United States v. Michael Busch

NOT RECOMMENDED FOR PUBLICATION FILE NAME: 21A0507N.06

Case Nos. 20-4065/4066/4067

UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT

) FILED UNITED STATES of AMERICA, Nov 04, 2021 ) DEBORAH S. HUNT, Clerk ) Plaintiff-Appellee, ) ) ON APPEAL FROM THE UNITED v. ) STATES DISTRICT COURT FOR ) THE SOUTHERN DISTRICT OF MICHAEL D. BUSCH (20-4065); BUSCH’S ) OHIO COUNTRY CORNER (20-4066); ) AMANDA JO BUSCH (20-4067), ) ) Defendants-Appellants. )

Before: SUTTON, Chief Judge; BATCHELDER and LARSEN, Circuit Judges.

ALICE M. BATCHELDER, Circuit Judge. In a joint trial, a jury convicted Michael Busch, Amanda Busch, and Busch’s Country Corner, Inc. (collectively, “the Busches”) on charges of conspiracy to commit offenses against the United States, food-stamp fraud, and wire fraud. In a consolidated appeal, they claim a Brady violation, an incorrect jury instruction, and prosecutorial misconduct. We find no reversible error as to any of these claims and AFFIRM.

I. Background

The United States Department of Agriculture’s (USDA’s) modern food-stamp program is known as the Supplemental Nutrition Assistance Program (SNAP). It provides food-purchasing assistance to eligible, low-income individuals and households through an Electronic Benefit Transfer (EBT) account with an associated debit card.1 The USDA—in conjunction with a corresponding state agency—deposits a monthly dollar-value allotment of food-purchasing credit

1 See United States v. Busch, No. 3:18-cr-00079 (S.D. Ohio) (Dkt. No. 48, “Second Superseding Indictment,” Feb. 28, 2019); see also USDA “Supplemental Nutrition Program (SNAP) Fact Sheet” (FNS-813, Dec. 2018), available at https://www.usda.gov/sites/default/files/documents/snap_fact_sheet.pdf (last visited Oct. 1, 2021).

Case Nos. 20-4065/4066/4067, United States v. Busch, et al. (money) into the EBT account and the account holder uses the EBT debit card to purchase eligible food items at authorized retailers. To become a SNAP-authorized retailer, the store—e.g., butcher, grocer, supermarket, convenience store—must apply for and obtain a SNAP license from the USDA. All EBT transactions are conducted through an electronic point-of-sale (POS) system, which debits the user’s monthly allotment, electronically deposits the corresponding payment to the retailer’s bank account, and records the transaction. Retailers may accept EBT payment “only in exchange for eligible food,” not “in exchange for cash.” 7 C.F.R. § 278.2(a). Nor may a retailer accept EBT “in payment for items sold to a household on credit.” 7 C.F.R. § 278.2(f).

Michael and Amanda Busch owned and operated Busch’s Country Corner (BCC), a specialty butcher and grocery store in the Findlay Market, in Cincinnati, Ohio. BCC was a SNAP-authorized retailer and it processed 195,113 EBT transactions (i.e., about 500 per week) between October 2010 and March 2018.2 Randall Busch, Michael’s brother, managed the day-to-day operations at BCC. Michael was the boss but was ordinarily present at BCC only in the mornings. Amanda was the BCC bookkeeper but worked from a home office and was seldom at BCC.

In May 2017, a taskforce of USDA Agents, Secret Service Agents, and local police received a tip that BCC was trading cash for EBT charges and opened an investigation. In the 13 months between May 2017 and April 2018, an informant working for the taskforce traded EBT credit for cash (i.e., conducted “controlled buys”) at BCC on eleven different days. For each of those controlled buys, the taskforce provided the informant with one or more EBT cards (as many as four), each with a different, fictitious accountholder name. The taskforce also equipped the informant with multiple audio-video recording devices: one in his hat, another in his coffee cup,

2 The conspiracy alleged in this case, as detailed in the indictment, spanned January 2010 through April 2018 (100 months). But for reasons that have no bearing on the case or any issue in this appeal, the database queries used to assess BCC’s EBT transactions spanned only October 2010 through March 2018 (90 months).

