United States v. David Lonich

23 F.4th 881

• Court: Court of Appeals for the Ninth Circuit
• Decided: January 10, 2022
• Docket: 18-10298
• Status: Published

Opinion

FOR PUBLICATION

UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT

UNITED STATES OF AMERICA, Nos. 18-10298 Plaintiff-Appellee, 18-10395

v. D.C. No. 3:17-cr-00139-SI-3 DAVID LONICH, Defendant-Appellant.

UNITED STATES OF AMERICA, Nos. 18-10299 Plaintiff-Appellee, 18-10408

v. D.C. No. 3:17-cr-00139-SI-2 BRIAN SCOTT MELLAND, Defendant-Appellant.

UNITED STATES OF AMERICA, Nos. 18-10300 Plaintiff-Appellee, 18-10394

v. D.C. No. 3:14-cr-00139-SI-2 DAVID LONICH, Defendant-Appellant. 2 UNITED STATES V. LONICH

UNITED STATES OF AMERICA, Nos. 18-10301 Plaintiff-Appellee, 18-10407

v. D.C. No. 3:14-cr-00139-SI-4 BRIAN SCOTT MELLAND, Defendant-Appellant.

UNITED STATES OF AMERICA, Nos. 18-10303 Plaintiff-Appellee, 18-10405

v. D.C. No. 3:14-cr-00139-SI-3 SEAN CLARK CUTTING, Defendant-Appellant.

UNITED STATES OF AMERICA, Nos. 18-10304 Plaintiff-Appellee, 18-10390

v. D.C. No. 3:17-cr-00139-SI-1 SEAN CLARK CUTTING, Defendant-Appellant. OPINION

Appeal from the United States District Court for the Northern District of California Susan Illston, District Judge, Presiding

Argued and Submitted February 10, 2021 San Francisco, California

Filed January 10, 2022 UNITED STATES V. LONICH 3

Before: Andrew D. Hurwitz and Daniel A. Bress, Circuit Judges, and Clifton L. Corker, * District Judge.

Opinion by Judge Bress

SUMMARY **

Criminal

The panel affirmed Sean Cutting’s, Brian Melland’s, and David Lonich’s convictions, but vacated their sentences and remanded for resentencing, in a complex case arising from fraudulent schemes concerning bank loans and real estate in Sonoma County, California.

The panel held the Sixth Amendment’s Speedy Trial Clause was not violated. Defendants claimed a Speedy Trial Clause violation as to all charges first brought in the October 2016 superseding indictment. Defendants then argued this court should reverse their convictions as to the charges in the original March 2014 indictment because of “prejudicial spillover” from evidence used to prove the charges in the allegedly unconstitutional superseding indictment. The panel had no occasion to consider defendants’ “prejudicial spillover” theory because the panel held that the government’s decision to file new charges in the superseding indictment did not infringe defendants’ Speedy Trial Clause

* The Honorable Clifton L. Corker, United States District Judge for the Eastern District of Tennessee, sitting by designation. ** This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader. 4 UNITED STATES V. LONICH

rights. As to the first factor in the balancing test set forth in Barker v. Wingo, 407 U.S. 514 (1972), the length of the delay, the parties disagreed on when defendants’ Speedy Trial Clause rights attached for the new charges first brought in the superseding indictment. Defendants argued the original indictment should be used as the start date for the new charges in the superseding indictment. The government contended the date it filed the superseding indictment should be used. The panel did not need to resolve that debate because it concluded that, even assuming the clock started at the time of the original indictment, there was no Speedy Trial Clause violation because the delay caused no relevant prejudice to defendants.

Defendants challenged the jury instructions on the money laundering (18 U.S.C. § 1957) and misapplication of bank funds (18 U.S.C. § 656) charges, contending that the instructions’ overarching definition of “knowingly” conflicted with the required mental states for the two charged offenses. The panel held that the district court’s general “knowingly” instruction was permissible and that defendants in any event did not show prejudice from the instruction.

Melland argued that there was insufficient evidence to support his conviction for bribery by a bank employee (18 U.S.C. § 215(a)(2)), which was based on his securing a $50,000 investment in Melland’s energy-drink start-up. The panel held that, as the parties effectively agree, the district court appropriately stated the law when it instructed the jury that, to find Melland “acted corruptly,” as required under § 215(a)(2), the jury must determine he “intend[ed] to be influenced or rewarded in connection with any business or transaction of” a financial institution. Noting that the UNITED STATES V. LONICH 5

circumstantial evidence was plentiful, the panel held that there was sufficient evidence to support the conviction.

Lonich argued that there was insufficient evidence to support his conviction for attempted obstruction of justice (18 U.S.C. § 1512(c)(2)) by encouraging a straw buyer to mislead the grand jury about his role in a scheme to gain control of a real estate development. The panel held that § 1512(c)(2) requires a showing of nexus to an official proceeding, but rejected Lonich’s argument that no reasonable jury could have found the required nexus here. Noting that neither party disputes using a “consciousness of wrongdoing” mens rea requirement for purposes of evaluating the sufficiency of the evidence, the panel held that a reasonable jury could find that the government met its burden of proof in demonstrating Lonich’s criminal intent.

The panel held that defendants’ sentences must be vacated. The district court applied several enhancements that dramatically increased defendants’ recommended Guidelines sentencing ranges. These enhancements were premised on a critical factual finding: that defendants caused Sonoma Valley Bank (SVB) to fail, making defendants responsible for associated losses. Addressing the standard of proof that the government was required to meet to demonstrate whether defendants caused SVB to fail, the panel focused on factors five and six of the non-exhaustive factors set forth in United States v. Valencia, 222 F.3d 1173 (9th Cir. 2000). Given the extremely disproportionate sentences that the disputed enhancements produced, the panel held that a clear and convincing evidence standard applies to the factual underpinnings for these enhancements. The panel concluded that the government did not demonstrate by clear and convincing evidence that defendants caused SVB to fail, where the district court made 6 UNITED STATES V. LONICH

no independent findings about the cause of the bank’s collapse beyond adopting the Presentence Investigation Reports (PSRs) and rejecting defendants’ objections without explanation, and neither the PSRs nor the additional materials the government now cites sufficiently show that defendants were responsible for SVB failing, especially given indications in the record that other factors internal and external to the bank may have contributed to the bank’s collapse. This meant that the government did not sufficiently support defendants’ 20-level loss enhancement under U.S.S.G. § 2B1.1(b)(1)(K). The panel wrote that its determination that the government did not adequately prove defendants caused SVB to fail means that enhancements under U.S.S.G. § 2B1.1(b)(2)(A)(i) (ten or more victims) and § 2B1.1(b)(17)(B)(i) (jeopardizing the safety and soundness of a financial institution) are infirm as well. The panel wrote that the same is true of defendants’ approximately $20 million restitution orders, which were likewise premised on the government’s theory that defendants caused the bank to fail. The panel vacated defendants’ sentences and remanded for resentencing on an open record.

The panel rejected defendants’ remaining challenges to their convictions in a memorandum disposition. UNITED STATES V. LONICH 7

COUNSEL

George C. Harris (argued), The Norton Law Firm PC, Oakland, California, for Defendant-Appellant David Lonich.

Juliana Drous (argued), Law Office of Juliana Drous, San Francisco, California, for Defendant-Appellant Brian Scott Melland.

Steven J. Keoninger (argued), Assistant Federal Public Defender; Steven G. Kalar, Federal Public Defender; Office of the Federal Public Defender, San Francisco, California; for Defendant-Appellant Sean Clark Cutting.

Francesco Valentini (argued), Trial Attorney; Brian C. Rabbitt, Acting Assistant Attorney General; United States Department of Justice, Criminal Division, Appellate Section, Washington, D.C.; Adam A. Reeves and Robert David Rees, Assistant United States Attorneys; David L. Anderson, United States Attorney; United States Attorney’s Office, San Francisco, California; for Plaintiff-Appellee. 8 UNITED STATES V. LONICH

OPINION

BRESS, Circuit Judge:

This complex criminal appeal arises from fraudulent schemes concerning bank loans and real estate in Sonoma County, California. The three defendants, Sean Cutting, Brian Melland, and David Lonich, appeal their convictions and sentences, raising numerous issues for our review.

We affirm defendants’ convictions. Among other things, we hold that the government did not infringe defendants’ rights under the Sixth Amendment’s Speedy Trial Clause; that the district court’s jury instructions for money laundering and misapplication of bank funds do not require reversal; and that sufficient evidence supported Melland’s conviction for bribery by a bank employee and Lonich’s conviction for attempted obstruction of justice.

However, we vacate defendants’ sentences and remand for resentencing. The defendants’ advisory Sentencing Guidelines ranges increased substantially based on sentencing enhancements that hinged on finding that defendants caused the Sonoma Valley Bank to fail. We hold that because the government has not sufficiently demonstrated that defendants caused the bank’s failure, the enhancements are not supported by the record. 1

1 In a concurrently filed opinion, we reject a third party’s ancillary challenge to the district court’s criminal forfeiture order. See United States v. 101 Houseco, LLC, No. 18-10305 (9th Cir. 2021). UNITED STATES V. LONICH 9

I. Facts and Procedural History

A

This case involves two overarching fraudulent schemes involving bank officers, a real estate developer, and the developer’s lawyer. Defendants Sean Cutting and Brian Melland were officers at Sonoma Valley Bank (SVB). Cutting was SVB’s Chief Lending Officer between 2005 and 2011. He joined its board of directors in 2008 and became its CEO in 2009. Melland was a commercial loan officer at SVB. Bijan Madjlessi was a real estate developer who died shortly after being indicted in this case. Defendant David Lonich worked as Madjlessi’s in-house lawyer between 2009 and 2012.

Defendants’ extensive fraudulent schemes took place over many years. In the first scheme, which we will call the “legal lending limit scheme,” Cutting and Melland conspired with Madjlessi and Lonich to induce SVB to approve, over a period of years, millions of dollars in bank loans to Madjlessi and entities he controlled. These loans exceeded SVB’s legal lending limit—the maximum amount that California law permits a bank to lend a borrower or his affiliates.

Cutting and Melland recommended that SVB approve these loans without disclosing to the bank’s loan committee that Madjlessi was the beneficiary. Madjlessi then used the fraudulently obtained loans to pay the interest on preexisting SVB loans. In one instance, Melland secured a loan for Madjlessi only after Madjlessi (through a $50,000 payment from his wife) agreed to invest in Melland’s side business, an energy drink start-up known as Magnus Innovations Group. 10 UNITED STATES V. LONICH

In connection with this scheme to keep Madjlessi’s businesses afloat through fraudulent loans, Cutting and Melland also conspired to conceal from the Federal Deposit Insurance Corporation (FDIC) SVB’s overall financial exposure to Madjlessi. Ultimately, Cutting and Melland enabled Madjlessi and his related entities to receive over $35 million in loans from SVB (although the government did not claim all these loans were fraudulent).

In the second scheme, which we will call the “101 Houseco scheme,” Madjlessi and Lonich conspired with Cutting and Melland to gain control of Park Lane Villas East (PLV East), a Madjlessi real estate development in Santa Rosa, California. By March 2009, Madjlessi had defaulted on a separate $32 million loan from another bank, IndyMac, secured by PLV East. IndyMac was in FDIC conservatorship, and the FDIC scheduled a sale of Madjlessi’s defaulted note at an auction.

FDIC rules prohibited Madjlessi and his related entities from participating in the auction. Nevertheless, the defendants used a straw buyer, James House, and a sham entity, 101 Houseco, LLC, to buy the IndyMac note at the auction and thereby secure control of PLV East. House, the straw buyer, was a contractor to whom Madjlessi owed around $200,000. House took part in the scheme so that Madjlessi would pay the $200,000.

The defendants created 101 Houseco, LLC solely for bidding on the note, naming House as its owner on paper. Lonich also had House fax to DebtX, the company managing the FDIC auction, an eligibility certification in which House falsely certified that he was not using the auction to “benefit directly or indirectly” anyone “who otherwise would be ineligible to purchase assets from the FDIC.” To fund the deposit on the sale, Lonich had Melland transfer $100,000 UNITED STATES V. LONICH 11

from Madjlessi’s daughter’s account into House’s business account, further concealing Madjlessi’s role from DebtX. To finance the rest of the purchase of the note, Melland and Cutting fraudulently secured for 101 Houseco a $5.4 million loan from SVB. Using the loan proceeds, 101 Houseco successfully bid $4.2 million to obtain Madjlessi’s defaulted IndyMac note, which had a face value exceeding $27 million.

After the auction, SVB’s loans to House continued under the guise of allowing House to construct the Park Lane Villas. SVB continued to increase the loan amount until it reached $9.4 million. Of this $9.4 million, about $4.5 million was passed to Madjlessi through one of his construction companies. Madjlessi kept his side of the bargain with House, paying him the $200,000 owed for past contracting work. In line with the plan, Lonich later transferred effective control of 101 Houseco to Madjlessi. Lonich became 101 Houseco’s sole manager. Madjlessi’s wife became the beneficiary of the trust that held a 99% interest in 101 Houseco.

