flow from' his unlawful conduct" and therefore that a defendant who sold (or intended to sell) cards to a third party without making any charges himself could be held to have intended to inflict the loss of their aggregate limit. 444 F.3d at 38 (quoting United States v. Jacobs, 117 F.3d 82, 95 (2d Cir. 1997)). In reaching this holding, the panel rejected our requirement that a sentencing court's intended loss calculation be based on "actual, not constructive, intent." Id. (citing Morrow, 177 F.3d at 301). We do not need to decide whether the First Circuit's examination of "losses that might naturally and probably flow" from a defendant's conduct is equivalent to our examination of what property was recklessly jeopardized. We also do not need to determine whether Alli's holding relied upon the fact that there was no actual loss inflicted in Alli. These inquiries are irrelevant, because the record reflects that the district court cited Alli to support the proposition that, generally, a defendant who transfers stolen property to a third party whom he does not control recklessly jeopardizes the face value of that property. This proposition is amply supported by our case law. See, e.g., Morrow, 177 F.3d at 300-01; Sowels, 998 F.2d at 251. Therefore, the district court did not err by referring to Alli to support it.

3. Harris's Control Over the Parties to the Conspiracy

Harris's final argument is that, even if Alli is good law in this circuit, its holding is inapplicable to her case because the record reflects that she controlled the third parties whom she provided with credit card account information. If the assertion that she controlled her co-conspirators' actions were supported by the district court's findings of fact, the rule of Wimbish might be inapplicable to her case, since it then likely could not be shown that she had recklessly jeopardized the full credit limits of the cards she compromised. If reckless jeopardy could not be shown, then arguably her intended loss could only have been assessed as the aggregate limit if a finding were made that the intent or hope of the conspiracy had been to ultimately charge the cards to their limits. That may have been difficult to establish, given that Harris's co-conspirators did not charge most of the cards to their limits and usually did not make any charges on a card after the first day it was used.

However, the record does not compel acceptance of Harris's assertion that she controlled the actions of her co-conspirators. In fact, the district court explicitly found that "she had no control over the amount of charges other people made to the victim's [sic] accounts." This is a finding of fact, which we must review for clear error. See Cisneros-Gutierrez, 517 F.3d at 764. While Harris was able to remove obstacles to the processing of her co-conspirators' fraudulent charges, there is no indication in the record that she ever intervened to erase a fraudulent charge that she felt was too large, nor is there any indication that she even had the power to do so. In fact, when Harris's co-conspirators exceeded the limit of one of the cards, there is no indication in the record that Harris did anything about it. If she were truly the controlling figure in a conspiracy with an understood goal of never reaching the limits of any cards, this inaction would be difficult to explain.

But we think this inaction makes sense, because the record supports the inference that Harris was little more than a pawn in Turner's conspiracy. Harris was pressured into helping Turner by her (Harris's) boyfriend. Harris's only knowledge of Turner's plan to purchase gift cards came through this boyfriend. Harris did not even know that there was another participant in the conspiracy, Christine Wright. Given these facts, we cannot conclude that the district court clearly erred in failing to find that Harris controlled the actions of the third parties to whom she gave the account information.

That Harris's co-conspirators might have avoided overcharging the cards does not change this analysis. The question before us is whether or not the district court clearly erred in determining that Harris intended to inflict a loss of the credit limits of the cards. This determination was supported by the fact that Harris recklessly jeopardized this amount by transferring it to third parties whom she does not appear to have known or controlled. That the record might support a finding that these third parties never intended to inflict a loss of the credit limits does not excuse Harris's recklessness in giving the credit card information to them.

4. Conclusion as to Harris's Intended Loss

Under the rule of Wimbish, a sentencing court may infer intent to inflict a loss equal to the face value of property based on the fact that the defendant recklessly jeopardized that property during the commission of his crime. See Wimbish, 980 F.2d at 315-16. That a defendant recklessly jeopardized property that he obtained fraudulently may be reasonably supported by a finding that he transferred it to a third party whom he did not control. See Morrow, 177 F.3d at 301. We find that the record in Harris's case supports a finding that she transferred confidential credit card information to third parties whom she did not control.