Case Nos. 20-4065/4066/4067, United States v. Busch, et al. and an open line on his cell phone so that agents could watch and listen in real time. At BCC, the informant interacted with Randall, who processed the fraudulent EBT transactions through BCC’s POS card reader, debiting an amount from each of the EBT cards (accounts) and giving the informant half of that amount back in cash. Because the full amount of the charge was deposited electronically into BCC’s bank account, BCC kept the other half of the fraudulent charge as profit. For example, if the informant offered a card with the name “Jane Smith” and an available credit of $911, Randall would charge $910 (which the USDA would transfer electronically to BCC’s bank account) and pay the informant $455 in cash, thus retaining $455 in profit for BCC.3 Michael was present in the videos for about half of the events and though he did not interact directly with the informant, he twice participated by providing Randall with additional cash needed to complete the trade.

Meanwhile, as part of its standard procedure in SNAP-fraud investigations, the taskforce obtained a summary of BCC’s average monthly EBT activity: dollar amount transacted, number of transactions, and dollar amount per transaction. The USDA logs every EBT transaction in its “Antifraud Locator Using EBT Retailer Transaction” (ALERT) database, which includes itemized transaction information, such as the date, time, retailer identification number, EBT card/account number, card starting balance, charge, and card ending balance. Here, a query of the ALERT database produced a summary report of the total aggregate and monthly averages of BCC’s EBT transactions and dollar amounts for the 90-month period from October 2010 through March 2018.

3 Due to the use of multiple EBT cards, the eleven undercover visits produced 20 EBT trades. Two of those trades were as depicted in the example here: an EBT card containing $911 of available EBT credit and a BCC charge of $910, leaving $1 on the card. To add some perspective, the average beginning amount on these 20 EBT cards was $706 (with a high of $1,300 and a low of $200) and the average BCC charge was $700, effectively charging the entire card amount (only one trade left more than $11 remining on the card and four left zero). Fifteen of the 20 charges were whole numbers (no cents). The other five were all in five-cent (nickel) increments, two of which simply emptied the card (e.g., a card containing $1,289.95 and a BCC charge of $1,289.95, leaving a zero balance on the card).

Case Nos. 20-4065/4066/4067, United States v. Busch, et al. The same report was produced for several other SNAP-authorized retailers located in Findlay Market that also had the same USDA designation (“meat specialty store”) as BCC.

A taskforce agent reviewed these summary reports, considered six of those other SNAP-authorized retailers to be appropriate “comparators,” and calculated a cumulative average of the monthly averages for those six comparators. BCC’s monthly averages for both EBT transactions and EBT dollar amounts were significantly higher than the corresponding cumulative average for the comparators over the designated period. The agent assumed that this difference was evidence of SNAP fraud. The taskforce used this information in preparing search-warrant applications.

In May 2018, taskforce agents executed three search warrants: one at BCC, one at Randall’s home, and a third at Michael and Amanda’s home. Among the evidence seized were bookkeeping records from Amanda’s home office, which included, among other things, some year-to-year accounting comparisons prepared by BCC’s external accountants. Two of them (2017 and 2018) had handwritten notes from the accountant asking Amanda how BCC’s profits were increasing while its sales were decreasing. The agents also seized $191,135 in cash from a locked safe in Michael and Amanda’s basement. Meanwhile, agents obtained call and text records from the Busches’ cell phones, and video recordings from the Findlay Market surveillance system. BCC had its own video surveillance system, which the Busches (particularly Amanda) could access remotely, but the agents did not obtain recordings from that system.

The grand jury indicted Michael, Amanda, Randall, and BCC on numerous counts, comprising multiple charges of conspiracy to commit offenses against the United States, SNAP fraud, and wire fraud.4 See 7 U.S.C. § 2024(b), 18 U.S.C. §§ 2, 371 & 1343. The simple version of the prosecution’s theory was that Randall and Michael used BCC’s SNAP authorization to trade

4 The final indictment—the second superseding indictment—included a count charging conspiracy to commit money laundering, 18 U.S.C. § 1956(h), but the court dismissed that money-laundering charge prior to trial.