Madjlessi and Lonich wanted to refinance PLV East through Fannie Mae or Freddie Mac programs for multi- family housing. In the meantime, however, Fannie Mae had repossessed several condos in PLV East and was selling them at auction. Fannie Mae preferred buyers who would occupy the condos over outside investors. To get around Fannie Mae’s preferences, Madjlessi and Lonich used straw buyers—including House, Madjlessi’s personal assistant, and the assistant’s two sons—to purchase the condos. The straw buyers then transferred the units to 101 Houseco. 12 UNITED STATES V. LONICH

Madjlessi and Lonich arranged the financing for these straw purchases through Cutting and Melland. Lonich drafted asset verification letters falsely stating that the buyers had sufficient assets with SVB to fund the purchases in full. Cutting and Melland then gave these letters back to Lonich on SVB letterhead with Cutting’s signature.

After the FDIC and the California Division of Financial Institutions (DFI) examined SVB, DFI gave SVB the lowest rating it could give a bank without closing it. In August 2010, California’s Commissioner of Financial Institutions seized control of SVB, ordering that the bank be liquidated and its assets turned over to the FDIC.

When federal agents interviewed House, he admitted wrongdoing and agreed to cooperate. In subsequent secretly recorded meetings, Lonich advised House on how he should testify before a grand jury. House later pleaded guilty to bank and wire fraud charges for making false statements in connection with the 101 Houseco application for the SVB loan and the bid on the FDIC-owned note.

B

In March 2014, a federal grand jury returned a 29-count indictment against Cutting, Melland, Lonich, and Madjlessi for the 101 Houseco scheme (Madjlessi died soon after). In October 2016, the grand jury returned a superseding indictment adding charges for defendants’ legal lending limit scheme and their concealing from the FDIC SVB’s risk exposure to Madjlessi.

In the fall of 2017, and after a 31-day jury trial, the jury convicted defendants on nearly all counts. This chart summarizes the convictions: UNITED STATES V. LONICH 13

Count 18 Offense Defendants U.S.C. § 1 371 Conspiracy to commit Cutting, bank fraud Lonich, and Melland 2 1344 Bank Fraud Cutting, Lonich, and Melland 3–4 1005 Making a false bank Melland entry for certain (Cutting Madjlessi-related acquitted) loans 5 1005 Making a false bank Cutting and entry for certain other Melland Madjlessi-related loans 6 371 Conspiracy to make Cutting and false statements to the Melland FDIC 7 656 Misapplication of Cutting and bank funds Melland 8 1007 Making a false Cutting and statement to the FDIC Melland 9 215 Receiving a gift for Melland procuring loans 10 1349, Conspiracy to commit Cutting, 1343 wire fraud Lonich, and Melland 14 UNITED STATES V. LONICH

11–15 1343 Wire Fraud for the Cutting, 101 Houseco scheme Lonich, and Melland 19–30 1957 Money laundering for Cutting, transferring loan Lonich, and proceeds House Melland controlled to Madjlessi 32–36 1005 Making a false bank Cutting, entry for Cutting’s Lonich, and false asset- Melland verification letters relating to purchases of PLV East condos 37 1512(c) Attempted obstruction Lonich of justice

All three defendants were acquitted of Count 16, a wire fraud charge. The government withdrew Counts 17, 18, and 31.

The advisory Sentencing Guidelines range adopted by the district court for both Cutting and Melland was 235–293 months. The Guidelines range for Lonich was 292–365 months. The district court sentenced Cutting and Melland each to 100 months in prison, and Lonich to 80 months. The district court also ordered approximately $20 million in restitution and the forfeiture of PLV East.

In this appeal, the defendants raise many challenges to their convictions and sentences. We review the district court’s legal conclusions de novo. See United States v. Gregory, 322 F.3d 1157, 1160 (9th Cir. 2003). And we UNITED STATES V. LONICH 15

review its factual determinations for clear error. See id. at 1161.

II. The Speedy Trial Clause

Defendants’ lead argument is that their convictions are invalid under the Sixth Amendment’s Speedy Trial Clause. Defendants claim a Speedy Trial Clause violation as to all charges first brought in the October 2016 superseding indictment. Defendants then argue we should reverse their convictions as to the charges in the original March 2014 indictment because of “prejudicial spillover” from evidence used to prove the charges in the allegedly unconstitutional superseding indictment.

We have no occasion to consider defendants’ “prejudicial spillover” theory because we hold that the government’s decision to file new charges in the superseding indictment did not infringe defendants’ Speedy Trial Clause rights.

Some additional background on the proceedings below is necessary to understand our resolution of the Speedy Trial Clause issue. In March 2014, the grand jury indicted defendants on 29 charges related to the 101 Houseco scheme. In May 2016, the district court ordered the government to file any superseding indictment by October 28, 2016, and set a trial date of March 2017.

On October 27, 2016, the day before the court’s deadline, the grand jury returned the superseding indictment. Besides the charges for the 101 Houseco scheme from the original indictment, the superseding indictment included new allegations for the legal lending limit scheme and 16 UNITED STATES V. LONICH

defendants’ related fraud on the FDIC regarding SVB’s financial exposure to Madjlessi.

Criminal defendants enjoy certain protections from post- indictment delays. Some are statutory, codified in the Speedy Trial Act. See 18 U.S.C. § 3161; see also, e.g., United States v. Murillo, 288 F.3d 1126, 1131 (9th Cir. 2002). Others are constitutional, rooted in the Sixth Amendment’s Speedy Trial Clause. See, e.g., Barker v. Wingo, 407 U.S. 514, 530–33 (1972); Murillo, 288 F.3d at 1131. “The specific time limits set by the Speedy Trial Act are . . . different from the broader limits of the [S]ixth [A]mendment” because the constitutional analysis “is governed by the more flexible consideration of prejudice caused by delay.” Murillo, 288 F.3d at 1131 (quoting United States v. Pollock, 726 F.2d 1456, 1460 n.5 (9th Cir. 1984)).

Defendants did not assert either below or on appeal an independent Speedy Trial Clause or Speedy Trial Act violation for the 101 Houseco scheme charges that were in the original indictment. The defendants had instead long agreed to the approximately 3-year period between the original March 2014 indictment and the original March 2017 trial date, based on their need to prepare a defense against the indictment’s complex allegations. See United States v. Aguirre, 994 F.2d 1454, 1457 (9th Cir. 1993) (“The Speedy Trial Clause primarily protects those who assert their rights, not those who acquiesce in the delay—perhaps hoping the government will change its mind or lose critical evidence.”). Defendants contended below, however, that there was a Speedy Trial Clause violation based on the new charges first brought in the October 2016 superseding indictment UNITED STATES V. LONICH 17

associated with the legal lending limit scheme and the related fraud on the FDIC. 2

The district court initially granted defendants’ motion asserting a Speedy Trial Clause violation and dismissed the superseding indictment without prejudice. The court calculated the period of delay using the March 2014 original indictment date as the starting point and March 2017 (the original trial date) as the end date. This three-year delay, the district court held, presumptively prejudiced the defendants.

The district court also found that “[t]he government could have filed all of the charges contained in the superseding indictment when it filed the original indictment, and the government has not adequately explained why it did not do so.” According to the district court:

This is not a case where new evidence has come to light that prompted the need to supersede. Rather, the government simply chose to seek indictment on some of the charges of which it was aware, while holding back on others. Then later—much later, some 31 months later—it decided to supersede to add the charges it had been holding back, including a new conspiracy and new substantive charges. Although the Court does not find bad faith on the part of the government, the Court does find that the

2 Defendants do not argue there was excessive pre-indictment delay under the Fifth Amendment. See United States v. Corona-Verbera, 509 F.3d 1105, 1112 (9th Cir. 2007) (“The Fifth Amendment guarantees that defendants will not be denied due process as a result of excessive pre-indictment delay.” (quoting United States v. Sherlock, 962 F.2d 1349, 1353 (9th Cir. 1989))). 18 UNITED STATES V. LONICH

government acted deliberately and intentionally with regard to charging the new crimes added in the superseding indictment.

The district court further found that the government’s delay had prejudiced the defendants. Citing the government’s production of millions of pages of additional documents, the poor quality of its electronic document production, its general discovery delays, and the fact that the broadened charges would require defendants to re-review documents previously produced, the district court found that the government had put defendants in an “untenable position.” If the new charges remained in this case, “the [March 2017] trial date would almost certainly need to be continued in order to allow the defense time to prepare.” The district court dismissed the superseding indictment without prejudice, allowing the government to refile the new charges in a new case.

The government moved for reconsideration, noting that Speedy Trial Clause violations require dismissals with prejudice. See, e.g., Strunk v. United States, 412 U.S. 434, 440 (1973); United States v. Saavedra, 684 F.2d 1293, 1297 (9th Cir. 1982). The government therefore suggested that the district court’s order would more properly be grounded in Federal Rule of Criminal Procedure 48, which allows dismissals without prejudice. See United States v. Yuan Qing Jiang, 214 F.3d 1099, 1103 (9th Cir. 2000). Rule 48 permits a court to dismiss an indictment “if unnecessary delay occurs in: (1) presenting a charge to a grand jury; (2) filing an information against a defendant; or (3) bringing a defendant to trial.” Fed. R. Crim. P. 48.

The district court granted the government’s motion for reconsideration, explaining that it had “clearly intended that UNITED STATES V. LONICH 19

the dismissal be without prejudice, and the Court erred by not grounding its order in Rule 48.” The court reiterated its previous finding that “the government had unnecessarily delayed in seeking the superseding indictment.” And it again noted “the prejudice defendants would experience if forced to proceed to trial in March 2017 on the new charges, given the technical problems with the government’s discovery production as well as the new discovery produced regarding the new charges.” The court then dismissed the superseding indictment without prejudice under Rule 48.

In March 2017, in response to the district court’s order, the government filed a new action against defendants based on an entirely new indictment concerning the legal lending limit scheme. Not wanting two trials, defendants requested that the district court consolidate the two cases. Defendants also requested that the district court vacate the March 2017 trial date so they would have sufficient time to prepare for a consolidated trial. The district court granted defendants’ request and re-set the trial for October 2017.

The Sixth Amendment provides that “[i]n all criminal prosecutions, the accused shall enjoy the right to a speedy and public trial, by an impartial jury . . . .” U.S. Const. amend. VI. The Speedy Trial Clause limits the government’s ability to delay criminal trials once it has “arrested or formally accused” a defendant of a crime. Betterman v. Montana, 578 U.S. 437, 441 (2016). The purpose of the Clause is to “prevent[] undue and oppressive incarceration prior to trial, minimiz[e] anxiety and concern accompanying public accusation, and limit[] the possibilities that long delay will impair the ability of an accused to defend 20 UNITED STATES V. LONICH

himself.” Id. at 1614 (quoting United States v. Marion, 404 U.S. 307, 320–21 (1971)).

To assess whether the Speedy Trial Clause was violated, we apply the four-part balancing test from Barker v. Wingo, 407 U.S. 514 (1972), considering (1) the length of the delay, (2) the reason for the delay, (3) whether the defendant asserted his rights, and (4) the prejudice to the defendant. Id. at 530–33; see also Doggett v. United States, 505 U.S. 647, 651 (1992) (explaining that “[o]ur cases . . . have qualified the literal sweep of the [Speedy Trial Clause] provision by specifically recognizing the relevance of four separate enquiries” set forth in Barker); United States v. King, 483 F.3d 969, 976 (9th Cir. 2007).

Importantly, “none of the four [Barker] factors . . . [i]s either a necessary or sufficient condition to the finding of a deprivation of the right of speedy trial. Rather, they are related factors and must be considered together with such other circumstances as may be relevant,” as part of “a difficult and sensitive balancing process.” Barker, 407 U.S. at 533; see also United States v. Mendoza, 530 F.3d 758, 762 (9th Cir. 2008) (“None of [the Barker] factors are either necessary or sufficient, individually, to support a finding that a defendant’s speed[y] trial right has been violated.”); Gregory, 322 F.3d at 1161–65 (discussing the Barker factors).

The first Barker factor, the length of delay, is “a double enquiry,” serving both as a triggering mechanism for the rest of the Speedy Trial Clause evaluation and a factor in that analysis. Doggett, 505 U.S. at 651–52. The “general consensus” is that an eight-month delay “constitutes the threshold minimum” to initiate the full Barker inquiry. Gregory, 322 F.3d at 1162 n.3. If the delay crosses that threshold, we generally proceed to the four-factor Barker UNITED STATES V. LONICH 21

test. Id. at 1161. “Although there is no bright-line rule, courts generally have found that delays approaching one year are presumptively prejudicial.” Id. at 1161–62.

