UNITED STATES of America, Plaintiff-Appellee, v. Judy YEUNG, a/k/a Mui Wan Yeung, Defendant-Appellant, 672 F.3d 594

OPINION IKUTA, Circuit Judge: Judy Yeung appeals from a restitution order imposed by the district court after a jury convicted Yeung of various crimes associated with her involvement in a fraudulent real estate investment scheme. We have jurisdiction pursuant to 28 U.S.C. § 1291. We hold that the district court failed to provide an adequate explanation of its reasoning in calculating the amount of restitution owed to two of the victims and, therefore, vacate that portion of the restitution order.

We remand for recalculation and explanation of the award pursuant to the Mandatory Victims Restitution Act of 1996 (“MVRA”), 18 U.S.C. § 3663A. I From 2005 until 2007, Judy Yeung and her co-conspirators developed a scheme to make money in the then-booming housing market and to pay off various debts. Yeung recruited five individuals with good credit scores to act as straw buyers to purchase and refinance residences in San Francisco and Gilroy, California.

Yeung and other members of the conspiracy falsified information on the straw buyers’ loan applications, for example by materially overstating the supposed borrower’s income and assets and misstating their employment information and rental obligations. As part of the scheme, the conspirators developed a false letter from an accountant and a false “verification of deposit” letter from Hang Seng Bank to support the fraudulent loan applications.

Despite her promises to the straw buyers, Yeung stopped making mortgage payments and defaulted on the loans, which triggered foreclosure of - the properties held as collateral. After she was convicted, the district court determined that the victims of two of the five fraudulent schemes (the “Ferrari loans” and the “Lam loans”) were entitled to restitution in the amount of $1,321,564. The relevant facts about the two schemes at issue, drawn from the record, are as follows. A The scheme involving the Ferrari loans began in 2005.

At the suggestion of real estate broker Alex Yee (an' alleged co-conspirator), Yeung obtained loans in the name of Kenneth Ferrari, Yee’s friend and former co-worker, in order to repurchase the property at 1351 Third Street from a previous straw buyer who wanted out of the scheme. Yee and Yeung prepared a loan application for Ferrari that falsely stated that 1351 Third Street would be Ferrari’s primary residence and falsely inflated Ferrari’s income and present rental obligations.

Relying on this loan application, in December 2005, Long Beach Mortgage Company made two separate loans to Ferrari, the first for $664,000 and the second for $166,000. The loans were secured by the 1351 Third Street property. Ferrari paid the first mortgage payment himself, and Yee, who felt responsible for involving Ferrari, made the second payment. Yeung reimbursed Ferrari for the first month’s mortgage payment and also made the third month’s payment. She made no further payments.

According to the government’s expert witness, in March 2006, Long Beach Mortgage Company sold the two Ferrari loans to Long Beach Mortgage Securities Corporation, which then sold or transferred the loans, as part of a pool of loans, to the Long Beach Mortgage Trust 2006-2 (“Long Beach Trust”), a corporate entity. No evidence was submitted regarding the price Long Beach Mortgage Securities Corporation or the Long Beach Trust paid for the loans; the government witness at the evidentiary hearing testified that she did not know the amount.

The Long Beach Trust sold interests in the securitized pool of loans to investors. On December 4, 2006, after Ferrari defaulted on the loans, foreclosure proceedings commenced. The unpaid principal balance of the two loans at the time of Ferrari’s default was $829,382. Deutsche Bank National Trust Company (“Deutsche Bank”), as trustee for the Long Beach Trust, took title to the 1351 Third Street property at the foreclosure sale for a credit bid of $707,388.

In March 2008, Deutsche Bank sold the real estate to a third party for $395,000, receiving net proceeds (i.e., the sales price less the costs of the sale) of $363,863. B The scheme involving the Lam loans began in August 2006, when Yeung recruited straw buyer Dinh Lam to purchase Yeung’s home at 261 San Fernando Way, San Francisco, California. Lam’s loan application falsely listed 261 San Fernando Way as his primary residence and stated that Lam was the actual borrower, instead of Yeung.

The application also overstated Lam’s income and fraudulently stated that Lam had a Hang Seng Bank account with $480,168 in assets. In January 2007, J.P. Morgan Chase Bank, NA, loaned Lam $1,732,500, secured by the residential property at 261 San Fernando Way. According to the government’s expert witness, J.P. Morgan Chase Bank then sold the loan to J.P. Morgan Alternative Loan Trust 2007-A1 (“J.P. Morgan Trust”), where it was pooled with other loans, and interests in the trust were sold to investors. Again, no evidence was submitted regarding the price the J.P.