Case Nos. 20-4065/4066/4067, United States v. Busch, et al. cash for EBT payments on a 50% basis, and Michael and Amanda restocked the cash to facilitate those fraudulent trades by withdrawing cash from the BCC bank account into which the USDA deposited the EBT payments electronically. The indictment relied on the 13-month investigation, including witness statements, the undercover controlled buys, and the results of the search warrants. The indictment also referred to and relied on Ohio’s EBT transaction database, known as “Conduent,” which is the functional equivalent of the USDA’s ALERT database, to estimate the economic loss and accuse the Busches of almost $3.5 million in SNAP and wire fraud. That is, the prosecutor—like the taskforce agent—assumed that the difference between BCC’s EBT sales and the cumulative average of EBT sales for the six comparators was all due to fraud. Specifically, BCC’s average monthly EBT sales ($60,151 per month) minus the cumulative average for the six comparator businesses ($21,807 per month) was the amount directly attributable to SNAP fraud ($38,344 per month). Applying that number to the 90-month period produced a total estimated fraud amount of $3,450,960, which was 63.74% of BCC’s total EBT sales of $5,413,629.

Michael, Amada, and BCC pleaded “not guilty” to all charges. Randall entered a guilty plea pursuant to a plea agreement and agreed to testify against the others. The court accepted his plea, convicted him of 20 counts of SNAP fraud, and sentenced him to five years of probation.

The prosecutor tried the remaining defendants to a jury. In a trial that spanned two weeks— from June 10 to 21, 2019—the prosecutor presented 18 witnesses. The key witness was the informant, who authenticated, narrated, and discussed the recordings of the controlled buys. From the recordings, the jurors saw for themselves that the transactions were conducted at the BCC counter, observable by anyone in the small BCC shop and with Michael’s concurrence or cooperation. The informant testified that both Randall and Michael participated, that everyone at BCC knew it was happening, and that he had seen others selling their EBT benefits for cash at *6 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. BCC. Another witness testified that he had been a “runner”: he brought EBT cards to BCC for Randall or Michael to execute a transaction, after which they gave him a paper bag to return to the card’s owner. He testified that he only once saw what was in the paper bag (cash) but that the cardowners would routinely withdraw the contents immediately and discard the bag. He estimated that he did these runs 30 to 40 times per week for upwards of 200 different people. He explained that he had a long relationship with the Busches and had worked for BCC occasionally, so he was a constant presence at the store, entering and exiting through the back door, lingering there during slow times, and routinely going behind the customer counter. This, combined with his familiarity with the local community, caused Michael to approach him to initiate his role in “running” these cards and paper bags. Another witness, a local EBT cardholder, testified that he had a “tab” at BCC, so he left his EBT card (and PIN) there for Michael and Randall to charge it as necessary to settle his tab.

Randall testified that Michael had taught him the fraudulent-EBT-transaction scheme. He estimated that he and Michael had each done five to ten of these transactions per week. He explained that all the proceeds went directly to BCC; he received no personal benefit. He testified that Amanda was aware of the scheme and could watch via BCC’s surveillance cameras. Another former BCC employee testified that she had seen Michael and Randall conducting the fraudulent cash-for-EBT trades, including those described by the “runner” who had testified earlier; and she testified that Michael and Amanda had authorized her to grant customers credit against future EBT charges, specifically corroborating the testimony by the cardholder who ran a “tab” and left his card at BCC. An adjacent business owner testified that he believed BCC had been trafficking EBT charges, that he frequently saw the “runner” making trades with Randall or Michael, and that when he would refuse a request for an illegal trade at his store, that cardholder would go to BCC. *7 Case Nos. 20-4065/4066/4067, United States v. Busch, et al.

Several taskforce agents testified about the investigation, including the eleven controlled buys, the executions of the search warrants, witness interviews, and other evidence obtained. The final taskforce witness discussed the ALERT database reports, the comparison of BCC’s monthly averages to the cumulative average for the six comparator stores, and the assumption that the difference—the amount of $3,450,960—was directly attributable to SNAP fraud. After crossexamination highlighted the flaws in this analysis and assumption, the prosecutor questioned the witness on re-direct about a summary chart prepared by the taskforce using the Conduent database and titled “Busch’s Country Corner Suspicious EBT Transaction January 2010 to April 2018,” which alleged $494,000 in suspicious transactions. This total was divided into categories of transactions that the taskforce and prosecutor considered suspicious, namely transactions executed in the morning before the store opened (before 8:00 a.m.) and transactions for exact dollar amounts of $100, $200, $300, or $400. On re-cross, the Busches’ defense counsel established that the agent had not interviewed the EBT cardholders associated with those transactions to inquire as to whether they were valid or fraudulent. This chart was not admitted as evidence.