The parties spend considerable effort dueling over the first Barker factor. They agree that the relevant end date for our analysis is the original trial date in March 2017. But they disagree on when defendants’ Speedy Trial Clause rights attached for the new charges first brought in the superseding indictment. Defendants argue we should use the original indictment as the start date for the new charges in the superseding indictment. The government contends we should use the date it filed the superseding indictment because, in its view, the Speedy Trial Clause is “offense specific” and the new charges involved different offenses under Blockburger v. United States, 284 U.S. 299 (1932), the seminal precedent in the Double Jeopardy context. Essentially, the government argues that its superseding indictment reset the Speedy Trial Clause clock here because the new charges were not barred under Blockburger’s Double Jeopardy test.

We need not resolve that debate today because we conclude that, even assuming the clock started at the time of the original indictment, there was no Speedy Trial Clause violation because the delay caused no relevant prejudice to the defendants.

If we assume that the Speedy Trial Clause clock on the charges in the superseding indictment started to run when the initial indictment was filed, the delay from that point to the original trial date was three years. That is a substantial delay. But under Barker, it is not conclusively a Speedy Trial Clause violation. It is not nearly as egregious as other cases 22 UNITED STATES V. LONICH

in which courts have found Speedy Trial Clause violations. See, e.g., Doggett, 505 U.S. at 657 (8.5-year delay); United States v. Black, 918 F.3d 243, 248–49 (2d Cir. 2019) (5.75- year delay); United States v. Handa, 892 F.3d 95, 107 (1st Cir. 2018) (6.5-year delay); United States v. Shell, 974 F.2d 1035, 1036 (9th Cir. 1992) (5-year delay).

Indeed, when considering the other Barker factors, courts have held much longer delays than the one here permissible under the Speedy Trial Clause. See, e.g., United States v. Loud Hawk, 474 U.S. 302, 314–17 (1986) (more than 7-year delay); Barker, 407 U.S. at 533–34 (“well over five year[]” delay); United States v. Alexander, 817 F.3d 1178, 1183 (9th Cir. 2016) (per curiam) (5-year delay); Corona-Verbera, 509 F.3d at 1116 (“nearly eight-year delay”); Aguirre, 994 F.2d at 1456–58 (5-year delay); Rayborn v. Scully, 858 F.2d 84, 89 (2d Cir. 1988) (over 7- year delay). The three-year delay that we assume occurred here was thus not dispositively unconstitutional, but instead “presumptively prejudicial.” Gregory, 322 F.3d at 1162. This means it is “sufficient to trigger inquiry into the other three [Barker] factors.” Corona-Verbera, 509 F.3d at 1114.

The defendants argue that, combined with the government’s intentional decision to delay the superseding indictment, the 3-year delay requires the government to show that defendants were not prejudiced by the delay. We assume defendants are correct on that point. See Shell, 974 F.2d at 1036. But as we have explained, “presumptive prejudice is simply ‘part of the mix of relevant facts, and its importance increases with the length of the delay,’” Gregory, 322 F.3d at 1162 (quoting United States v. Beamon, 992 F.2d 1009, 1013 (9th Cir. 1993)), as well as the reason for the delay, see Alexander, 817 F.3d at 1182. Here the delay, while notable, was not on the high end of the range UNITED STATES V. LONICH 23

where Speedy Trial Clause violations typically lie. And most critically, regardless of who bears the burden to show prejudice (or lack thereof), there is simply no basis to find any Speedy Trial Clause prejudice on the facts of this case. See Doggett, 505 U.S. at 658 n.4 (noting that the government may “affirmatively prove[] that the delay left [the defendant’s] ability to defend himself unimpaired”).

The critical feature of this case is that the trial on the charges in the original March 2014 indictment was already set for March 2017. Defendants had no Speedy Trial Clause (or Speedy Trial Act) objection to trying the charges in the original indictment on that date. Indeed, they agreed to the March 2017 trial date based on their need to prepare for trial in a complex case, the amount of anticipated discovery, and defendants’ counsel’s schedules.

Thus, even if the government had brought all the charges in the original indictment, there is no reason to believe the trial date would have been set any earlier than March 2017. So, the government’s filing of the superseding indictment at most added seven months of delay, from March 2017 to the eventual trial date in October 2017—a date to which all parties agreed after the two cases were joined for trial at the defendants’ request.

The government has sufficiently demonstrated that the extra seven-month delay did not cause defendants any identifiable prejudice. Defendants were not incarcerated pending trial. See Betterman, 578 U.S. at 442. And while defendants presumably experienced some anxiety during the entire pretrial period, there is no suggestion that the added seven-month delay resulting from the new charges in the superseding indictment created any material increase in anxiety. See id. 24 UNITED STATES V. LONICH

Nor did the incremental delay “impair the ability” of defendants to defend themselves. Id. Defendants did have to defend against a broader set of charges involving a broader body of evidence. But the “prejudice with which we are concerned is prejudice caused by the delay that triggered the Barker inquiry, not simply any prejudice that may have occurred before the trial date but unrelated to the fact of the delay itself.” Gregory, 322 F.3d at 1163.

The prejudice associated with defending against the broader charges here is unrelated to the fact of the delay itself. Gregory is relevant on that point. There, we held that there was no Speedy Trial Clause violation when the government filed a third superseding indictment after the defendant had already pleaded guilty and served a prison sentence on earlier related charges. Id. at 1159–65. As we explained, even if the sequencing of the government’s charging decisions was “unusual,” the government’s filing of additional charges “is not in and of itself a constitutional violation.” Id. at 1161.

That was because, as here, “[t]he government was free to file a new indictment, rather than a superseding indictment,” which “would have presented no constitutional speedy trial problems.” Id. Indeed, that was so in Gregory even though there (unlike here) the new charges in the superseding indictment arose “out of the same course of conduct” as that charged in the original indictments, and there (unlike here) the defendant had even served time in prison on the earlier charges. Id.

As in Gregory, any prejudice to the defendants here “results solely from the government’s choice not to bring” the different charges together initially, which “has nothing to do with the delay itself from the time of the indictment until the time of trial.” Id. at 1164. If the government had UNITED STATES V. LONICH 25

merely filed a second action in October 2016 instead of a superseding indictment, defendants would have no basis to complain under the Speedy Trial Clause. It should not matter that the government initially joined the charges together in one case through a superseding indictment, broke them out to comply with the district court’s Rule 48 order, and then rejoined them for trial upon the defendants’ request.

In response, defendants lean heavily on the district court’s prejudice analysis. But that analysis shows why the district court was correct in concluding that dismissal with prejudice was not warranted. The district court found that the superseding indictment caused prejudice because it would require defendants to prepare a defense against a whole new set of allegations in time for a March 2017 trial. That was indeed prejudicial, justifying the district court’s decision to order the government to file the new charges in a new case. See Fed. R. Crim. P. 48(b). But that is not the type of prejudice the Speedy Trial Clause protects against: the Speedy Trial Clause relates to trial delay, not being rushed into a trial. Cf. Loud Hawk, 474 U.S. at 311 (“The Speedy Trial Clause does not purport to protect a defendant from all effects flowing from a delay before trial.”); United States v. MacDonald, 435 U.S. 850, 861 (1978) (“It is the delay before trial, not the trial itself, that offends against the constitutional guarantee of a speedy trial.”).

When defendants focus on areas of prejudice that could implicate the Speedy Trial Clause, the weakness of their prejudice theory becomes further apparent. Defendants note that two SVB directors suffered from memory issues and that a box of FDIC records from the May 2008 SVB examination had gone missing. But these assertions of prejudice are speculative, especially in view of the overall scope of the government’s prosecution and the 26 UNITED STATES V. LONICH

overwhelming evidence that defendants participated in fraudulent schemes. See Loud Hawk, 474 U.S. at 315 (“[T]he possibility of ‘impairment of a fair trial that may well result from the absence or loss of memory of witnesses’ . . . is not sufficient to support respondents’ position that their speedy trial rights were violated.” (quoting United States v. Loud Hawk, 741 F.2d 1184, 1193 (9th Cir. 1984)).

Even if defendants had known to contact the two SVB directors earlier, it is unclear if the witnesses’ memories would have been better then, since the events in question were already somewhat dated. The same is true of other witnesses who could not recall certain details at trial. As to the box of documents, it is unclear when it went missing. Plus, defendants had already extensively cross-examined the government’s investigator about the 2008 FDIC examination. In short, there is no “non-speculative proof as to how [defendants’] defense was prejudiced by the” additional seven-month delay. Alexander, 817 F.3d at 1183.

Defendants also point to the district court’s findings that the government “deliberately and intentionally” delayed pursuing the new charges in the superseding indictment. This argument resonates most centrally in Barker’s second factor—the reason for the delay—which we have described as “the focal inquiry” of the Barker analysis. Alexander, 817 F.3d at 1182 (quoting United States v. Sears, Roebuck & Co., Inc., 877 F.2d 734, 739 (9th Cir. 1989)).

We agree with defendants that the government’s reasons for delay augment the presumed prejudice that we must infer. See Doggett, 505 U.S. at 656–58; Shell, 974 F.2d at 1036. But defendants overstate the significance of this Barker factor on the particular facts of this case. The UNITED STATES V. LONICH 27

government did not act with reasonable diligence here, but the district court also emphasized that it did not find any bad faith. While “prejudice may be presumed” when the government “intentionally delayed or negligently pursued the proceedings,” Alexander, 817 F.3d at 1182, prejudice does not somehow drop out of the analysis altogether simply because the government may have acted strategically. Instead, “the amount of prejudice” required to trigger a Speedy Trial Clause violation “is inversely proportional to the length and reason for the delay.” Id. at 1183. And ultimately, and even if prejudice is presumed, the Speedy Trial Clause and Barker inquiry seek to ensure that defendants are not unduly harmed by excessive post- indictment delay. See MacDonald, 456 U.S. at 8 (“The speedy trial guarantee is designed to minimize the possibility of lengthy incarceration prior to trial, to reduce the lesser, but nevertheless substantial, impairment of liberty imposed on an accused while released on bail, and to shorten the disruption of life caused by arrest and the presence of unresolved criminal charges.”).

Here, the problem for defendants remains the same: the trial was already going to happen at least three years after the original indictment. In this case, there is no basis to conclude that the added delay resulting from the superseding indictment worked any relevant prejudice to the defendants. That is especially so when the government could have filed these new charges in an entirely separate case, see Gregory, 322 F.3d at 1161—as the government in fact did in response to the district court’s Rule 48 dismissal. We may not abstract one Barker factor from the others without considering the 28 UNITED STATES V. LONICH

overall facts and circumstances associated with the delay. See Barker, 407 U.S. at 533. 3

We do not minimize the district court’s concerns with the timing of the government’s charging decisions. The government’s answering brief admits that it “takes the district court’s criticism seriously and is committed to avoid[ing] such situations in the future.” We fully expect the government to abide by that representation. And, the result might well be different if meaningful Speedy Trial prejudice resulted from the government’s actions.

But the issue here is whether the defendants should effectively reap a windfall when the timing of the superseding indictment worked no meaningful Speedy Trial Clause prejudice. Neither the Sixth Amendment nor Barker support, much less require, that result. Given the effective seven-month delay and the lack of material prejudice, we hold that the Speedy Trial Clause was not violated.

III. Jury Instructions for Money Laundering and Misapplication of Bank Funds

Defendants next challenge the jury instructions on the money laundering and misapplication of bank funds charges. Specifically, defendants contend that the instructions’ overarching definition of “knowingly” conflicted with the required mental states for the two charged offenses. We hold, however, that the district court’s general “knowingly”

3 The Third Circuit’s decision in United States v. Battis, 589 F.3d 673 (3d Cir. 2009), on which defendants rely, is distinguishable. That case involved a 45-month delay, at least 35 months of which were attributable to the government. Id. at 683. Battis also did not involve a defendant who had already agreed to a substantial delay in the trial, as here. UNITED STATES V. LONICH 29

instruction was permissible and that defendants in any event have not shown prejudice from the instruction.

Defendants were charged with twelve counts of money laundering under 18 U.S.C. § 1957. These counts were based on the 101 Houseco scheme for defendants transferring the loan proceeds from House’s construction company and 101 Houseco to Masma Construction, a Madjlessi construction company. Cutting and Melland were also charged with one count of misapplication of bank funds under 18 U.S.C. § 656 for arranging the fraudulent loans.

The district court proposed count-specific instructions for the money laundering and misapplication of bank funds charges that followed the Ninth Circuit’s model jury instructions. See Manual of Model Criminal Jury Instructions for the Ninth Circuit § 8.41 (misapplication of bank funds); id. § 8.150 (money laundering). The district court also proposed defining “knowingly” for all charges in a manner that tracked the model instruction. See id. § 5.7.

The defendants objected to part of this proposed “knowingly” instruction, arguing that a phrase in the general instruction—“[t]he government is not required to prove that a defendant knew that his or her acts or omissions were unlawful”—conflicted with the mens rea requirements in the instruction specific to misapplication of bank funds. To address that concern, the district court modified the general “knowingly” instruction to clarify, with the language underlined below, that the challenged sentence only applied to commission of the act itself:

An act is done knowingly if the defendant is aware of the act and does not act through 30 UNITED STATES V. LONICH

ignorance, mistake or accident. To prove that an act is done knowingly, the government is not required to prove that the defendant knew that his or her acts were unlawful.