Morgan Trust paid for the loans, and the government witness testified that she did not know the details of the transaction, including the purchase amount. Lam defaulted on the J.P. Morgan Chase Bank loan in April 2007. The government presented evidence that the unpaid principal balance of the loan at the time of the default was the full loan amount, $1,732,500. The 261 San Fernando Way property was appraised in late 2007 and determined to have a market value of $1,290,000 (per an August 2007 appraisal) and $1,855,500 (per a September 2007 appraisal).

On October 1, 2007, Chase Home Finance LLC took title to the property at the foreclosure sale with a credit bid of $1,710,771. According to the bidding instructions, Chase Home Finance LLC foreclosed “in the name of HSBC Bank USA, National Association, as Trustee of J.P. Morgan Alternative Loan Trust 2007-A1.” After taking title to the property, Chase Home Finance LLC sold the property on February 27, 2008, to a third party for $1.5 million. The net proceeds to Chase Home Finance LLC were some $1,343,955. Concurrently with J.P.

Morgan Chase Bank making a $1.73 million loan to Lam secured by a first mortgage on 261 San Fernando Way, Cal State 9 Credit Union (“Cal State 9”) loaned Lam $467,500, secured by a second mortgage on the same property. Cal State 9’s junior lien was extinguished by the foreclosure in October 2007. Cal State 9 charged off the full value of the loan and associated expenses on November 9, 2007. The state subsequently put Cal State 9 into a conservator-ship, and National Credit Union Administration (“NCUA”) took over Cal State 9’s operations.

C Following Yeung’s conviction, the district court scheduled an evidentiary hearing to determine the amount of loss for the purposes of sentencing and restitution. In advance of the hearing, the probation office submitted a Presentence Investigation Report containing a chart of the victims’ losses, and the government filed a sentencing memorandum that defined and calculated loss pursuant to the U.S. Sentencing Guidelines.

Both the probation office and sentencing memorandum calculated the loss to the victims as the outstanding principal balance on the defaulted loans less any money recovered from a sale of the properties used as collateral for the loans. The government stated that it did not include “any interest owed, or any fees or costs associated with the foreclosures and short sales” in the loss amount. At the August 5, 2010 evidentiary hearing, the government argued that it had proved a loss amount of $1,375,502, which required a sixteen-level upward adjustment in Yeung’s offense level under § 2Bl.l(b)(l)(

I) of the U.S. Sentencing Guidelines (applying to losses that exceed $1 million but are less than $2.5 million). While conceding that it had no direct proof regarding how much the alleged victims (Long Beach Trust and J.P. Morgan Trust) paid to purchase the loans from the original lenders, the government argued, consistent with the Presentence Investigation Report and its own sentencing memorandum, that the loss was equivalent to the outstanding principal balance of the loan minus any proceeds recovered from the eventual sale of the collateral.

Following the hearing, the district court determined that the evidence was sufficient to allow it to estimate losses for purposes of the Sentencing Guidelines, even though — due to the “structure of the financial industry” during the period in which the fraudulent scheme was taking place — it was difficult to trace the money and it would be possible to get lost “in the maze of the financial transactions that occurred after the fraudulent loans were obtained and prior to the time that the losses were realized by the various sales and foreclosure sales.” The court adopted the government’s calculation of loss, but ultimately made a downward departure from the sentencing range of 87 to 108 months established under the Sentencing Guidelines and sentenced Yeung to 24 months in custody.

The court then calculated restitution. Using the same calculations developed to determine the amount of loss under § 2Bl.l(b)(l)(

I) of the Sentencing Guidelines, and with no mention of the MVRA, the court ordered restitution for the three fraudulent loans at issue here in the amount of $1,321,564. Specifically, for the Ferrari loans, the court ordered Yeung to pay $465,519 to Deutsche Bank, as trustee for the Long Beach Trust, an amount equal to the difference between the unpaid principal balance of the loans at the time of default ($829,382) and the net proceeds from the sale of the 1351 Third Street property to a third-party purchaser ($363,-863).