The overall defense theory was that Randall had conducted the fraudulent EBT transactions on his own, without Michael’s or Amanda’s knowledge. The defense called seven witnesses, including Michael and Amanda, and four family members who had worked at BCC.5 The four family members testified consistently that they had never seen any fraudulent EBT transactions, were unaware of any fraud, and were shocked to learn of Randall’s perpetration of the fraudulent EBT transactions. Given this emphatic (and consistent) testimony that they never saw Randall commit any fraudulent EBT transactions, the obvious but likely unintended inference is that they

5 The other defense witness was an employee of Findlay Market, who provided maintenance and security services at the Market. Under questioning from defense counsel, he agreed that he had been called to BCC on occasion for security purposes, but he testified that he did not remember why.

Case Nos. 20-4065/4066/4067, United States v. Busch, et al. would not have seen Michael doing likewise. Amanda testified about her role at BCC as the bookkeeper, which included her accounting for the cash. One aspect of BCC’s cash management was referred to as “pocket cash” or “pocket,” which comprised large denomination bills ($100, $50, or $20) that Randall or Michael would remove from the register throughout the day for safekeeping, until these bills could be delivered to Amanda for accounting and deposit. An incongruous aspect of BCC’s cash management was, as Amanda testified, that she would withdraw cash from the BCC business bank account by writing and cashing a BCC business-account check, in order to provide change for the BCC cash register. She did this about once per week and a “typical amount” was about $1,250. She testified that Michael did likewise and would tell her when he did, so that she could record it in her bookkeeping notebook. When asked why she and Michael made such frequent trips to the bank to write checks for cash even though she had large amounts of cash on hand and was receiving cash deliveries daily via the large denomination “pocket,” her only answer was: “Because that’s how I was taught to do it.”

Michael testified that he would go to the bank two or three times per month to write and cash checks from the BCC account to provide change for the BCC register. He said the typical amount was between $500 and $2,000. Counterintuitively, he would correspondingly deliver large denomination “pocket” cash to Amanda, of $1,000 to $2,000, four or five days per week. When asked why he did not just exchange the large denomination “pocket” bills for smaller denomination bills, rather than writing checks from the BCC account, his answer was: “I probably didn’t have any pocket on me to do that.” Michael candidly admitted that he extended credit to EBT cardholders with the understanding that he would charge their cards when they received allotment from USDA, but that he did not consider it wrong.

The jury convicted the defendants as charged: Michael, Amanda, and BCC on the conspiracy charges; and Michael on the SNAP- and wire-fraud charges. Because the prosecutor *9 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. intended to argue that the $3.5-million amount was the proper “estimate of loss” for sentencing, the Busches hired an expert to prepare an opposing estimate. The expert requested, and the prosecutor provided, the “raw” or particularized EBT transaction data from the ALERT database. The expert used this data set, his personal experience, and certain assumptions, to conduct his own statistical analyses, from which he drew three conclusions. One, the prosecutor’s $3.5 million estimate was unreliable and demonstrably incorrect. Two, “[t]he correct estimate of loss based on the relevant data from the ALERT system is, statistically, effectively zero.” That is, according to the expert’s analysis, there was no fraud at all. It bears recognizing that this conclusion is irreconcilable with the direct evidence of fraud produced at trial, such as the videos of the fraudulent transactions and the testimony of several witnesses. Turning this around, if there was some fraud, as the evidence showed and the jury found, then this statistical analysis cannot be a reliable means of assessing fraud or estimating loss. But the expert did that analysis anyway, offering as his third conclusion an estimated loss-due-to-fraud of $15,526.75, based on the total of the recorded undercover transactions ($14,123.67), two suspicious $50 transactions ($100), and nine suspicious transactions that were executed on Mondays, when BCC was closed, and when Randall would likely have been at BCC alone ($1,303.08).