The government agreed with the revised instruction and defendants did not object further. The district court used this revised definition for “knowingly” to instruct the jury. 4

“We review the formulation of jury instructions for abuse of discretion, but review de novo whether those instructions correctly state the elements of the offense and adequately cover the defendant’s theory of the case.” United States v. Liew, 856 F.3d 585, 595–96 (9th Cir. 2017). “We must determine whether the instructions, viewed as a whole, ‘were misleading or inadequate to guide the jury’s deliberation.’” United States v. Kaplan, 836 F.3d 1199, 1215 (9th Cir. 2016) (quoting United States v. Moore, 109 F.3d 1456, 1465 (9th Cir. 1997) (en banc)). Absent contemporaneous objection, we review for plain error. United States v. Soto, 519 F.3d 927, 930 (9th Cir. 2008).

“Jury instructions, even if imperfect, are not a basis for overturning a conviction absent a showing that they prejudiced the defendant.” Kaplan, 836 F.3d at 1215 (quoting United States v. Christensen, 828 F.3d 763, 786 (9th Cir. 2015)). If an instruction is erroneous, “[w]e apply harmless error analysis to determine whether an improper

4 In the final instructions, the underlined language was revised to replace the word “done” with “committed.” UNITED STATES V. LONICH 31

instruction constitutes reversible error.” United States v. Munguia, 704 F.3d 596, 598 (9th Cir. 2012).

Although defendants did not object anew after the district court revised its general “knowingly” instruction, the government concedes that defendants adequately preserved their objection as to the misapplication of bank funds instructions. Thus, we review this claim for abuse of discretion. Liew, 856 F.3d at 595–96. Although it is less clear that the defendants preserved their arguments for the money laundering instructions, we assume for purposes of this appeal that they did. But even under abuse of discretion review, defendants’ claims fail.

We discern no error in the general “knowingly” instruction as applied to the specific money laundering instructions. Defendants do not contest that the district court correctly stated the mens rea for money laundering, see 18 U.S.C. § 1957, when it instructed the jury that as to that offense, the government must prove (as relevant here):

First, the defendants knowingly engaged or attempted to engage in a monetary transaction; [and]

Second, the defendants knew the transaction involved criminally derived property.

Defendants contend, however, that the second sentence of the overarching “knowingly” instruction—“[t]o prove that an act is committed knowingly, the government is not required to prove that a defendant knew that his or her acts or omissions were unlawful”—contravened the requirement that, for money laundering, the government must prove that 32 UNITED STATES V. LONICH

defendants “knew the transactions involved criminally derived property.” United States v. Turman, 122 F.3d 1167, 1169 (9th Cir. 1997), abrogated on other grounds by Henderson v. United States, 568 U.S. 266 (2013).

We disagree. The general “knowingly” definition, as the district court revised it, only applied to whether “an act is committed knowingly.” By the terms of that instruction, “[a]n act is done knowingly if the defendant is aware of the act and does not act through ignorance, mistake or accident.” That provides the set-up for the key language as the district court revised it: “To prove that an act is committed knowingly, the government is not required to prove that the defendant knew that his or her acts were unlawful.”

This general definition of “knowingly” did not extend to the money laundering charge’s second element, which required the jury to find that defendants “knew the transaction involved criminally derived property.” Rather, the general “knowingly” instruction only applied to the money laundering charge’s first element, whether defendants “knowingly engaged or attempted to engage” in a monetary transaction. In other words, these instructions “are substantively different because they address two distinct types of subjective knowledge.” United States v. Greer, 640 F.3d 1011, 1019–20 (9th Cir. 2011).

The district court also elsewhere reinforced that the government must prove that defendants knew their underlying conduct was illegal to convict them for money laundering. After reciting the elements for money laundering, the court instructed the jury that the “government must prove that the defendants knew that the property involved in the monetary transaction constituted, or was derived from, proceeds obtained by some criminal offense.” This dispelled “the possibility that the jury could UNITED STATES V. LONICH 33

have omitted the second element of money laundering and convicted without finding [defendants] knew the money represented illegal . . . proceeds.” United States v. Knapp, 120 F.3d 928, 932 (9th Cir. 1997).

This case thus differs from United States v. Stein, 37 F.3d 1407 (9th Cir. 1994), on which defendants principally rely. In Stein, we reversed the defendant’s money laundering convictions because the general jury instructions defined “knowingly” in a way that conflicted with the jury instruction specific to money laundering, which required that the government prove defendant knew the money represented illegal proceeds. Id. at 1410. There, the district court’s “knowingly” instruction broadly specified that “[t]he Government is not required to prove that the defendant knew that his acts or omissions were unlawful.” Id. We thus held that the instruction was impermissible because the general definition for “knowingly” “conflict[ed] with the district court’s previous specific instruction on money laundering,” meaning that “a jury could convict Stein without finding that he knew his predicate acts of fraud were unlawful.” Id.

The problematic unqualified instruction in Stein is different than the more tailored “knowingly” instruction that the district court gave here. Unlike this case, the district court in Stein did not limit the general “knowingly” definition to whether an act was committed knowingly. See id..

This case more closely resembles Knapp, 120 F.3d at 931, in which we found no conflict between a general “knowingly” definition and the money laundering instructions. In Knapp, the district court provided a general “knowingly” definition that: 34 UNITED STATES V. LONICH

An act is done knowingly if the defendant is aware of the act and does not act or fail to act through ignorance, mistake or accident. As to money laundering, the government is not required to prove that the defendant knew that his acts or omissions were unlawful.

Id. at 931. We held that this instruction was “distinguish[able] . . . from the one given in Stein.” Id. at 932. It created no impermissible dissonance with the money laundering mens rea requirements because the “added phrase ‘as to money laundering’ made sufficiently clear that the corresponding sentence applied only to the crime itself.” Id.

If anything, Knapp presented a closer case than this one. Here, the jury instruction explicitly stated that knowledge of unlawfulness is unnecessary only when the question is whether “an act is committed knowingly.” By focusing specifically on the “act,” the added language was sufficient to avoid interference with the specific mens rea requirements for money laundering. See United States v. Golb, 69 F.3d 1417, 1428 (9th Cir. 1995) (holding that “general knowledge instructions” did not “negate[] the scienter element of the money-laundering offense”).

Our holding here is also consistent with cases in which we have found no conflict between two mens rea instructions that, like here, are “thematically similar” but “substantively different.” Greer, 640 F.3d at 1019–20. For example, we have held that an instruction that the government must prove “the defendant knew that the firearm [he possessed] was a machine gun,” did not conflict with a general knowledge instruction stating that “[t]he government is not required to prove that the defendant knew that his acts . . . were UNITED STATES V. LONICH 35

unlawful.” United States v. Gravenmeir, 121 F.3d 526, 529– 30 (9th Cir. 1997). We reasoned that “a general instruction that requires a defendant’s awareness of his acts” did not negate the requirement that the defendant “knew that the firearm was a machine gun” because they dealt with two different requirements. Id. at 530 (emphasis added); see also Greer, 640 F.3d at 1020 (no conflict when general “knowingly” instruction dealt with “knowledge of unlawful activity,” while specific instruction dealt with “knowledge of legal entitlement to the property the alleged extortionist tried to obtain”).

Defendants point to a comment in the Manual of Model Criminal Jury Instructions that, in discussing a different money laundering provision, noted that “it is reversible error to give [the model ‘knowingly’ instruction] in a money laundering case” because “it is a specific intent crime.” Ninth Circuit Manual of Model Criminal Jury Instructions § 8.146. But the district court did not give the model “knowingly” instruction. Rather, it modified the definition to limit it only to “acts” committed knowingly, negating the concerns the Model Jury Instructions comment addressed.

For substantially the same reasons, we reject defendants’ contention that the general “knowingly” instruction conflicts with the willfulness requirement for the misapplication of bank funds offense.

The district court correctly instructed the jury that under the misapplication of bank funds statute, 18 U.S.C. § 656, the government had to prove that “defendants knowingly and willfully stole, embezzled, or misapplied funds or credits belonging to the bank or entrusted to its care in excess of $1,000.” The district court also properly instructed the jury 36 UNITED STATES V. LONICH

that “‘[w]illfully’ as used in this instruction means undertaking an act with a bad purpose,” so that “in order to establish a willful violation of a statute, the government must prove that the defendant acted with knowledge that his conduct was unlawful.” Once again, nothing in the general “knowingly” instruction—relating to whether “an act is committed knowingly”—undermined the specific mens rea requirements applicable to misapplication of bank funds. See, e.g., Knapp, 120 F.3d at 931–32.

Even assuming the jury instructions were incorrect, the error did not prejudice defendants. There was overwhelming evidence that defendants had the required mens rea for both sets of offenses: defendants clearly knew that the 101 Houseco disbursements were based on the proceeds of a fraudulent loan to 101 Houseco, LLC.

We can see this most directly in defendants’ convictions for wire fraud and conspiracy to commit wire fraud (Counts 10 to 15). Those convictions were based on fraudulent communications surrounding the 101 Houseco scheme. To find defendants guilty of these wire fraud charges, the jury had to conclude that defendants knowingly misrepresented House as 101 Houseco’s legitimate owner when they knew that Madjlessi was the real buyer. Put differently, the jury must have found that defendants knew House was a straw purchaser.

Defendants do not contest this conclusion. They instead note that knowing House was a straw buyer did not mean that the defendants also “knew the proceeds of the HouseCo loan were unlawful.” This argument is not tenable. It is “clear beyond a reasonable doubt that a rational jury” would find that defendants—members of the legal and banking UNITED STATES V. LONICH 37

professions—knew it was illegal knowingly to help a straw buyer secure a $10 million loan. See Munguia, 704 F.3d at 603–04.

IV. Sufficiency of the Evidence on Bribery and Obstruction of Justice Charges

Melland argues that there was insufficient evidence to support his conviction for bribery by a bank employee (Count 9). See 18 U.S.C. § 215(a)(2). Lonich contends that there was insufficient evidence to sustain his conviction for attempted obstruction of justice (Count 37). See 18 U.S.C. § 1512(c)(2). We must “determine whether ‘after viewing the evidence in the light most favorable to the prosecution, any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt.’” United States v. Nevils, 598 F.3d 1158, 1163–64 (9th Cir. 2010) (en banc) (quoting Jackson v. Virginia, 443 U.S. 307, 319 (1979)). We hold that sufficient evidence supported the convictions for bribery and attempted obstruction of justice.

Melland’s bribery charge was based on his securing from Madjlessi a $50,000 investment in Magnus Innovations, Melland’s energy-drink start-up. Madjlessi wired Magnus Innovations $50,000 using an overdraft check from Madjlessi’s wife, drawn from a Madjlessi entity’s business account. The very next day, Melland sought approval from SVB’s loan committee for the third pair of $1.86 million Greenbriar loans (loans for another Madjlessi development). In doing so, Melland misrepresented the borrower’s liability and hid Madjlessi’s interest in the loans. After the jury returned a guilty verdict on this charge, Melland moved for a judgment of acquittal, arguing that the evidence against 38 UNITED STATES V. LONICH

him was insufficient. The district court did not err in denying that motion.

Under 18 U.S.C. § 215(a)(2), whoever “as an officer, director, employee, agent, or attorney of a financial institution, corruptly solicits or demands for the benefit of any person, or corruptly accepts or agrees to accept, anything of value from any person, intending to be influenced or rewarded in connection with any business or transaction of such institution,” is subject to punishment. This offense— specific to banking officials, employees, and agents—is not one we have had much previous occasion to consider. But as we remarked in the context of § 215’s predecessor provision, “[t]here can be no doubt that Congress intended, by the enactment of this statute, to remove from the path of bank officials the temptation to enrich themselves at the expense of the borrowers or the bank, and also to prevent improvident loans.” Ryan v. United States, 278 F.2d 836, 838 (9th Cir. 1960); see also United States v. Brunson, 882 F.2d 151, 155 (5th Cir. 1989) (identifying the elements of a § 215(a)(2) violation).

Melland’s challenge to his conviction centers on only one aspect of the government’s required § 215(a)(2) showing: the requirement that Melland acted “corruptly,” i.e., with corrupt intent. The mens rea “corruptly” has a lengthy historical lineage. While the Model Penal Code seems to resist it, see Model Penal Code §§ 240.1–7 Explanatory Note, the term “corruptly” still appears with some frequency in the federal criminal code. See Eric Tamashasky, The Lewis Carroll Offense: The Ever- Changing Meaning of ‘Corruptly’ Within the Federal Criminal Law, 31 J. Legis. 129, 130–33 (2004). We encounter it most commonly in criminal statutes involving bribery, obstruction of justice, and tampering with UNITED STATES V. LONICH 39

witnesses. See, e.g., 18 U.S.C. § 201 (bribery of public officials and witnesses); id. § 226 (bribery affecting port security); id. § 666 (theft or bribery concerning programs receiving federal funds); id. § 1505 (obstruction of proceedings before departments, agencies, and committees); id. §§ 1512(b)–(c) (obstructing official proceedings); id. § 1517 (obstructing examination of a financial institution).