Therefore, we hold that the district court did not err in calculating her intended loss as being equal to the credit limits of the cards she compromised.

D. Williams's Intended Loss

Williams also argues that the district court erred in calculating his intended loss for three reasons. First, he contends that the district court could not properly calculate his loss as being equal to the aggregate limit of the cards without finding that he subjectively intended to inflict a loss of that amount. Second, he argues that potential risk alone cannot support sentencing enhancements under the Guidelines. Finally, he argues that the trial court improperly sentenced him based on his co-conspirators' subjective intent.

We have already addressed Williams's first argument above. We do not interpret the language from our previous opinions that required "actual, not constructive, intent" as preventing a sentencing court from inferring intent from recklessness or willful blindness. See Tedder, 81 F.3d at 551. Thus, where a defendant's crime recklessly jeopardized property, the defendant may properly be held to have intended a loss of the amount jeopardized. Wimbish, 980 F.2d at 315-16. A defendant may recklessly jeopardize property by transferring it to a third party whom he does not control. Morrow, 177 F.3d at 300-01. In this case, Williams's PSR indicated that he had given the information from over 500 credit cards to an individual whom he knew only as "D" in exchange for $2,000.00. If this did not recklessly jeopardize the credit limits of these cards, it is hard to imagine what would. Therefore, we do not think that the district court erred in adopting the PSR's determination that Williams had intended to inflict a loss equal to the aggregate credit limit of the cards he had compromised.

Williams's second argument, that potential victim risk alone cannot support sentence enhancements, is also without merit. We have held numerous times in the past that a defendant's sentence may be enhanced where his crime recklessly jeopardizes property, even if that property survives the crime intact. E.g., Wimbish, 980 F.2d at 315-16. In these cases, the sentencing court may infer the intent to inflict loss from the defendant's reckless disregard for the value of the property that was jeopardized. Id.

Williams's third argument is a variation on his first. He argues that the court "transferred" his co-conspirators' subjective intent to steal money from the cards onto him. He asserts that his intent was only to record the information off of the cards and give it to Davis. Therefore, he concludes, it was error for the district court to find that he intended to steal any amount of money. This argument lacks merit, because a sentencing court may reasonably infer intent from recklessness or willful blindness. Id. We refuse to find that Williams must be treated as having intended no loss simply because he happened to be a criminal middleman. He illegally obtained information that allowed him to steal an amount of money equal to the aggregate limit of the cards he skimmed. He then gave this information to Davis, in effect transferring to Davis an option to take the aggregate limit of the cards that had been skimmed.

Whether or not Davis intended to exercise this option is irrelevant to Williams's intended loss, because Williams had no control over Davis's actions. The district court could reasonably find that Williams knew that Davis planned to steal money (indeed, it is hard to conceive of any other reasonable finding), and that he enabled Davis to steal the full amount that could be obtained with the cards. By holding Williams responsible for an intended loss of the aggregate limit of the cards, the district court held him responsible for his own intent, not that of his co-conspirators.

Therefore, we find that the record in Williams's case supports a finding that he intended to inflict a loss equal to the credit limits of the cards he compromised. The district court did not err in holding him responsible for this amount.

E. Conclusion as to the District Courts' Method

As we discussed in detail above, neither Sowels nor Wimbish allows district courts to always automatically assess intended loss as being equal to the limits of stolen credit cards. Rather, the method used in each case was contingent upon a factual background wherein the defendant evidenced his intent to cause a certain amount of loss by recklessly jeopardizing property worth that amount.

The records in both Harris's and Williams's cases support the intended loss findings ultimately made by the district courts, because both records reasonably support the implicit findings that the defendants recklessly jeopardized the aggregate credit limits of the cards they compromised. For this reason, we find no error in the district courts' intended loss calculations.