The court held an evidentiary hearing prior to sentencing, at which this expert presented his report and testified about it. Ultimately, his expert opinion was a reiteration of the report’s conclusions: his statistical analysis showed that there was no EBT fraud or, at most, only a minimal amount of loss due to fraud. At this hearing, defense counsel raised an accusation that the prosecutor’s failure to provide the full set of transaction data from the ALERT database prior to trial was a violation of Brady v. Maryland, 373 U.S. 83 (1963). The taskforce agent testified that he did not have, much less use or rely on, the raw transaction data that the Busches’ expert used; he had had and relied on only the monthly summaries. Therefore, he explained that while the raw *10 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. data was available in the ALERT database, it was not provided to the defense prior to trial because it was not provided to the taskforce or the prosecutor.

Following the hearing, the Busches moved to vacate their convictions, claiming that the prosecutor’s failure to provide the particularized transaction data prior to trial was a Brady violation. They argued that the data was exculpatory because their expert’s analysis concluded that, “statistically speaking, SNAP fraud at Busch’s Country Corner was non-existent.” In response, the prosecutor reminded the court that the taskforce agent had neither had nor used the ALERT transaction data; he had only monthly summaries of that data, which the prosecutor had provided to the defense prior to trial. And, when the agent had used raw transaction data from Conduent (the state equivalent database) in preparation for trial, the prosecutor had given that data to the defense.6 More importantly, the prosecutor argued, the data was not relevant to the defendants’ guilt. At trial, the prosecutor proved his case based on the videos and testimony about the controlled buys, testimony by co-conspirators (i.e., Randall, the “runner,” and the cardholder with the tab), testimony by the former employee, testimony by BCC’s external accountant, and testimony by the investigators and SNAP administrators. The monthly sales comparison did not prove guilt, it merely estimated the amount of the loss attributable to the overall fraud scheme.

The district court held that the ALERT transaction data was not “material” because it was not exculpatory: it did not concern, much less disprove, the prosecution’s charge that the Busches conspired to defraud the United States with respect to eleven specific transactions. United States v. Busch, No. 3:18-cr-00079, 2020 WL 1703191, at *5 (S.D. Ohio, Apr. 8, 2020). Nor was the transaction data impeaching: it did not contradict or call into question any testimony as to that

6 This Conduent data set, which the prosecutor provided to the defense prior to trial, contained the raw transaction data for the 195,113 EBT transactions at BCC. The only articulated difference between the Conduent data provided before trial and the ALERT data provided after trial was that the ALERT data also had data for EBT transactions at the comparator stores, whereas the Conduent data set had data for EBT transactions at only BCC.

Case Nos. 20-4065/4066/4067, United States v. Busch, et al. charge. Id. at *6. Because the data was not material, there was no prejudice, no question of confidence in the jury’s verdict, and no Brady violation. Id. The court denied the motion.

At the ensuing sentencing hearing, the court heard argument about the appropriate means of estimating loss for purposes of sentencing and rejected both parties’ approaches, explaining:

As stated, the government has the burden to prove the amount of the loss by a preponderance of the evidence. And in situations where . . . the losses occasioned by financial frauds are not easy to quantify, the district court need only make a reasonable estimate of the loss given the available information. ... The [c]ourt finds that none of the amounts set forth by the parties is sufficiently reasonable in estimating a loss in this case. [The Busches’] expert report provides a number of reasons why the government’s loss methodology and calculation is flawed and not reasonable. The [c]ourt agrees with some of those reasons. The [c]ourt also notes that the government’s calculation did come from a comparative analysis that was used to obtain a search warrant. However, the [Busches’] loss calculation of $14,123.67, which is merely what it claims to be the amount of the actual fraud actually proven to have taken place, the [c]ourt finds is also unreasonable. The $14,123.67 is only the . . . total amount from the transactions performed by the confidential informant who testified at trial. The [c]ourt does find . . . that amount to be unreasonable in light of the testimony and the evidence. And the [Busches’] expert[’]s suggested [] loss calculation only adds $1,303 from the Monday transactions to the $14,123.67 and makes some assumptions that the [c]ourt finds also are not reasonable.