With respect to the Lam loans, the district court ordered Yeung to pay $388,544 to HSBC as trustee for the J.P. Morgan Trust, an amount equal to the difference between the unpaid principal of the J.P. Morgan Chase loan ($1,732,500) and the net proceeds from the sale of the 261 San Fernando Way property to a third-party purchaser ($1,343,955), and to pay $467,500 to NCUA, an amount equal to the full value of the Cal State 9 loan at the time of charge-off. On appeal, Yeung argues that the district court erred in all three restitution orders.

With respect to the Ferrari loans, Yeung argues that the Long Beach Trust is not a victim under MVRA and that the court erred in calculating both the amount of the Long Beach Trust’s loss and the amount it received from sale of the collateral. Yeung makes similar arguments with respect to the Lam loan from J.P. Morgan Chase Bank. Finally, with respect to the Lam loan from Cal State 9, Yeung argues that there was insufficient evidence that Cal State 9 took a charge-off on the loan before going into conservator-ship, or that the NCUA took over after Cal State 9 did so.

II We review the legality of a restitution order, including the district court’s valuation method, de novo. United States v. Davoudi, 172 F.3d 1130, 1134 (9th Cir. 1999). “[I]f the order is within the statutory bounds, we review the amount for abuse of discretion.” United States v. Phillips, 367 F.3d 846, 854 (9th Cir.2004). We review factual findings supporting an order of restitution for clear error. United States v. Gordon, 393 F.3d 1044, 1051 (9th Cir.2004).

Under the MVRA, 18 U.S.C. § 3663A, a court must order a defendant to make restitution to a victim of certain specified offenses without considering the defendant’s economic circumstances. See §§ 3663A(a)(1), 3664(f)(1)(A). The MVRA defines the word “victim” as “a person directly and proximately harmed as a result of the commission of an offense for which restitution may be ordered.” § 3663A(a)(2). Although a defendant’s conduct need not be the sole cause of the loss, “ ‘any subsequent action that contributes to the loss, such as an intervening cause, must be directly related to the defendant’s conduct.

The causal chain may not extend so far, in terms of the facts or the time span, as to become unreasonable.’” United States v. Peterson, 538 F.3d 1064, 1074 (9th Cir.2008) (quoting United States v. Gamma Tech Indus., Inc., 265 F.3d 917, 928 (9th Cir.2001)). The MVRA also provides guidance regarding the calculation of the restitution amount. 18 U.S.C. § 3663A(b). The basic rule is that the victim is entitled to be made whole.

See Gordon, 393 F.3d at 1053 (“ ‘[T]he primary and overarching goal of [the MVRA] is to make victims of crime whole, to fully compensate these victims for their losses and to restore these victims to their original state of well-being.’”) (emphasis in original) (quoting United States v. Simmonds, 235 F.3d 826, 831 (3d Cir.2000)); see also Hughey v. United States, 495 U.S. 411, 416, 110 S.Ct. 1979, 109 L.Ed.2d 408 (1990) (observing that the “meaning of ‘restitution’ is restoring someone to a position he occupied before a particular event”).

Accordingly, in the case of an offense resulting in damage to or loss of a victim’s property, the victim is entitled to return of the property, § 3663A(b)(1)(A), or if the return of the property is “impossible, impracticable, or inadequate,” the victim is entitled to the value of the property less “the value (as of the date the property is returned) of any part of the property that is returned,” § 3663A(b)(1)(B). The government has the burden of proving these values by a preponderance of the evidence. § 3664(e).

Using the framework set forth in § 3663A(b), we have developed some guidelines for calculating the restitution amount in a case involving a defendant’s fraudulent scheme to obtain secured real estate loans from lenders. Generally, district courts calculating a direct lender’s loss in this context begin by determining the amount of the unpaid principal balance due on the fraudulent loan, less the value of the real property collateral as of the date the direct lender took control of the property. United States v. Hutchison, 22 F.3d 846, 856 (9th Cir.1993), abrogated on other grounds by United States v.

Wells, 519 U.S. 482, 117 S.Ct. 921, 137 L.Ed.2d 107 (1997) (construing the VWPA); United States v. Smith, 944 F.2d 618, 625-26 (9th Cir.1991) (construing the VWPA).

Because restitution should address a victim’s “actual losses,” see Smith, 944 F.2d at 626, we have approved restitution awards that included other amounts in the calculation of loss, such as prejudgment interest (using the governmental loan rate), id., interest still due on the loan, Davoudi, 172 F.3d at 1136, and expenses associated with holding the real estate collateral that were incurred by the lender before it took title to the property, Hutchison, 22 F.3d at 856.