II. Number of Victims in Williams's Case

Williams's second assignment of error is that the district court enhanced his sentence four levels based on its finding that his offense had involved fifty or more victims. Williams objected to this enhancement, because only eight of the sixty-three financial institutions identified as "victims" by the district court suffered an actual loss. The Government correctly concedes that it was error to apply this enhancement in its brief, explaining that:

"In the application notes to § 2B1.1, ``victim' is defined to include ``any person who sustained any part of the actual loss determined under subsection (b)(1).' USSG § 2B1.1, comment. (n.1) (emphasis added). ``Actual loss' is ``the reasonably foreseeable pecuniary harm that resulted from the offense.' Id. (n.3(A)(I)). ``Pecuniary harm' means ``harm that is monetary or that otherwise is measurable in money.' Id. (n.3(A)(iii)). Here, only eight of the 63 financial institutions that had issued the credit cards improperly skimmed by Williams suffered an ``actual loss' .... Thus, only those eight financial institutions would qualify as ``victims' under the plain language of § 2B1.1, and it was error for all 63 banks to be counted in calculating the four-level enhancement for more than 50 ``victims.' See, e.g., United States v. Conner, 537 F.3d 480, 489 (5th Cir.2008) (recognizing that, under language of § 2B1.1, ``actual loss' must be suffered to qualify as a ``victim' under § 2B1.1(b)(2))."

However, the Government argues that this error does not require remand, because Williams's sentence fell within the correct range, and because it asserts that the record adequately demonstrates that the district court would have imposed the same sentence without the error. Williams argues that the record does not show that the district court would have selected the same sentence if it had calculated the correct range.

Under the district court's erroneous interpretation of the Guidelines, Williams's recommended range would have been seventy-eight to ninety-seven months, had he not pleaded guilty only to conspiracy. Because Williams pleaded guilty only to conspiracy, the Guidelines recommended a sentence of sixty months under USSG § 5G1.1(a).^[16] If the district court had correctly assessed the number of victims, Williams's recommended range normally would have been fifty-one to sixty-three months, which would have been reduced by the statutory maximum for conspiracy to a range of fifty-one to sixty months. See USSG § 5G1.1(c)(1).

A procedural error made during sentencing is harmless if the error did not affect the district court's selection of the sentence imposed. United States v. Delgado-Martinez, 564 F.3d 750, 753 (5th Cir.2009) (citing Williams v. United States, 503 U.S. 193, 112 S. Ct. 1112, 1120-21, 117 L. Ed. 2d 341 (1992)). The burden of establishing that an error is harmless rests on the party seeking to uphold the sentence. Delgado-Martinez, 564 F.3d at 753. The proponent of the sentence "must point to evidence in the record that will convince us that the district court had a particular sentence in mind and would have imposed it, notwithstanding the error made in arriving at the defendant's guideline range." Id. (quoting United States v. Huskey, 137 F.3d 283, 289 (5th Cir.1998)). We use this method of analysis even when the sentence selected under a district court's erroneous interpretation of the Guidelines would have fallen within the correct range. Delgado-Martinez, 564 F.3d at 753. "[T]he crux of the harmless-error inquiry is whether the district court would have imposed the same sentence, not whether the district court could have imposed the same sentence." Id. (emphasis in original). The fact that the actual sentence fell within the properly calculated Guidelines range may sometimes be relevant to the harmless-error inquiry, but it is not dispositive. Id.

After reviewing the record, we cannot say that the district court would have imposed the same sentence if it had properly calculated Williams's Guidelines range. The Government points to several statements made by the district court at Williams's sentencing hearing as support for its argument that the district court would have imposed the same sentence if it had correctly calculated the guidelines range:

"[Y]ou are dealing with over 500 cards that were skimmed improperly, that is why the Guidelines are so high, and I think this qualifies as a serious offense.

* * *

So I think the Guideline range is appropriate in this case. It is reasonable for the conduct. It is a serious offense.