Having rejected those competing approaches, the court began its own approach by finding 2,700 to 3,600 fraudulent EBT transactions, based on Randall’s trial testimony of 30 to 40 per month. The court calculated a range of $74,736 to $99,648 in total fraud, based on an overall average EBT charge of $27.68 per transaction. And the court deemed the middle of that range to be the appropriate estimate of loss, making it $87,192.

The court set Amanda’s total offense level at 14, her criminal history as category I, and her guidelines’ advisory range at 15 to 21 months in prison. But it sentenced her to just five years of probation with the first eight months under home confinement. The court set Michael’s total *12 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. offense level at 17, his criminal history as category I, and his advisory range at 24 to 30 months. It sentenced him to just 12 months in prison, imposed a fine of $50,000, and ordered forfeiture of $87,192 in currency seized from his home. The court also ordered restitution in the amount of $87,192, applied to Michael, Amanda, and BCC jointly and severally.

On appeal, the Busches allege three errors that would warrant reversal: a Brady violation, an incorrect jury instruction, and prosecutorial misconduct. None of these has merit.

II. Brady

The Busches claim that the prosecutor violated Brady v. Maryland by failing to provide them the full array of raw transaction data from the ALERT database.

To prove a Brady violation, a defendant must show that the undisclosed evidence is both favorable to the defendant—meaning either exculpatory or impeaching—and material either to guilt or to punishment. Evidence is material when there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different. And a reasonable probability is a probability sufficient to undermine confidence in the outcome of the proceeding.

United States v. Lucas, 2021 WL 4099241, at *8 (6th Cir. Sept. 9, 2021) (quotation marks and citations omitted); see United States v. Bagley, 473 U.S. 667, 682 (1985); Strickler v. Greene, 527 U.S. 263, 281 (1999); Kyles v. Whitley, 514 U.S. 419, 433-34 (1995); Brady, 373 U.S. at 87.

The district court ruled that the ALERT transaction data “was not ‘material’ when viewed in the context of the evidence presented at trial and the crimes for which the [Busches] were charged (and convicted).” Busch, 2020 WL 1703191, at *5. The court explained that this data does not show whether “any particular transaction is fraudulent or is not fraudulent.” Id. At most, it enabled the Busches’ expert to construct a statistical analysis from which the Busches’ attorney could: (1) argue “that there was not nearly as much food stamp fraud as the Government believed,” and (2) “impeach [certain] witnesses’ estimates of how much food stamp fraud was occurring at BCC.” Id. at *6. But the prosecutor “did not need to establish, and the jury did not need to find, *13 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. widespread food stamp fraud occurring at BCC in order for the [Busches] to be convicted of the counts on which they were found guilty,” namely “conspiracy to defraud the United States, and . . . fraud with respect to eleven specific transactions.” Id. at *5 (quotation marks and underlining omitted; emphasis added). The court found that the evidence at trial “provides more than sufficient confidence in the jury’s verdict, even in light of the transactional data at issue and the [Busches’] expert’s opinion of what such data shows.” Id. at *6. Therefore, the court concluded that the ALERT transactional data was “not Brady material,” nor had the Busches “shown that prejudice must have ensued” from the presumptive suppression of that data. Id. (quotation marks omitted).

The Busches fail to address, much less overcome, the district court’s reasoning. For example, they say the ALERT transactional data was favorable to them because it would have “allowed [them] to show by statistical analysis that ‘widescale’ fraud was not occurring at BCC— the core of the government’s case against them.” But that was not the “core” of the case against them. In fact, the district court’s premise was that widespread fraud was not even part of the charges against them. Thus, favorable though it might be in a general sense, this was immaterial to proving or disproving the charges against them.