To calculate the value of the real property collateral “as of the date the property is returned,” § 3663A(b)(l)(B)(ii), courts use the value of the collateral “as of the date the victim took control of the property,” Davoudy 172 F.3d at 1134. The lender does not take control of the collateral merely by triggering the foreclosure process. See United States v. Gossi 608 F.3d 574, 578 (9th Cir.2010). Rather, the lender generally takes control on the date the lender either (1) receives the net proceeds from the sale of the collateral to a third party at the foreclosure sale, see United States v.

James, 564 F.3d 1237, 1246 (10th Cir.2009), or (2) takes title to the real estate collateral at the foreclosure sale, at which time “the new owner had the power to dispose of the property and receive compensation,” see Smith, 944 F.2d at 625. The direct lender’s losses may also be reduced by amounts recouped from resale of the loan or from other types of “return” of property. See, e.g., Hutchison, 22 F.3d at 856. These guidelines require some adjustment when a victim purchased a loan in the secondary market, that is, where the victim is the loan purchaser as opposed to the loan originator.

In the language of § 3663A, the loan itself is the “property” that has lost value due to the fraudulent conduct of the defendant. Because the value of that loan is not necessarily its unpaid principal balance, but may vary with the value of the collateral, the credit rating of the borrower, market conditions, or other factors, the loan purchaser may have purchased the loan for less than its unpaid principal balance. See United States v. Winstar Corp., 518 U.S. 839, 851-52, 116 S.Ct. 2432, 135 L.Ed.2d 964 (1996) (discussing factors causing loans to be worth less than their face value); Huffington v. T.C.

Grp., LLC, 637 F.3d 18, 20 (1st Cir.2011) (noting the value of mortgage-backed securities depends “on the cash flow generated by the mortgages and the prospects that the principal and interest will be paid”); Blueberry Land Co. v. Comm’r, 361 F.2d 93, 95 n. 7 (5th Cir.1966) (considering mortgage loans sold at a discount of 15% of the principal unpaid balance).

To calculate a victim’s restitution award using the outstanding principal balance of the loan, if the victim only paid a fraction of that amount to obtain the loan on the secondary market, would cause the victim to receive an amount exceeding its actual losses. See, e.g., United States v. Caputo, 517 F.3d 935, 943 (7th Cir.2008) (explaining that restitution for a machine with a list price of $100,000, but sold at a $20,000 discount, is $80,000, not the listed $100,000; and restitution for a machine that was billed at $100,000 but never actually paid for is zero).

Awarding such an amount would constitute plain error. See United States v. Rizk, 660 F.3d 1125, 1137 (9th Cir.2011) (“A district court may not order restitution such that victims will receive an amount greater than their actual losses; to do so is plain error.”); see also United States v. Boccagna, 450 F.3d 107, 117 (2d Cir.2006) (“[The court] cannot award the victim ‘a windfall,’ i.e., more in restitution than he actually lost.”).

Therefore, a district court calculating the loss suffered by a victim who purchased a fraudulent loan may begin by determining how much the victim paid for the fraudulent loan (or the value of the loan when the victim acquired it), less the value of the real property collateral as of the date the victim took control of the collateral property. As in a case with an original lender, the district court may consider other relevant factors that may increase or decrease the amount of the loan purchaser’s loss. See, e.g., Davoudi, 172 F.3d at 1136; Hutchison, 22 F.3d at 856; Smith, 944 F.2d at 626.

Although district courts possess a “ ‘degree of flexibility in accounting for a victim’s complete losses,’” remand is appropriate where the restitution award lacks an adequate evidentiary basis and the district court failed to explain its reasoning. United States v. Waknine, 543 F.3d 546, 557 (9th Cir.2008) (quoting Gordon, 393 F.3d at 1053); see also United States v.

Tsosie, 639 F.3d 1213, 1223 (9th Cir.2011); Peterson, 538 F.3d at 1077-78 (affirming restitution order notwithstanding lack of factual findings by the district court where it was clear the district court relied on expert’s declaration reporting specific loss to victims on each property). A We begin by considering Yeung’s arguments with respect to the two loans issued by Long Beach Mortgage Company to Ferrari.