And frankly, Mr. Williams, you received a break by the fact that your lawyer was able to work out a plea agreement to get you to plead guilty to a conspiracy. If you had pled guilty to the underlying statute, you would be looking at 78 to 97 months or six to eight years confinement. That is where your Guidelines are. By working a plea agreement for conspiracy, you limited your exposure to 60 months. ...

So I think the Guideline range is reasonable in light of the conduct that is involved here, the type of crime, and the magnitude of the crime ...." (emphasis added).

We find that these statements are ambiguous, because it is unclear to us whether, in speaking of the "Guideline range," the district court was referring to the seventy-eight to ninety-seven month range that would normally have been recommended or to the sixty month range that was technically recommended in Williams's case due to the fact that he pleaded guilty to conspiracy. While the sixty month range was the official Guidelines range for Williams's case, the district court also referred to the seventy-eight to ninety-seven month range as "where your Guidelines are."

Furthermore, that the district court found its miscalculated Guidelines range "reasonable" would not necessarily mean that it would have issued the same sixty month sentence if it had been apprised of the correct range. This is true whether the range to which the district court referred was the sixty month sentence or the seventy-eight to ninety-seven month sentence. We have nothing that indicates to us, with the confidence needed to satisfy the standard of harmless error review on this issue, that the district court would have sentenced Williams to sixty months' imprisonment if it had calculated his Guidelines range correctly.

Therefore, we must vacate Williams's sentence and remand it to the district court, so that it may sentence him after considering the correct Guidelines range for his offense, fifty-one to sixty months. The district court may decide not to alter Williams's original sentence, but the sentence must be reconsidered.

CONCLUSION

For the foregoing reasons, we affirm the district court's sentence in Harris's case. In Williams's case, we affirm the district court's intended loss determination, but vacate and remand his sentence on the issue of the number of victims of his crime, with instructions that the district court consider the correct Guidelines range of fifty-one to sixty months before sentencing him again.

No. 08-11121, USA v. Harris, AFFIRMED; No. 08-11151, USA v. Williams, Sentence VACATED and REMANDED for resentencing with instructions.

NOTES

[1] Section 1344 reads as follows:

"§ 1344. Bank fraud

Whoever knowingly executes, or attempts to execute, a scheme or artifice—

(1) to defraud a financial institution; or

(2) to obtain any of the moneys, funds, credits, assets, securities, or other property owned by, or under the custody or control of, a financial institution, by means of false or fraudulent pretenses, representations, or promises;

shall be fined not more than $1,000,000 or imprisoned not more than 30 years, or both." 18 U.S.C.A. § 1344 (2000) (emphasis in original).

[2] Section 371 reads as follows:

"§ 371. Conspiracy to commit offense or to defraud United States

If two or more persons conspire either to commit any offense against the United States, or to defraud the United States, or any agency thereof in any manner or for any purpose, and one or more of such persons do any act to effect the object of the conspiracy, each shall be fined under this title or imprisoned not more than five years, or both.

If, however, the offense, the commission of which is the object of the conspiracy, is a misdemeanor only, the punishment for such conspiracy shall not exceed the maximum punishment provided for such misdemeanor." 18 U.S.C.A. § 371 (2000) (emphasis in original).

[3] The relevant portions of Section 1029 read as follows:

"§ 1029. Fraud and related activity in connection with access devices

(a) Whoever—

* * *

(2) knowingly and with intent to defraud traffics in or uses one or more unauthorized access devices during any one-year period, and by such conduct obtains anything of value aggregating $1,000 or more during that period;

* * *

shall, if the offense affects interstate or foreign commerce, be punished as provided in subsection (c) of this section." 18 U.S.C.A. § 1029(a) (2000) (emphasis in original).

[4] Both Harris and Williams were sentenced under the November 2007 version of the Guidelines.

[5] The two remaining accounts that were compromised did not sustain fraudulent charges.

[6] See USSG § 2B1.1, comment note 3(F)(i). The record suggests that the limits of these 208 cards were not known because the relevant credit card companies did not respond to requests for information about these accounts.