The district court instructed the jury that, to find Melland “acted corruptly,” it must determine he “intend[ed] to be influenced or rewarded in connection with any business or transaction of” a financial institution. While we have not specifically defined “corruptly” in the context of § 215(a)(2), we conclude that the district court’s instruction appropriately stated the law, as the parties effectively agree.

“As used in criminal-law” statutes, the term “corruptly” usually “indicates a wrongful desire for pecuniary gain or other advantage.” Black’s Law Dictionary 435 (11th ed. 2019). In United States v. Rodrigues, 159 F.3d 439 (9th Cir. 1998), we commented—in the case of § 215(a)(2) in particular—that “‘[t]o corrupt’ is a standard synonym for ‘to bribe,’” so that “[t]o accept corruptly something of value as a reward is to accept a payoff or bribe.” Id. at 450.

That is essentially how we have interpreted “corruptly” in the context of similar federal criminal statutes. See, e.g., Martinez-de Ryan v. Whitaker, 909 F.3d 247, 250 (9th Cir. 2018) (explaining that for purposes of 18 U.S.C. § 666, which prohibits bribery in connection with programs receiving federal funds, “‘[a]n act is done ‘corruptly’ if it is performed voluntarily, deliberately, and dishonestly, for the purpose of either accomplishing an unlawful end or result or of accomplishing some otherwise lawful end or lawful result by an unlawful method or means’”) (quoting United States v. McNair, 605 F.3d 1152, 1193 (11th Cir. 2010)); United 40 UNITED STATES V. LONICH

States v. Leyva, 282 F.3d 623, 626 (9th Cir. 2002) (explaining for purposes of 18 U.S.C. § 201, which prohibits bribery of public officials, that “corruptly” “refers to the defendant’s intent to be influenced to perform an act in return for financial gain”). The district court’s instruction thus appropriately captured the plain meaning of “corruptly” that we have set forth in our prior cases.

We now turn to whether the evidence was sufficient to show that Melland acted “corruptly” in procuring Madjlessi’s $50,000 investment in Melland’s energy-drink enterprise. The government did not need a “smoking gun” document or admission to show that Melland acted “corruptly” through an apparent quid pro quo arrangement with Madjlessi. Instead, it could prove Melland’s criminal intent through circumstantial evidence. See, e.g., United States v. Kincaid-Chauncey, 556 F.3d 923, 937 (9th Cir. 2009), abrogated in part on other grounds, Skilling v. United States, 561 U.S. 358 (2010).

The circumstantial evidence here was plentiful. Earlier, and while Melland was pushing for SVB to approve the first two pairs of Greenbriar loans that exceeded the legal lending limit, Madjlessi hosted Melland and his business partner on his yacht to discuss Magnus Innovations. Later, on March 25, 2008, Melland emailed Madjlessi with an “URGENT!!!” solicitation to invest in that business. The very next day after Madjlessi wired Melland a $50,000 investment in Magnus Innovations, Melland presented the third pair of Greenbriar loans (that exceeded the legal lending limit) to SVB’s loan committee. The timing of events strongly suggests that Melland sought and accepted $50,000 from Madjlessi for Melland’s energy drink side business in return for helping Madjlessi secure a large loan from SVB. UNITED STATES V. LONICH 41

The government also showed that Melland and Madjlessi each needed what the other could provide. Melland needed seed money for his drink company. Melland knew that Madjlessi needed the SVB loans because of Madjlessi’s own cash-flow issues. Melland also knew that Madjlessi could pay for the Magnus Innovations investment using an overdraft check because Melland could approve the overdraft, which he evidently did.

Melland nonetheless contends it would be unreasonable to infer any corrupt intent on his part because SVB had previously approved several loans for Madjlessi. But a properly instructed jury could certainly conclude otherwise. Indeed, that SVB approved previous loans for Madjlessi above the legal lending limit meant, if anything, that Madjlessi had an even greater reliance on Melland. Melland, as noted, had helped lock down the first two pairs of Greenbriar loans while he negotiated with Madjlessi for an investment in his start-up. Melland also knew that Madjlessi needed him to recommend the third pair of loans to SVB’s loan committee, even though it far exceeded SVB’s legal lending limits for Madjlessi.

Melland responds that it was merely coincidental that he took $50,000 from an investor whose loan requests Melland was actively pressing before his bank. But in a sufficiency of the evidence challenge it is immaterial that “there is an equally plausible innocent explanation.” Nevils, 598 F.3d at 1169. We thus hold that sufficient evidence supported Melland’s conviction for bribery by a bank employee. 42 UNITED STATES V. LONICH

We turn next to Lonich’s claim that insufficient evidence supported his conviction for attempted obstruction of justice by encouraging House to mislead the grand jury about his role in the 101 Houseco scheme. See 18 U.S.C. § 1512(c)(2). Lonich’s conviction on this charge arose out of two meetings involving Lonich, House, and Madjlessi. Working with the government, House secretly recorded these meetings. During trial, the government played the video recordings for the jury.

At the first meeting, House handed Lonich and Madjlessi a subpoena House had received commanding him “to appear and testify before the Grand Jury” in the Northern District of California. The subpoena also required House to produce documents about a condominium he bought through Madjlessi in Reno, Nevada.

House told Lonich and Madjlessi that the subpoena had made him “a fucking wreck.” Lonich and Madjlessi then tried to downplay the subpoena’s significance. Lonich told House that because the subpoena only referenced the Reno condominium, House would not need to discuss 101 Houseco. But House responded that the agents “want to talk about 101 Houseco to me.” This pattern repeated throughout the conversation: Lonich returned to the subpoena’s Reno-related document request, while House stressed that the agents wanted to talk about 101 Houseco. House also informed Lonich that when he told the agents he could not discuss 101 Houseco, they told him to “go tell it to the Grand Jury.” UNITED STATES V. LONICH 43

The next day, the three resumed their discussion. This time, Lonich provided guidance to House on what he should tell the grand jury. Among other things, Lonich instructed House: “I think it’s good for you to be able to say, I had a guy running [101 Houseco]. I’m running my business. And this was an investment that I made.” Later in the meeting House asked: “What about a straw buyer for [101 Houseco]? . . . I received money that was owed me.” Lonich again instructed: “I think you are better off to say you received money, not money that was owed you. You made an investment. You know—you know, I don’t care.” Lonich reiterated: “My sense is I think it’s much better that you got paid. . . . Not that it was an old debt, but that you got paid.”

18 U.S.C. § 1512(c)(2) punishes anyone who “corruptly . . . obstructs, influences, or impedes any official proceeding, or attempts to do so.” Section 1512(f)(1) clarifies that “an official proceeding need not be pending or about to be instituted at the time of the offense.” An “official proceeding” includes “a proceeding before . . . a Federal grand jury.” 18 U.S.C. § 1515(a)(1)(A).

Lonich first contends that the government did not prove a nexus between his actions and the grand jury proceeding. We agree that § 1512(c)(2) has a nexus requirement. Namely, § 1512(c)(2) “require[s] that (1) the obstructive conduct be connected to a specific official proceeding . . . that was (2) either pending or was reasonably foreseeable to [the defendant] when he engaged in the conduct . . . .” United States v. Young, 916 F.3d 368, 385 (4th Cir. 2019).

That nexus requirement is firmly rooted in law. The Supreme Court has identified a nexus requirement in two related obstruction provisions that employ similar statutory 44 UNITED STATES V. LONICH

language. See Arthur Andersen LLP v. United States, 544 U.S. 696, 707–08 (2005); United States v. Aguilar, 515 U.S. 593, 598–99 (1995). In Aguilar, the Court considered the catchall provision in 18 U.S.C. § 1503, a grand jury tampering statute. See 515 U.S. at 599. Section 1503 provides that a person may not “corruptly or by threats or force . . . influence[], obstruct[], or impede[] . . . the due administration of justice.” The only evidence the government had shown to support the conviction in Aguilar was that the defendant made false statements to an investigating agent “who might or might not testify before a grand jury.” 544 U.S. at 600. In reversing the conviction, the Court held that § 1503 required a nexus showing, namely, “that the act [had] a relationship in time, causation, or logic with the judicial proceedings.” Id. at 599.

In Arthur Andersen, the Supreme Court applied Aguilar to a related provision of the obstruction statute. See 544 U.S. at 707–08. Arthur Andersen held that 18 U.S.C. § 1512(b)(2)(A), which prohibits tampering with documents that would be used in an official proceeding, requires proof of a “nexus between the ‘persuasion’ to destroy documents” and the official proceeding. Id. at 707 (alterations omitted). The Court thus held that a jury cannot convict a defendant under § 1512(b)(2)(A) “when he does not have in contemplation any particular official proceeding in which those documents might be material.” Id. at 708.

Every other circuit to have considered the issue has found that § 1512(c) requires a showing that the obstruction was connected to a pending or reasonably foreseeable official proceeding. See Young, 916 F.3d at 386 (citing, e.g., United States v. Martinez, 862 F.3d 223, 237 (2d Cir. 2017); United States v. Petruk, 781 F.3d 438, 445 (8th Cir. 2015); United States v. Tyler, 732 F.3d 241, 249–50 (3d Cir. 2013); UNITED STATES V. LONICH 45

United States v. Friske, 640 F.3d 1288, 1292 (11th Cir. 2011); United States v. Phillips, 583 F.3d 1261, 1263–64 (10th Cir. 2009)). And United States v. Ermoian, 752 F.3d 1165 (9th Cir. 2013), although not directly resolving the issue, we noted in construing § 1512(c)(2) that “Supreme Court precedent requir[es] a nexus between the obstructive act and criminal proceedings in court.” Id. at 1172 (citing Arthur Andersen, 544 U.S. at 708).

We thus hold that § 1512(c)(2) requires a showing of nexus. Nevertheless, we reject Lonich’s argument that no reasonable jury could have found the required nexus here. Lonich asserts he did not contemplate an official proceeding concerning 101 Houseco because he thought the grand jury was investigating a separate issue about the purchase of a condominium in Reno. Lonich’s comments at the first meeting with House and Madjlessi did focus on the Reno property. But a jury need not accept his version of what he allegedly believed.

In the meetings, House stressed to Lonich and Madjlessi that federal agents wanted to talk to him about 101 Houseco before the grand jury. And Lonich provided House with specific guidance about what House should say to the grand jury about the 101 Houseco arrangement. Given the explicit discussions at the meetings about 101 Houseco and what House should say on that topic, a reasonable jury could easily conclude that when Lonich sought to frame the subpoena as focused on a Reno property, he was seeking to calm House and minimize both the significance of the federal proceeding and the likelihood that House would offer testimony adverse to Lonich and Madjlessi. 46 UNITED STATES V. LONICH

Lonich also maintains that the government did not prove the statute’s mens rea requirement that a defendant “corruptly . . . obstructs, influences, or impedes any official proceeding, or attempts to do so.” We have not yet defined “corruptly” in § 1512(c). See United States v. Watters, 717 F.3d 733, 735 (9th Cir. 2013) (reserving the issue). We have, however, affirmed an instruction stating that “‘corruptly’ meant acting with ‘consciousness of wrongdoing’” because it, “if anything, . . . placed a higher burden of proof on the government than section 1512(c) demands.” Id. The district court provided an analogous instruction to the jury, and neither party disputes using a “consciousness of wrongdoing” mens rea requirement for purposes of evaluating the sufficiency of the evidence.

Under that articulation, a reasonable jury could find that the government met its burden of proof in demonstrating Lonich’s criminal intent. We focus on two exchanges on the second day of meetings. The first involved Lonich instructing House on “the story” he should tell the grand jury. Lonich said he thought it was “good for [House] to be able to say, [House] had a guy running [101 Houseco]” and that House should say he entered the 101 Houseco arrangement as “an investment.” Yet at trial, House repeatedly testified that he only agreed to take part in the 101 Houseco scheme to get the $200,000 Madjlessi owed him, and that House would only be the owner “on paper.”

In the second exchange, House asked Madjlessi and Lonich how he should testify about being a “straw buyer.” Madjlessi told House to say that he “made money,” but House countered: “I received money that was owed me.” To this Lonich said: “I think you are better off to say you received money, not money that was owed you. You made UNITED STATES V. LONICH 47

an investment.” Lonich soon after reiterated, “My sense is I think it’s much better that you got paid. . . . Not that it was an old debt, but that you got paid.”

A reasonable jury could draw from these exchanges that Lonich instructed House to deceive the grand jury. It is more than reasonable to think, based on all the evidence presented at trial, that Lonich knew House only took part in the scheme to get the $200,000 Madjlessi already owed him. House testified that Lonich knew about the arrangement. And that Madjlessi and Lonich only “permitted [House] to retain” approximately $200,000 reflected Madjlessi’s existing debt to House. A rational jury could conclude that House did not make an “investment” to buy a property he knew would net him only money already owed him. And that same jury could also think it was deceitful for Lonich to tell House to testify this was “not money that was owed to you.”