As noted above, the district court awarded Deutsche Bank, as trustee for the Long Beach Trust, an amount equal to the difference between the unpaid principal balance of the loans at the time of Ferrari’s default and the net proceeds from the sale of the collateral to a third party, approximately sixteen months after Deutsche Bank obtained title to it in the foreclosure. Yeung argues that the district court erred in awarding restitution to the Long Beach Trust because it is not the “victim” for purposes of the MVRA. We disagree.

Here, the evidence showed that the Long Beach Trust purchased the two loans originally issued by Long Beach Mortgage Company before Yeung’s fraud had come to light. Further, the evidence showed that, due to Yeung’s conduct, the borrower for each loan had misrepresented his financial ability to repay the loan. Because the Long Beach Trust purchased the loan without an awareness of its true value due to Yeung’s fraud, the district court could reasonably conclude that Yeung’s fraudulent conduct proximately harmed the Long Beach Trust.

See, e.g., James, 564 F.3d at 1243 (approving a restitutionary award to loan purchaser). Therefore, the district court did not abuse its discretion in deeming the Long Beach Trust to be a victim for purposes of the MVRA and in making an award to Deutsche Bank as the trustee for the Long Beach Trust. Yeung next argues that the district court erred in calculating the amount of restitution owed with respect to these loans.

The district court did not explain its reasons for calculating that Deutsche Bank was entitled to the difference between the $829,382 unpaid principal balanee of the loan and $363,863, the net proceeds that Deutsche Bank (as trustee for the Long Beach Trust) received from its sale of the collateral to a third party in 2008, more than sixteen months after it acquired title. The district court’s calculation raises two concerns.

First, the district court did not make a finding that the Long Beach Trust paid an amount equal to the unpaid principal balance of the loans when purchasing the loans, and the government witness acknowledged that she did not have any information on that point. In fact, there is no evidence in the record establishing the value of the loans at the time they were acquired by the Long Beach Trust. We disagree with the government’s assertion that documents filed with the Securities and Exchange Commission show that the loans were purchased for the amount of unpaid principal.

The government concedes that these documents were never entered into evidence; accordingly, they cannot be used to support the government’s claims on appeal. See, e.g., Bareamerica Int’l USA Trust v. Tyfield Imps., Inc., 289 F.3d 589, 593 n. 4 (9th Cir.2002). In the absence of any evidence as to value of the loans at the time the victim acquired them, we cannot conclude that the district court’s restitutionary award was free from error.

As explained above, if the Long Beach Trust purchased the loans for less than the unpaid principal balance of the loan, it is not entitled to receive a windfall by receiving an award of the loans’ full amount. See Rizk, 660 F.3d at 1137. Second, Yeung argues that the district court erred in determining that the value of the collateral was $363,863, the net proceeds that Deutsche Bank (as trustee for the Long Beach Trust) received from its sale of the collateral to a third party in 2008, more than sixteen months after it acquired title. We agree.

The property returned must be valued “as of the date the victim took control of the property,” Davoudi 172 F.3d at 1134, which generally occurs when the victim “received title to the property and the corresponding ability to sell it for cash,” Smith, 944 F.2d at 625. In this case, Deutsche Bank took control of the real property collateral at the foreclosure sale in 2006, when it acquired title to the real property via a credit bid.

Therefore, the court cannot rely on the subsequent sales price of the real property unless it provides reasons why that sales price reflects the value of the real property on the date Deutsche Bank took control of the property. See Waknine, 543 F.3d at 556-57. We reject Yeung’s further argument that Deutsche Bank’s credit bid of $707,388 dis-positively determines the value of the collateral at the time of the foreclosure sale. A lender’s credit bid may not reflect the value of the collateral in all circumstances.

See Green, 648 F.3d at 584 (holding that where the fraud involved obtaining a loan worth “far more than the market value of the property, ---- [u]sing a credit bid based on the fraudulently inflated loan amount would surely understate the actual loss”); cf. Boccagna, 450 F.3d at 109 (holding that “the nominal sale price of property with a higher fair market value cannot be used to calculate offset value because such a calculation impermissibly awards a victim restitution in excess of its compensable loss”).

Therefore, the court would have to provide some explanation as to why the value of the real property collateral in a particular case was equal to the victim’s credit bid at the foreclosure sale. The district court did not do so here. Even where the district court fails to make pertinent factual findings, we may uphold a restitution order when the basis of the district court’s calculations is clear. Cf Peterson, 538 F.3d at 1077-78. But the district court’s determinations of loss for Sentencing Guidelines purposes here are insufficient to support a restitutionary award under the MYRA.