[7] Presumably, no fraudulent charges had been made on these 208 cards.

[8] The parties do not contend, and the record does not reflect, that in either case the district court intended to either make a Guideline authorized departure or a non-guideline sentence (under Gall or Kimbrough v. United States, 552 U.S. 85, 128 S. Ct. 558, 169 L. Ed. 2d 481 (2007)).

[9] We recognize that some of the language found in our earlier opinions may be in tension with Klein. See, e.g., United States v. Hill, 42 F.3d 914, 918-19 (5th Cir.1995) (concluding "that the district court did not clearly err" where it found that the defendant's intended loss was the face value of securities in which the defendants had no legal interest but nevertheless fraudulently rented to their victims); United States v. Wimbish, 980 F.2d 312, 316 (5th Cir.1992) (holding that the district court "did not clearly err in calculating the loss value under USSG § 2F1.1 as the face value of the checks deposited" where the defendant only stole the portions of the face values which he requested as "cash back" from his fraudulent deposits). We read these opinions as primarily examining the factual bases of the loss determinations that were at issue. Under this reading, we regard the panels that wrote these opinions as implicitly having approved de novo the methods used by the district courts before explicitly declining to find clear error in the factual bases for the district courts' calculations.

This reading avoids conflict with our later cases and allows us to treat a district court's method of determining loss as an application of the Guidelines, rather than as a finding of fact. We think the former approach is superior, because the issue of whether or not a particular method of determining loss is proper, though often fact-dependent, is itself a sufficiently abstract and general proposition to amount to a question of law. Therefore, we think that a district court's method of determining loss should be reviewed de novo, while clear error review should be applied to the background findings of fact that determine whether or not a particular method is appropriate, as well as to the mathematical calculations and the appraisals of value that are made using the method chosen.

[10] In early versions of the Guidelines, "OFFENSES INVOLVING FRAUD OR DECEIT" were addressed in USSG § 2F1.1, instead of § 2B1.1. See, e.g., United States Sentencing Commission, Guidelines Manual, § 2F1.1 (Nov. 1997).

[11] The opinion in Stinson v. United States did not abrogate Wimbish on grounds that are relevant to the issues presented by Harris's and Williams's cases.

[12] In United States v. Tedder, we implicitly distinguished Henderson on the basis of the fact that the repayment of the fraudulent loans in Henderson was controlled by the defendant, while the repayment of the fraudulent loans in Tedder was in the hands of third parties whom the defendant did not control and whose repayments he did not intend to make in the event of a default. Tedder, 81 F.3d at 551.

[13] Accordingly, in remanding Henderson, we did not require the district court to find that the defendant had intended no loss. We ordered it to consider the possibility that the defendant had intended no loss and to make a factual finding one way or the other. Henderson, 19 F.3d at 928-29.

[14] We do not here deal with or address the elements of a criminal offense or the statutory maximum sentence, but rather with the Sentencing Guidelines in the post-Gall and Kimbrough world.

[15] We also disagree with Harris that our opinion in United States v. Ismoila, 100 F.3d 380 (5th Cir. 1996), requires this interpretation. In Ismoila, we merely characterized Sowels as having held "that available credit limit can be used as a measure of loss when the credit cards were stolen but not used." Ismoila, 100 F.3d at 396 (emphasis in original). We did not state that the available credit limit could not also be used as a measure of loss where the credit cards were used and the full amount of their limit was recklessly jeopardized. The central point of our opinion in Ismoila was that sentencing courts have a large amount of flexibility and discretion in making these determinations. See id. ("Available credit is simply one way of determining intended loss."). That is likewise significant in our holding here.

[16] USSG § 5G1.1 states:

"Sentencing on a Single Count of Conviction

(a) Where the statutorily authorized maximum sentence is less than the minimum of the applicable guideline range, the statutorily authorized maximum sentence shall be the guideline sentence." (Emphasis in original).