Lonich points to other aspects of the taped conversations potentially more favorable to him, including his repeated assertions to House that Lonich’s version of the facts was the “truth.” But a jury could focus on the parts of Lonich’s guidance that are more problematic. And Lonich representing his advice as “truthful” does not make it so.

Lonich also argues that he was Houseco’s lawyer and that, under 18 U.S.C. § 1515(c), the prohibition on obstructing an official proceeding “does not prohibit or punish the providing of lawful, bona fide, legal representation services in connection with or anticipation of an official proceeding.” This argument is unavailing. Lawyers of course have some latitude in helping clients frame their anticipated testimony in a light most favorable to them, consistent with the truth. The problem for Lonich is that even assuming he had an attorney-client relationship with House or a legal relationship with 101 Houseco (an 48 UNITED STATES V. LONICH

entity created to perpetuate a fraud), a rational jury could find that Lonich’s recorded conversations with House go far beyond “lawful, bona fide” legal advice. Lonich urged House to testify that he made an “investment” while avoiding saying that money was owed to House—advice that invited House to give grand jury testimony that was either outright false, seriously misleading, or both.

Lonich also errs in relying on United States v. Liew, 856 F.3d 585, 604 (9th Cir. 2017), in which we reversed an obstruction of justice conviction. In Liew, the only evidence that the defendant tried to influence testimony was that he told a potential witness not to discuss an issue with anyone “because doing so would not be good for” the witness or his family. Id. We found this insufficient to convict because it was “the same advice that many criminal attorneys would [give] in that situation[.] . . . Sometimes the best advice for a potential criminal defendant is not to talk to anyone about anything.” Id. But Lonich did not tell House to avoid discussing 101 Houseco. He instead counseled House on how to testify about 101 Houseco in ways that a jury could find reflected a “consciousness of wrongdoing.” Arthur Andersen, 544 U.S. at 706. 5

V. Sentencing and Restitution

The principal issue we consider concerning sentencing is the district court’s adoption of several enhancements that dramatically increased defendants’ recommended Guidelines sentencing ranges. These enhancements were 5 In a separate unpublished memorandum disposition, we reject defendants’ other challenges to their convictions. And because none of defendants’ arguments demonstrate error, the cumulative error doctrine does not apply. See United States v. Lindsey, 634 F.3d 541, 555 (9th Cir. 2011). UNITED STATES V. LONICH 49

premised on a critical factual finding: that defendants caused SVB to fail, making defendants responsible for associated losses. Most prominently, in calculating the financial loss that defendants caused, the Presentence Investigation Reports (PSRs) and the district court did not calculate loss simply based on the amounts of the defaulted Madjlessi- related loans, but instead based on the total amount of loss the federal government allegedly sustained because of SVB’s collapse.

We hold that defendants’ sentences must be vacated. The jury did not determine whether defendants caused SVB’s failure. And the government at sentencing did not sufficiently prove that point. It may be that, on remand, the government will be able to justify its requested enhancements. But on this record, it has not done so.

Not only did the jury through its verdict never decide whether defendants caused SVB’s failure, the district court prohibited the government from implying that causal relationship when presenting its case. A pretrial order provided that “[t]he government may not argue that defendants’ offense conduct caused the failure of SVB,” “agree[ing] with defendants that because the government need not prove causation for any of the charged crimes, the probative value of evidence and arguments about what caused the bank’s failure is substantially outweighed by the risk of undue prejudice and waste of time.” The government’s witnesses were thus not permitted to draw any causal connection between the Madjlessi loans and the bank’s later collapse because, the court explained, “there were other loans out there and other issues out there. This is the biggest recession since the Great Depression. It would be very complicated to prove what caused what.” 50 UNITED STATES V. LONICH

Nevertheless, after the jury returned its verdict, the government’s theory that defendants caused the bank’s failure resurfaced. The PSRs assigned each defendant a base offense level of 7. But the PSRs recommended much higher total offense levels for each defendant: 38 for Cutting and Melland, and 40 for Lonich.

What accounted for the dramatic increases was the PSRs’ recommended sentencing enhancements. The PSRs recommended a 20-level enhancement for each defendant under U.S.S.G. § 2B1.1(b)(1)(K), asserting they were responsible for a loss between $9.5 million and $25 million, which the PSRs estimated at upwards of $20 million. The PSRs also recommended a 2-level enhancement for each defendant because the offenses involved ten or more victims, see id. § 2B1.1(b)(2)(A)(i), and a 4-level enhancement for each defendant for substantially jeopardizing the safety and soundness of a financial institution, see id. § 2B1.1(b)(17)(B)(i).

The PSR did not base the 20-level loss enhancement on amounts owed SVB on the Madjlessi-related loans. Instead, the estimated losses were premised on the theory that defendants “caused the eventual failure and closure of SVB, which resulted in a loss totaling $20,120,000.” That figure represented amounts the federal government allegedly lost because SVB failed.

The PSRs rejected defendants’ arguments that “events independent of the offense conduct,” including a “cataclysmic financial crisis,” caused SVB’s demise. Instead, the PSRs attributed the bank’s downfall—and the government’s resulting losses—to defendants’ schemes. The critical paragraph in each of the PSRs stated: UNITED STATES V. LONICH 51

In total, loans provided to Bijan Madjlessi, directly to himself or through straw buyers, totaled approximately $35,000,000, which was approximately $24,700,000 over the bank’s lending limit in 2010. According to the FDIC report, on December 21, 2009, they were forced to downgrade SVB’s capital due to the relationship between the bank and Bijan Madjlessi. The report specifically notes that “Prior to the loss classification [i.e., the FDIC downgrade], the $27,195,000 Bijan Madjlessi relationship represented 74% of Total Risk Based capital. Several large borrower concentrations (Brian Madjlessi and [another borrower]) expose the institution to significant risk.” After this downgrade from the FDIC, the bank failed as it was unable to provide the necessary funding to stay afloat because 74% of their risks were taken up by Bijan Madjlessi.

The “FDIC report” referred to in the above paragraph in the PSRs downgraded SVB to the lowest possible rating and was issued before SVB failed. While the FDIC report noted that SVB had concentrated too much capital in Madjlessi- related loans, it also mentioned a host of other problems that SVB faced. These included liquidity issues, “deteriorating market conditions,” and concentrated capital with another large borrower.

Lonich also objected to the PSR’s recommended loss enhancement for reasons more specific to him. He argued that he had not caused any loss to SVB because he was only involved with the 101 Houseco loan, and that loan was repaid in full. Lonich also maintained that he “should not be 52 UNITED STATES V. LONICH

held accountable for any loans prior to his involvement in the conspiracy.”

The PSR disagreed, responding that “Lonich’s significant role within the conspiracy negatively impacted the safety and soundness of the bank.” The PSR concluded that Lonich was still responsible for the bank’s collapse because the bank failed after Lonich had joined the conspiracy, and Lonich “is responsible for all acts of the conspiracy.”

After determining that defendants were responsible for SVB’s failure, the PSRs calculated the total loss amount— approximately $20 million—based on amounts the federal government lost because of the bank’s failure. This loss fell into two categories: $8.65 million that SVB was unable to repay to the federal Troubled Asset Relief Program (TARP), and $11.47 million that the FDIC paid out of its Deposit Insurance Fund after SVB was forced to write off various Madjlessi-related loans. The PSRs adopted the same losses for restitution purposes as had “been determined by the Government,” but did not specify the calculation method.

Before the sentencing hearing, the government filed a sentencing memorandum. While maintaining that defendants caused SVB’s collapse, the memorandum did not rely on the FDIC report referenced in the PSRs. Instead, it attached law enforcement investigative reports summarizing interviews of three witnesses bolstering the theory that defendants were responsible for SVB’s failure.

The government also argued for a much larger loss amount than the PSRs originally recommended. Specifically, it argued that because of SVB’s failure, the FDIC lost $39.18 million—meaning that, combined with the TARP losses, defendants caused $47.84 million in losses to UNITED STATES V. LONICH 53

the federal government. Despite the overall larger loss estimation, the government claimed that the FDIC’s insurance fund experienced only $10.54 million in loss, a discrepancy from the PSRs’ calculation of $11.47 million.

At the sentencing hearing, the district court agreed with the sentencing recommendations in the PSRs, rejecting the government’s request to increase the total loss amount to $47.84 million. (Several months later in its restitution order, the court explained why it had not adopted the government’s larger number. The court recounted how the government had not objected to the PSRs’ loss calculations, and how the government had provided “no explanation” about why it did not provide its much larger numbers sooner.)

At the sentencing hearing, the district court overruled all objections to the PSRs, including defendants’ objections to loss-causation. The court stated that “losses to FDIC and TARP [were] caused by the ultimate failure of the bank,” but did not explain its rationale for that finding. At sentencing, Lonich’s counsel pressed for an explanation as to how Lonich could be responsible for SVB failing when he did not work at SVB and was involved in only the 101 Houseco loan, which was repaid in full. The district court declined to offer its reasoning.

Based on defendants’ offense levels in light of the proposed sentencing enhancements, the recommended Guidelines ranges were 235–293 months each for Cutting and Melland, and 292–365 months for Lonich. The district court departed downward, sentencing Cutting and Melland each to 100 months in prison, and Lonich to 80 months.

Even though the district court departed below the recommended Guidelines ranges, it accepted those ranges as the starting point because it agreed with the enhancements 54 UNITED STATES V. LONICH

used in the PSRs and rejected defendants’ objections to them. We thus must consider whether the government sufficiently demonstrated that defendants should receive certain enhancements—most centrally a 20-level loss amount enhancement—based on their alleged role in causing SVB to fail.

“[T]he government bears the burden of proof on the facts underlying a sentence enhancement.” United States v. Zolp, 479 F.3d 715, 718 (9th Cir. 2007). We review de novo the district court “selecting and properly interpreting the right Guidelines provision.” United States v. Gasca-Ruiz, 852 F.3d 1167, 1170 (9th Cir. 2017) (en banc). We review the district court’s application of the Guidelines to the facts for abuse of discretion, and its factual findings for clear error. Id.

Before we examine the government’s evidence that defendants caused SVB to fail, we address the standard of proof that the government was required to meet to demonstrate this factual point. We hold that the clear and convincing standard applies.

“As ‘a general rule,’ factual findings underlying a sentencing enhancement need only be found by a preponderance of the evidence.” United States v. Parlor, 2 F.4th 807, 816 (9th Cir. 2021) (quoting United States v. Valle, 940 F.3d 473, 479 (9th Cir. 2019)). But in some instances, a sentencing enhancement has an “an extremely disproportionate impact on the sentence.” Valle, 940 F.3d at 479 (quoting United States v. Jordan, 256 F.3d 922, 930 (9th Cir. 2001)). In those circumstances, we have held that due UNITED STATES V. LONICH 55

process may require the government to demonstrate facts underlying disputed enhancements by clear and convincing evidence. Parlor, 2 F.4th at 816–17; Valle, 940 F.3d at 479. 6

In determining when the government must meet a clear and convincing standard of proof, we have said that “[w]e look to the totality of the circumstances.” Valle, 940 F.3d at 479 (citing United States v. Pike, 473 F.3d 1053, 1057 (9th Cir. 2007)). In United States v. Valensia, 222 F.3d 1173, 1182 (9th Cir. 2000), and United States v. Jordan, 256 F.3d 922, 928 (9th Cir. 2001), we condensed the circumstances requiring a heightened standard into six non-exhaustive factors:

(1) whether “the enhanced sentence fall[s] within the maximum sentence for the crime alleged in the indictment;” (2) whether “the enhanced sentence negate[s] the presumption of innocence or the prosecution’s burden of proof for the crime alleged in the indictment;” (3) whether “the facts offered in support of the enhancement create new offenses requiring separate punishment;” (4) whether “the increase in sentence [is] based on the extent of a conspiracy;” (5) whether “the increase in the number of offense levels [is] less than or equal to four;”

6 We recognize that other circuits have held that “due process does not require sentencing courts to employ a standard higher than preponderance-of-the-evidence, even in cases dealing with large enhancements.” United States v. Jones, 829 F.3d 476, 477 (6th Cir. 2016) (per curiam) (quoting United States v. Brika, 487 F.3d 450, 462 (6th Cir. 2007)); see also United States v. Grubbs, 585 F.3d 793, 802– 03 & n.5 (4th Cir. 2009). But we are bound by our precedent, which clearly requires a heightened standard in some circumstances. 56 UNITED STATES V. LONICH

and (6) whether “the length of the enhanced sentence more than double[s] the length of the sentence authorized by the initial sentencing guideline range in a case where the defendant would otherwise have received a relatively short sentence.”

Jordan, 256 F.3d at 928 (alterations in original) (quoting Valensia, 222 F.3d at 1182); see also Parlor, 2 F.4th at 817; Valle, 940 F.3d at 479–80. In evaluating these factors, “we consider only the cumulative effect of ‘disputed enhancements.’” Parlor, 2 F.4th at 817 (quoting Jordan, 265 F.3d at 927).