Nevertheless, in their briefing, they press three reasons why the ALERT data was material. None holds up under examination. First, they contend that the data “proved [their] theory at trial that the fraud at BCC was limited and solely perpetrated by the government’s cooperating witness Randy Busch without the knowledge or involvement of Michael or Amanda Busch.” But it did no such thing. As the district court pointed out, the data itself does not show whether any transaction is fraudulent or legitimate. At most, this data enabled the Busches’ expert to render an opinion, based on his unique manipulation of the data, his personal knowledge or experience, and certain partisan assumptions, some of which, the district court opined, were “not reasonable.” That is perhaps the most critical point. The expert did not prove that Randall alone conducted the *14 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. fraudulent transactions unknown to Michael and Amanda; he assumed that and extrapolated from it. He started with that assumption (that Randall alone was conducting fraudulent EBT transactions and hiding them from Michael) and applied it to two known facts—(1) the BCC shop is so small that anyone present would witness the fraudulent EBT transactions and (2) Randall typically worked alone at BCC on Mondays when it was closed to the public and, more importantly, Michael was absent—to conclude that Randall was conducting fraudulent EBT transactions on Mondays, unknown to Michael. Beyond being circular, this was contrary to the irrefutable video evidence in which none of the eleven controlled buys occurred on a Monday.7 All eleven occurred during ordinary business hours with other employees present, including Michael who twice participated. And Randall made little effort to hide it. Recall also that the former employee testified to witnessing this activity routinely.

Second, the Busches contend that the data “revealed that the government’s statistical method of determining that there was allegedly fraud committed by others than Randy was flawed.” The prosecution used its “statistical method” to estimate the total amount of loss that it deemed attributable to BCC’s SNAP fraud. It did not use that statistical method to prove that Michael, rather than Randall, committed the fraud. The prosecutor presented the videos of the controlled buys and testimony from the informant, Randall, the runner, and the former employee to prove that Michael participated in fraudulent EBT transactions.

And third, the Busches contend that the data was material because it “impeached the testimony of the government’s most important witnesses.” These “most important witnesses” were, in the Busches’ view, the runner and the agent who testified about the taskforce’s statistical comparison. Even if the expert’s opinion did impeach their testimony, those two were arguably

7 The eleven controlled buys occurred on Tuesdays (3), Wednesday (1), Thursdays (6), and Friday (1).

Case Nos. 20-4065/4066/4067, United States v. Busch, et al. the prosecution’s least important witnesses; they were certainly not the most important, nor would impeaching their testimony have created a reasonable probability of a different outcome. The prosecution’s critical witnesses were the informant who authenticated, narrated, and discussed the videos of the controlled buys, followed closely by Randall and the employee who witnessed Randall and Michael committing the fraudulent EBT transactions on numerous occasions. The expert opinion did not impeach their testimony. It did not even implicate their testimony in any way.

In their appeal brief, the Busches made certain other contentions that are mistaken or outright false. They say their “expert analysis caused the district court to reduce its calculation of loss from $3,450,960 . . . to only $87,192.” The district court never accepted the prosecutor’s tendered amount of $3.5 million, nor did the court ever calculate $3.5 million for itself, so the court did not reduce “its” calculation of loss. The court did expressly reject the prosecutor’s $3.5 million number and even “agree[d] with some of” the expert’s “reasons why the government’s loss methodology and calculation is flawed and not reasonable,” but the court also stated expressly that the Busches’ expert “makes some assumptions that . . . are not reasonable.” It is doubtful that the court relied on the expert in the way the Busches imply, and it is certain that the expert’s report did not cause the final $87,192 number.

The Bushes also say that “the only proof that the government introduced of SNAP fraud at trial was ten fraudulent SNAP transactions between a government confidential informant and Randy Busch.” As we have stated repeatedly, the prosecutor produced video of and testimony about the eleven controlled buys (20 EBT card transactions), testimony from Randall who admitted to the fraud and explained that Michael had trained him to do it, testimony from the outside accountant who questioned how profits were up when sales were down, testimony from a nearby *16 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. business owner about the EBT fraud, and testimony from a former employee who witnessed Michael and Randall committing EBT fraud on numerous occasions.

Another odd, apparently confused, contention is that: “Had the government produced the transaction data before trial, [the Busches] would have called [the expert] at trial to give the exact testimony he gave at the post-trial hearing.” But the Busches hired the expert after trial and then he requested the ALERT transaction data after he was hired. That is, they hired the expert even though they did not yet know about the ALERT transaction data. They could have hired him before trial in the same way they hired him after trial. The fact that the Bushes did not have the ALERT data had no bearing on their hiring of the expert.