We have previously held that a loss determination used for Sentencing Guidelines purposes could not be used to determine the amount of restitution, given the different methods of calculation and different purposes of the calculation. See Gossi 608 F.3d at 581-82; Gordon, 393 F.3d at 1052 n. 6; United States v. Catherine, 55 F.3d 1462, 1464-65 (9th Cir.1995). Nor can it be used in this case. As noted above, the government and probation office calculated loss to the victims as the outstanding principal balance on the defaulted loans less any money recovered from a sale of the real property collateral.

This is an appropriate approach for making an estimate of loss under the Guidelines, where the district court need not consider a victim’s actual loss but may use evidence of the defendant’s intended loss, or even the gain realized by the defendant, U.S.S.G. § 2B1.1 cmt. n. 3(A), (B), but for the reasons explained above, it does not provide a sufficient basis to establish the actual loss suffered by the victims here.

Because the district court did not provide reasoning to explain its determination of loss for purposes of § 3663A(b)(1)(B), and because the district court did not determine the value of the collateral at the time Deutsche Bank took title, we must remand for the district court to recalculate and provide its reasoning for this award. See Waknine, 543 F.3d at 557. B We apply similar reasoning to Yeung’s argument that the district court erred in determining that the J.P. Morgan Trust was a victim, and in calculating the amount of restitution.

As noted above, the district court awarded $388,544 to HSBC, as trustee for the J.P. Morgan Trust, an amount equal to the difference between the unpaid principal balance of the J.P. Morgan Chase loan at the time of Lam’s default and the net proceeds from the sale of the collateral to a third party approximately five months after the foreclosure. For the reasons explained above, we reject Yeung’s argument that the J.P. Morgan Trust was not a victim for purposes of the MVRA. See 18 U.S.C. § 3663A(a)(2). The evidence showed that the J.P. Morgan Trust purchased the loan originally issued by J.P.

Morgan Chase Bank before Yeung’s fraud had come to light. Further, the evidence showed that, due to Yeung’s conduct, Lam (the borrower for the loan) had misrepresented his financial ability to repay the loan. Therefore, Yeung’s fraudulent conduct proximately harmed the J.P. Morgan Trust, and it was a “victim” for purposes of the MVRA. However, as with the Ferrari loans, the district court did not explain its reasoning or point to any evidence establishing that the value of the loans held by the J.P. Morgan Trust was equal to the unpaid principal balance of the loan.

Accordingly, we must remand to the district court to recalculate this value or explain its reasoning. See Waknine, 543 F.3d at 557. We also agree with Yeung that the value of the real property collateral returned to the victim must be measured as of the date that Chase Home Finance LLC (in the name of HSBC, as trustee for the J.P. Morgan Trust) took title to the collateral at the foreclosure sale. Therefore, the district court erred in relying on the value of the collateral when it was sold to a third party several months later. See Gossi, 608 F.3d at 578; Davoudi, 172 F.3d at 1134—35.

The record does not establish the value of the collateral at the time Chase Home Finance LLC took title (and the amount of the credit bid is not dispositive), so we must remand for a redetermination of this issue as well. C Finally, we turn to the Cal State 9 loan to Lam. The district court did not abuse its discretion in holding that NCUA was a “victim” for purposes of the MVRA. The evidence presented at sentencing established that the initial lender, Cal State 9, lost the full amount of the loan and that its insurer, NCUA, took over the loss when Cal State 9 subsequently went into conservatorship.

To the extent NCUA acquired all of Cal State 9’s loss, NCUA would qualify as a victim. See Smith, 944 F.2d at 622 (holding that the insurance company, which had insured the loan institution’s accounts, although not directly victimized by the defendant’s actions, could receive restitution under the MVRA because it had “acquired the claims of a defunct savings and loan”); of. O’Melveny & Myers v. FDIC, 512 U.S. 79, 86, 114 S.Ct. 2048, 129 L.Ed.2d 67 (1994) (“[T]he FDIC as receiver ‘steps into the shoes’ of the failed S & L.”).