Even after our articulation of the six Valensia factors, we have frequently commented that in this area, “[o]ur case law has ‘not been a model of clarity.’” Parlor, 2 F.4th at 817 (quoting Valle, 940 F.3d at 479 n.6); see also United States v. Hymas, 780 F.3d 1285, 1289 (9th Cir. 2015) (same); United States v. Berger, 587 F.3d 1038, 1048 (9th Cir. 2009) (same). But some clarity can be found in how our cases apply the factors.

The first two factors—whether the enhanced sentence falls within the maximum sentence allowed and whether it negates the presumption of innocence, Valensia, 222 F.3d at 1182—are to some extent eclipsed by subsequent developments in Sixth Amendment case law, including that the Sentencing Guidelines are now merely advisory in nature. See United States v. Booker, 543 U.S. 220 (2005); Blakely v. Washington, 542 U.S. 296 (2004); Apprendi v. New Jersey, 530 U.S. 466 (2000). In fact, in articulating its first factor, Valensia itself noted the Supreme Court’s then- recent holding in Apprendi that “other than the fact of a prior conviction, any fact that increases the penalty for a crime UNITED STATES V. LONICH 57

beyond the prescribed statutory maximum must be submitted to a jury, and proved beyond a reasonable doubt.” Valensia, 222 F.3d at 1182 n.4 (quoting Apprendi, 530 U.S. at 490).

We have further commented that “it is not entirely clear how the first three Valensia factors were derived from our decision in United States v. Restrepo, 946 F.3d 654 (9th Cir. 1991) (en banc), which Valensia cited as their source.” Valle, 940 F.3d at 479 n.6. As a three-judge panel, we cannot eliminate certain parts of a six-part test. See Miller v. Gammie, 335 F.3d 889, 899–900 (9th Cir. 2003) (en banc). But at the same time, it is appropriate to recognize that in practice, and as our case law has developed, the first two Valensia factors appear to do little independent work in driving the analysis.

With an important caveat about the fourth Valensia factor relating to conspiracies—which we will discuss in a moment—the real action is in Valensia factors five and six. We have repeatedly recognized that our cases commonly turn on the last two factors: whether the enhanced sentence is four or more offense levels higher (factor 5) and more than double the initial sentencing range (factor 6). See, e.g., Parlor, 2 F.4th at 817 (“Later cases . . . have focused specifically on the last two factors.”); Valle, 940 F.3d at 479 (explaining that Jordan and Valensia themselves “disregarded the first four factors” and that “more recent cases have also relied on only the[] last two factors”).

Focusing on factors five and six makes sense when one considers why under our cases a clear and convincing standard of proof is sometimes required. In the typical case, the last two Valensia factors best capture the principle underlying our precedents, which is that when disputed sentencing enhancements significantly increase the sentence 58 UNITED STATES V. LONICH

that would otherwise apply, due process can require the government to make a stronger showing. See Valle, 940 F.3d at 480.

Most commonly, the fifth and six factors will coincide: when the offense levels go up substantially, this will at some point generate a sentence that is more than twice the length of the “relatively short sentence” that would have otherwise applied. See, e.g., id. (holding that the clear and convincing standard applied when disputed enhancements led to an 11- level increase in offense level and “far more than doubled [defendant’s] sentencing range”); United States v. Gonzalez, 492 F.3d 1031, 1039–40 (9th Cir. 2007) (requiring clear and convincing evidence when disputed enhancements resulted in a 9-level increase in offense level and raised Guidelines range “more than four times”); United States v. Mezas de Jesus, 217 F.3d 638, 643 (9th Cir. 2000) (applying the clear and convincing standard when challenged enhancements resulted in a 9-level increase and more than doubled original Guidelines range).

What happens when the fifth Valensia factor is met, but the sixth is not? Consistent with the objective of applying a heightened standard of proof only when “the combined effect of contested enhancements would have ‘an extremely disproportionate effect on the sentence imposed,’” United States v. Garro, 517 F.3d 1163, 1168–69 (9th Cir. 2008) (quoting United States v. Staten, 466 F.3d 708, 718 (9th Cir. 2006)), we have recognized that district courts may apply a preponderance of the evidence standard, notwithstanding an increase in the offense level of four or more, when the sentence did not otherwise double. See Parlor, 2 F.4th at 817 (holding that district court did not plainly err when the disputed enhancements “did increase [defendant’s] offense level by more than four points,” but when they “did not more UNITED STATES V. LONICH 59

than double his recommended Guidelines range”); United States v. Riley, 335 F.3d 919, 927 (9th Cir. 2003) (same).

If we consider only the fifth and six factors, the result for defendants here is obvious: the clear and convincing standard must apply. All defendants challenge the 20-level enhancement for losses between $9.5 and $25 million, see U.S.S.G. § 2B1.1(b)(1)(K), and the 2-level enhancement for 10 or more victims, see id. § 2B1.1(b)(2)(A)(i). Lonich (but not Cutting and Melland) also challenges his 4-level enhancement for substantially jeopardizing the safety and soundness of SVB. See id. § 2B1.1(b)(17)(B)(i). For two defendants, the disputed enhancements increased their Guidelines offense level by 22 levels. For Lonich, it was a 26-level increase.

Unsurprisingly, these substantial total offense-level increases produced advisory Guidelines ranges for all three defendants that were far more than double what they would have obtained absent the SVB-related enhancements. Absent just the 20-level enhancement for causing $9.5 million or more in loss, U.S.S.G. § 2B1.1(b)(1)(K), Cutting and Melland’s recommended sentencing ranges would have dropped from 235–293 months to 27–33 months. See U.S.S.G. Ch. 5, Pt. A (table). And Lonich’s recommended sentencing range would have plummeted from 292–365 months to 33–41 months. See id. Valensia factors five and six thus strongly counsel in favor of applying a clear and convincing standard to factual findings underlying defendants’ disputed enhancements.

But what if the conduct that produced the monetary loss is conduct for which the jury convicted the defendants? This 60 UNITED STATES V. LONICH

is where we encounter the government’s principal argument, and where the fourth Valensia factor comes into play.

The fourth factor is whether “the increase in sentence [is] based on the extent of a conspiracy.” Valensia, 222 F.3d at 1182. Although this factor is focused on conspiracy charges, it is important to appreciate that it is framed in that manner only because conspiracy convictions can present borderline cases about whether certain conduct was within the scope of convicted conduct. Ultimately, the fourth Valensia factor is just an example of another broader principle: if a defendant has already been convicted of certain conduct (whether through a jury verdict or a guilty plea), enhancements that are based on the conduct of conviction do not require proof by clear and convincing evidence. This is essentially what the third Valensia factor—whether the facts offered in support of the enhancement create new offenses—addresses in the non- conspiracy context. See Valensia, 222 F.3d at 1182; see also United States v. Barragan, 871 F.3d 689, 718 (9th Cir. 2017) (discussing third Valensia factor); United States v. Johansson, 249 F.3d 848, 854–55 (9th Cir. 2001) (same).

The justification for the broader principle, of which the fourth Valensia factor is an example, is that when an enhancement is based on conduct for which the jury found the defendant guilty beyond a reasonable doubt (or to which defendant pleaded guilty), any due process concerns associated with imposing enhancements based on this same conduct are correspondingly lower. “[T]he defendants had the opportunity at trial to challenge [the] evidence,” Hymas, 780 F.3d at 1292, and so a clear and convincing standard of proof is not warranted.

Our decision in United States v. Garro, 517 F.3d 1163 (9th Cir. 2008), is instructive. In Garro, the defendant was UNITED STATES V. LONICH 61

convicted of wire fraud and other offenses and sentenced to 135 months in prison. Id. at 1165. Applying the Guidelines enhancements in place at the time, the district court imposed a 16-level increase for the defendant causing losses that exceeded $20 million. Id. at 1167. This enhancement had a dramatic effect on defendant’s sentence because his crime had a base offense level of just six. Id. On appeal, the defendant argued that the government was required to prove the loss amount by clear and convincing evidence, rather than by a preponderance. Id. at 1168.

We disagreed, explaining that “[i]n identifying the appropriate standard of proof, we have distinguished between enhancements based upon charged conduct for which the defendant has been convicted, and enhancements based upon uncharged conduct.” Id. at 1169. Notwithstanding the substantial effect of a 16-level loss enhancement on defendant’s sentence, we held that because defendant’s “sentence for loss exceeding $20 million was based on conduct for which [defendant] was charged and convicted,” the government only had to prove the amount of the loss at sentencing by a preponderance of the evidence. Id.

Garro presented a straightforward case. Under our precedents, another classically straightforward case would arise if the conduct that forms the basis for the enhancement is clearly not the conduct for which defendant was charged and convicted. Our decision in United States v. Mezas de Jesus, 217 F.3d 638 (9th Cir. 2000), provides an example of this type of situation. There, the defendant was convicted of being an undocumented immigrant in possession of a firearm. Id. at 639. But at sentencing, the government argued that the defendant had committed this offense during a kidnapping, which was uncharged conduct. Id. The 62 UNITED STATES V. LONICH

district court found that the defendant possessed the firearm in connection with the uncharged kidnapping. This resulted in a 9-level upward adjustment and an increase in the defendant’s sentencing range from 21–27 months to 57–71 months. Id. at 643. This, we held, satisfied the “extremely disproportionate” impact test, such that the kidnapping would need to be established by clear and convincing evidence. Id. at 645.

Decisions like Garro and Mezas de Jesus show that in considering whether the clear and convincing or preponderance standard should apply, it is not enough simply to examine the effect on the sentence of the disputed enhancements. Many of our cases apply the preponderance standard to very large enhancements. See United States v. Treadwell, 593 F.3d 990, 1001 (9th Cir. 2010) (22-level increase), rev’d on other grounds by United States v. Miller, 953 F.3d 1095 (9th Cir. 2020); Garro, 517 F.3d at 1167 (9th Cir. 2008) (16-level increase); United States v. Sanchez, 967 F.2d 1383, 1384, 1385–87 (9th Cir. 1992) (14-level increase). Others have held that smaller enhancements nonetheless required the higher clear and convincing standard. See, e.g., Mezas de Jesus, 217 F.3d at 643 (9-level increase); United States v. Hopper, 177 F.3d 824, 833 (9th Cir. 1999) (7-level increase).

These holdings are consistent because the critical issue in these cases was whether the enhancements were based on the conduct of conviction. If they are based on such conduct, the preponderance of the evidence standard applies. See Garro, 517 F.3d at 1168–69. But if they are based on conduct for which the defendant was not convicted, the clear and convincing standard may apply. See Mezas de Jesus, 217 F.3d at 642–43. UNITED STATES V. LONICH 63

Our conspiracy cases—and our application of Valensia factor four—turn on this very same dichotomy, but often present difficult questions in more complicated factual scenarios about whether certain conduct is, in fact, based on the conduct of conviction. We have “declined to apply the clear and convincing standard of proof [when] the enhancement at issue was ‘based entirely on the extent of the conspiracy.’” Garro, 517 F.3d at 1169 (quoting Riley, 335 F.3d at 926); see also, e.g., Valle, 940 F.3d at 480 n.8 (summarizing this line of cases). That is true regardless of whether the disputed sentencing enhancements resulted in an increase of the offense level by more than four points or whether it resulted in a Guidelines range that more than doubled. See, e.g., Barragan, 871 F.3d at 718 (holding that the preponderance of the evidence standard applied when enhancement was based on the scope of conspiracy, even though it resulted in a 17-level increase in the offense level and quadrupled the length of the sentencing range); Treadwell, 593 F.3d at 1001–02 (holding that the preponderance of the evidence standard applied when enhancement was based on the scope of the conspiracy, even though it resulted in a 22-level increase that multiplied the sentencing range tenfold); Berger, 587 F.3d at 1041, 1048– 49 (holding that the preponderance of the evidence standard applied, even though enhancement resulted in 14-level increase in the offense level and more than quadrupled the length of the sentencing range).

The rationale for our approach in the conspiracy context once again lies in the distinction between those enhancements that are based on convicted conduct and those that are not. “Enhancements based on the extent of a conspiracy,” we have reasoned, “are ‘on a fundamentally different plane than’ enhancements based on uncharged or acquitted conduct.” United States v. Armstead, 552 F.3d 64 UNITED STATES V. LONICH

769, 777 (9th Cir. 2008) (quoting Riley, 335 F.3d at 926). The justification is the same as in non-conspiracy cases: “due process concerns . . . are satisfied by a preponderance of the evidence standard because the enhancements are based on criminal activity for which the defendant has already been convicted.” Id.

The government argues that a preponderance of the evidence standard should apply here because the jury found defendants guilty of a fraudulent conspiracy, and the district court could then find that this criminal conduct caused the bank to fail and produced the government’s related losses. We reject the government’s position for two reasons.

First, there is a notable mismatch between the scope of the criminal convictions and the losses that supposedly drove the bank’s failure. The PSRs described how the FDIC paid out of its Deposit Insurance Fund $11.47 million. Of this, approximately $10.3 million consisted of Madjlessi-related loans that the government required SVB to write off. But $3.27 million of these write-offs related to a loan for 132 Village Square, another Madjlessi-controlled entity, for which the defendants were not charged with wrongdoing. Also included within the $10.3 million was $3.88 million in loans for the first two pairs of Greenbriar loans. But the jury acquitted Cutting of making a false bank entry for these two loans. As to Cutting in particular, over $7 million of the $10.3 million in loans thus reflected uncharged or acquitted conduct. And all these loans pre-dated Lonich even joining the conspiracy.

This is therefore not a situation where the losses are “based entirely on the extent of the conspiracy.” Riley, 335 F.3d at 926. Instead, by the terms of the PSRs and the FDIC report on which they rely, a substantial amount of uncharged and acquitted conduct contributed to the bank’s UNITED STATES V. LONICH 65

collapse, and thus the claimed losses. When uncharged or acquitted conduct is in the mix to such an extent, the mere fact of the conspiracy convictions—and thus Valensia factor four—is not dispositive. See Hymas, 780 F.3d at 1290–93 (holding that a preponderance of the evidence standard applied for losses associated with convicted conduct, but that a clear and convincing standard applied for losses based on uncharged loans); Armstead, 552 F.3d at 777.

Second, unlike our past conspiracy cases, this case involves a substantial intermediate causation question: to conclude that defendants’ conspiratorial conduct caused the government’s losses, the district court had to determine that defendants’ criminal conduct caused the bank’s collapse. The logic of our decisions again does not support applying a preponderance of the evidence standard in this situation. The jury’s guilty verdicts do not compel the conclusion—or even plausibly demonstrate—that the defendants through their criminal conduct were responsible for SVB’s collapse. In fact, the jury was not even permitted to hear evidence on why the bank failed.

The preponderance of the evidence standard might have been appropriate if, for example, the loss enhancements were based on the value of defaulted Madjlessi loans. But here, the linchpin of the disputed enhancements turns on a wholly separate causal inquiry—the bank’s failure—that is thoroughly disconnected from the jury’s verdict. It thus cannot be said that defendants “had ample opportunity at trial to challenge the government’s evidence of the extent of losses caused by the conspiracy.” Treadwell, 593 F.3d at 1001.

In short, Valensia’s fourth factor does not govern here. The fifth and sixth factors do. Given the extremely disproportionate sentences that the disputed enhancements 66 UNITED STATES V. LONICH

produced, a clear and convincing standard applies to the factual underpinnings for these enhancements. This reflects the trajectory of our precedents interpreting Valensia. The first two Valensia factors are unlikely to add independent weight to the analysis and do not do so here. The third and fourth factors are effectively a threshold inquiry that asks whether the enhancement is based on the conduct of conviction, which, if so, means that the preponderance of the evidence standard applies. Because the third and fourth factors do not apply on these facts, we thus proceed to the fifth and sixth factors, which require a clear and convincing standard in this case.

We next turn to whether the government showed by clear and convincing evidence that defendants caused SVB to fail. This standard “requires that the government ‘prove [its] case to a higher probability than is required by the preponderance-of-the-evidence standard.’” Jordan, 256 F.3d at 930 (alteration in original) (quoting California ex rel. Cooper v. Mitchell Bros.’ Santa Ana Theater, 454 U.S. 90, 93 n.6 (1981)). Under this standard, the factfinder must have “‘an abiding conviction that the truth of [the] factual contentions’ at issue is ‘highly probable.’” Mondaca-Vega v. Lynch, 808 F.3d 413, 422 (9th Cir. 2015) (en banc) (alteration in original) (quoting Colorado v. New Mexico, 467 U.S. 310, 316 (1984)); see also Black’s Law Dictionary 698 (11th ed. 2019) (clear and convincing evidence requires “indicating that the thing to be proved is highly probable or reasonably certain”). We hold that the government did not show by clear and convincing evidence that defendants caused SVB to fail.

The Guidelines require that a defendant’s sentence “be based on ‘all harm that resulted from the acts or omissions’ UNITED STATES V. LONICH 67

of the defendant.” United States v. Hicks, 217 F.3d 1038, 1048 (9th Cir. 2000), as amended on denial of reh’g (July 31, 2000) (quoting U.S.S.G. § 1B1.3(a)(3)). The term “‘resulted from’ establishes a causation requirement,” which includes both cause-in-fact (but-for causation) and proximate cause. Id. at 1048–49; see also United States v. Peppel, 707 F.3d 627, 644 (6th Cir. 2013) (explaining that under the Guidelines, “[c]ausation includes two distinct principles, cause in fact, or what is commonly known as ‘but for’ causation, and legal causation.” (quoting United States v. Rothwell, 387 F.3d 579, 583 (6th Cir. 2004)).

These basic causation requirements apply to loss enhancements. The Guidelines define “loss” to include not only actual loss and intended loss, but also “reasonably foreseeable pecuniary harm.” U.S.S.G. § 2B1.1 cmt. n.3(A)(i). This “import[s] the legal concept of a causal relationship between the defendant’s conduct and the determined loss.” Rothwell, 387 F.3d at 583. The Guidelines’ loss rules thus “do[] not obviate the requirement to show that actual, defendant-caused loss occurred.” Berger, 587 F.3d at 1045; see also U.S.S.G. app. C vol. II amend. 617, at 178 (2003) (explaining that § 2B1.1 “incorporates [a] causation standard that, at a minimum, requires factual causation (often called ‘but for’ causation) and provides a rule for legal causation (i.e., guidance to courts regarding how to draw the line as to what losses should be included and excluded from the loss determination)”). Applying this principle, we have vacated sentences when the government failed to produce sufficient evidence to show proximate or but-for cause for asserted loss amounts. See Berger, 587 F.3d at 1046–47; Hicks, 217 F.3d at 1047–49. 68 UNITED STATES V. LONICH

In this case, the district court made no independent findings about the cause of the bank’s collapse beyond adopting the PSRs and rejecting defendants’ objections without explanation. But neither the PSRs nor the additional materials the government now cites sufficiently show that defendants were responsible for SVB failing, especially given indications in the record that other factors internal and external to the bank may have contributed to the bank’s collapse.

First, although the PSRs stated that defendants had caused the bank to fail, they cited only the FDIC report. Relying exclusively on that report, the PSRs explained that regulators “were forced to downgrade SVB’s capital due to the relationship between the bank and Bijan Madjlessi,” and that before the downgrade, the Madjlessi relationship “represented 74% of Total Risk Based capital.” The PSRs then found that “[a]fter this downgrade from the FDIC, the bank failed as it was unable to provide the necessary funding to stay afloat because 74% of their risks were taken up by Bijan Madjlessi.”

There are several difficulties with the reliance on the FDIC report. For one, the report did not find that defendants caused SVB’s failure. The FDIC issued its report before SVB’s failure, and the report instead concerned the FDIC downgrading SVB. The FDIC report also does not blame all the bank’s problems on the Madjlessi-related loans. While the FDIC report noted its concern for SVB concentrating too much capital in Madjlessi, it also mentioned a host of other issues besetting SVB, including poor management, deteriorating market conditions, concentrated lending to another larger borrower, poor financial reporting, and so on. Madjlessi is mentioned by name in only one paragraph in the eleven-page report. UNITED STATES V. LONICH 69

The government argues in its briefing that the other problems the FDIC report identifies are “all hallmarks of the Madjlessi relationship.” But setting aside that the FDIC report does not connect all the bank’s problems to Madjlessi, that the Madjlessi loans may have reflected the bank’s otherwise poor practices does not mean those loans (much less defendants’ criminal conduct) are, standing alone, what caused the bank to fail. Based on the FDIC report, it is not clear whether SVB’s collapse was caused by defendants’ conspiracy-related loans or by other “intervening” and “independent” factors, including outside economic forces. Hicks, 217 F.3d at 1049.

Second, neither the PSRs nor the FDIC report focused on those loans for which the jury convicted defendants. The PSRs referenced the total amount of loans to Madjlessi— $35,000,000. But not all the loans comprising this amount were claimed to be unlawful. The “74%” figure referenced in the PSR and FDIC report related to all Madjlessi-related loans. And even then, it is not apparent from the FDIC report that it is properly interpreted as meaning that 74% of the bank’s risks were taken up by Madjlessi-related loans.

The FDIC report similarly referenced the $10.3 million write-off that the FDIC ordered SVB to take on Madjlessi- related loans. But as we noted above, this write-off included an uncharged $3.27 million loan for 132 Village Square, and $3.88 million for the first two pairs of Greenbriar loans for which Cutting was acquitted. And all the loans that were written off pre-dated Lonich’s involvement in the conspiracy. From a causation perspective, it is thus unclear if the loans associated with defendants’ criminal wrongdoing led to the bank’s failure.

Third, the government on appeal now points to three “investigative reports” it included in its sentencing 70 UNITED STATES V. LONICH

memorandum. But the PSRs did not rely on any of these investigative reports. And there is no indication the district court did either. If anything, the court discounted the government’s sentencing memorandum after finding the government belatedly changed its calculations and provided conflicting numbers.

The three investigative reports have other important limitations. They are essentially summaries of interviews of two California bank examiners and one federal examiner. Although these examiners generally opined that the defendants caused the bank to fail, the written reports do not identify the bases for those conclusory opinions. For example, one state examiner reportedly “felt that the Bijan Madjlessi loans exceeding the [legal lending limit] caused” the bank’s failure. The reports also do not discuss the role of other potential factors flagged in the FDIC report, such as adverse economic conditions or the bank’s overall poor management practices.

Fourth, while the government’s causation theory lacked support as to all defendants, it was particularly lacking as to Lonich, who did not join the conspiracy until January 2009. All damages listed in the government’s calculation table related to loans issued before Lonich joined the conspiracy. “A defendant’s relevant conduct does not include the conduct of members of a conspiracy prior to the defendant joining the conspiracy.” U.S.S.G. § 1B1.3, cmt. n.3(B); see also United States v. Bad Wound, 203 F.3d 1072, 1077–78 (8th Cir. 2000) (vacating sentence because the district court did not make a finding on when the defendant joined the conspiracy, so that the court could not tell what percentage of the loss should be attributable to him).

In response, the government points out that it charged Lonich in connection with the 101 Houseco loan. But that UNITED STATES V. LONICH 71

loan was fully paid and accounted for zero loss according to the government’s own calculations. It was also secured by collateral worth twice the amount of the loan itself. It is thus hard to see how Lonich—who did not even work at the bank and was involved in only one loan—can be said to have caused the bank’s downfall, especially given the other potential causal factors at play. The government tries to advance a theory by which the 101 Houseco loan, while fully paid, nonetheless contributed to the bank’s problems. But neither the PSR nor the district court made findings on that point, and the government’s logic is far from self-evident.

In short, the government did not demonstrate by clear and convincing evidence that defendants caused SVB to fail. 7 This means that the government did not sufficiently support defendants’ 20-level loss enhancement. Perhaps the government, at resentencing on an open record, can prove its theory that defendants were responsible for SVB’s failure. We hold only that, on this record, defendants’ 20-level loss enhancement cannot be sustained. 8

Our determination that the government did not adequately prove defendants caused SVB to fail means that other aspects of defendants’ sentences are infirm as well. The district court imposed a 2-level sentencing enhancement under U.S.S.G. § 2B1.1(b)(2)(A)(i) because defendants’ offenses involved ten or more victims. But these victims consisted of the FDIC, TARP, and shareholders of the bank.

7 Although we question whether the government even met the preponderance of the evidence standard, we need not address that issue in light of our holding as to the appropriate standard. 8 We thus do not reach defendants’ alternative argument that even if they caused SVB to fail, there was insufficient evidence supporting the monetary losses identified in the PSR. 72 UNITED STATES V. LONICH

This enhancement thus likewise depended on the finding that defendants were responsible for the bank’s failure. Lonich’s 4-level enhancement for jeopardizing the safety and soundness of a financial institution, U.S.S.G. § 2B1.1(b)(17)(B)(i), encounters the same problem. 9

The same is true of defendants’ approximately $20 million restitution orders, which were likewise premised on the government’s theory that defendants caused the bank to fail. While the standard of proof for restitution is a preponderance of the evidence, Hymas, 780 F.3d at 1293 n.4, because we are already vacating defendants’ sentences due to the lack of demonstrated causal relationship between their offenses and the bank’s collapse and remanding for resentencing, we likewise vacate the restitution order as well. We thus do not reach defendants’ other assignments of error as to the restitution award.

* * *

For these reasons and those set forth in our accompanying memorandum disposition, we affirm defendants’ convictions. But we vacate defendants’ sentences and remand for resentencing on an open record. See United States v. Matthews, 278 F.3d 880, 885 (9th Cir. 2002) (en banc) (“[A]s a general matter, if a district court errs in sentencing, we will remand for resentencing on an open record—that is, without limitation on the evidence that the district court may consider.”).

AFFIRMED in part, VACATED in part, and REMANDED.

9 Cutting and Melland do not challenge this enhancement as to them.