Yeung also argues that the district court erred by relying on hearsay evidence that Cal State 9 wrote off the loan as uncollectible before going into conservatorship. This argument fails. The Federal Rules of Evidence, including the rule against hearsay, do not apply to sentencing hearings, see United States v. Littlesun, 444 F.3d 1196, 1199 (9th Cir.2006); see also 18 U.S.C. § 3661, and therefore the district court did not err in relying on hearsay in ordering restitution payable to NCUA.

We hold that the district court properly ordered Yeung to pay restitution to NCUA for $467,500 based on losses proximately resulting from her criminal conduct. Ill District courts possess a great deal of flexibility in applying the MVRA to unique factual circumstances and conducting the calculation required by § 3663A(b)(1)(B). See Gordon, 393 F.3d at 1053-54. But where we cannot discern from the record the district court’s reasoning regarding the key determinations required by the MVRA, we cannot uphold the award. Waknine, 543 F.3d at 557. Here, there was no evidence regarding what the victims (the J.P.

Morgan Trust and Long Beach Trust) had paid for the loan, and the district court did not explain its reasoning as to why the unpaid principal balance of the loan could be used to establish the amount of the victims’ losses. Nor was there any evidence regarding the value of the collateral when the victims asserted control (i.e., the date the trustees for the trusts took title to the property), and the district court did not explain its reasoning as to why the sales price of the property months or years later established that value.

Under these circumstances, we must vacate the restitution order with respect to the Long Beach Trust and the J.P. Morgan Trust awards and remand to the district court for a new determination of loss. AFFIRMED IN PART, VACATED AND REMANDED IN PART. . We affirm Yeung's conviction in a memorandum disposition filed concurrently with this opinion. In this opinion, we address only Yeung’s challenges to the restitution award. . A “credit bid” is "a bid that offers to cancel the outstanding principal, interest, and related fees in return for title to the property.” United States v.

Green, 648 F.3d 569, 584 (7th Cir.2011). Pursuant to California Civil Code § 2924h(b), the present beneficiary of the deed of trust for the property under foreclosure can bid up to the full amount of the debt owed, including the trustee’s fees and expenses. . The prior restitution statute, the Victim and Witness Protection Act of 1982 ("VWPA”), required courts to consider the economic circumstances of the defendant prior to ordering restitution, and the granting of restitution was discretionary, not mandatory. See 18 U.S.C. § 3663.

"With these exceptions, the two statutes are identical in all important respects, and courts interpreting the MVRA may look to and rely on cases interpreting the VWPA as precedent.” See Gordon, 393 F.3d at 1048. . The relevant portion of the MVRA reads a follows: (

b) The order of restitution shall require that such defendant— (1) in the case of an offense resulting in damage to or loss or destruction of property of a victim of the offense— (

A) return the property to the owner of the property or someone designated by the owner; or (

B) if return of the property under sub-paragraph (

A) is impossible, impracticable, or inadequate, pay an amount equal to— (

i) the greater of— (

I) the value of the property on the date of the damage, loss, or destruction; or (II) the value of the property on the date of sentencing, less (ii) the value (as of the date the property is returned) of any part of the property that is returned!)] § 3663A(b)(l). . In her reply brief, Yeung argues for the first time that the district court erred in concluding that the Long Beach Trust was a victim because any losses it suffered were the result of the general collapse of the housing market, rather than Yeung's fraud. Arguments raised for the first time in a reply brief are waived. United States v.

King, 257 F.3d 1013, 1029 n. 5 (9th Cir.2001). In any event, this argument fails. Although an intervening cause unrelated to the offense conduct may cut off the causal chain between the defendant's conduct and the victim's loss, see United States v. Brock-Davis, 504 F.3d 991, 1000-01 (9th Cir. 2007), here Yeung created the circumstances under which the harm or loss occurred through her use of false information that induced the Long Beach Trust to purchase the loan.

Because the Long Beach Trust’s loss is directly related to Yeung’s offense, the declining value of the real estate collateral, even if attributable to general financial conditions, does not disrupt the causal chain, see Peterson, 538 F.3d at 1077, and the victims of the fraud are entitled to restitution. . We reject the government’s argument that the loan's originator, J.P. Morgan Chase Bank, transferred the loan to its own subdivision rather than selling it to a separate entity, because it is contrary to the evidence presented at sentencing, where the government’s expert testified that J.P.

Morgan Chase Bank sold the Lam loan to the J.P. Morgan Trust, and that Chase Home Finance LLC handled the foreclosure proceedings on behalf of the trustee, HSBC Bank, for the J.P. Morgan Trust.