And, finally, they offer this misleading contention: “Still further, [the expert]’s testimony suggested that ‘there were confidential informant transactions that were attempted but were not successfully executed; in other words, the people at BCC declined to do fraudulent transactions.’” The expert said this while explaining a caveat that he had included in his report, in which he had warned of the possibility of one unsuccessful controlled buy that, if true, would require him to change his analysis. But upon further questioning, the expert conceded that his suspicion of this unsuccessful controlled buy had been refuted and he agreed that it had not occurred.

On the whole, the Bushes have not shown that the ALERT transactional data was necessarily favorable to them—inasmuch as it was neither exculpatory nor impeaching—or material to their guilt of the crimes charged. There is no reasonable probability that the result of the trial would have been different if the data had been provided. See Bagley, 473 U.S. at 682. They failed to prove a Brady violation.

III. Jury Instruction

The Busches claim the district court committed reversible error by instructing the jury to convict them if it found that they had violated civil or administrative regulations, and further erred *17 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. by refusing to instruct the jury to the contrary. We review de novo a claim that jury instructions misstate the law. Hurt v. Com. Energy, Inc., 973 F.3d 509, 523 (6th Cir. 2020).

The prosecutor charged Michael with multiple counts of SNAP fraud under 7 U.S.C. § 2024(b)(1), which says, in pertinent part: “whoever knowingly uses, transfers, acquires, alters, or possesses benefits in any manner contrary to this chapter or the regulations issued pursuant to this chapter shall . . . be guilty of a felony.” Under those regulations, SNAP-authorized retailers may not accept EBT payment “in exchange for cash,” 7 C.F.R. § 278.2(a), or “in payment for items sold to a household on credit,” § 278.2(f). Restating this law as a single, simplified proposition applicable to the prosecutor’s accusations in this case: it would be a felony for Michael (acting as the SNAP-authorized retailer) to knowingly trade cash or provide goods on credit in exchange for EBT payments. The court instructed the jury on those counts as follows:

For you to find the defendant guilty of SNAP fraud, you must find that the government has proved each and every one of the following elements beyond a reasonable doubt: First, that the defendant acquired, possessed, used, or trafficked SNAP benefits in a manner not authorized by law or the Department of Agriculture regulations; Second, that the defendant knew that he was acting unlawfully and intended to violate the law; Third, that the benefits in question had a value of at least $100. Under the law, the Department of Agriculture regulations are that the only authorized use of SNAP benefits is the recipient’s purchase of qualifying food at the authorized retail store at a price prevailing in such stores. No law or Department of Agriculture regulation allows anyone to sell or purchase benefits for cash or other non-food items of value. Further, no law or Department of Agriculture regulation allows anyone to extend credit to an individual with the promise of payment in future benefits. The government need not show that the defendant had knowledge of the specific law, only that he knew that his conduct was unlawful.

The Busches contend that the reference to “regulations in th[is] criminal jury instruction[] had the practical effect of rewriting the criminal law of SNAP fraud as enacted by Congress and *18 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. inviting jurors to convict [them] for violating administrative regulations, not committing crimes.” But the SNAP-fraud statute enacted by Congress, 7 U.S.C. § 2024(b)(1), expressly states that it is crime to knowingly violate “the regulations issued pursuant to this chapter.” We reject this contention out of hand.

It appears from a careful parsing of their brief, however, that the Busches’ actual contention is that, because “a regulatory violation does not require specific intent while [a] criminal offense does,” “the instructions were confusing and misleading about the most important issue for the jury to decide—[their] intent.” That is, they accuse the district court of instructing the jury to convict them of SNAP fraud even if it found that they did not specifically intend to violate the law, i.e., 7 U.S.C. § 2024(b)(1). That would be of concern if it were true, but it is not. In fact, it is impossible to reconcile this accusation with the specific-intent aspect of the given instruction, which is: “to find the defendant guilty of SNAP fraud, you must find . . . beyond a reasonable doubt . . . that the defendant knew that he was acting unlawfully and intended to violate the law.” There is no merit to this contention.

Finally, the Busches contend that a further “limiting instruction was necessary because the government was permitted to introduce evidence at trial in a way that conflated regulatory infraction with crime to the jury.” This is the same meritless argument in a different wrapper. Given that 7 U.S.C. § 2024(b)(1) specifically sets out one element of SNAP fraud as a violation of the SNAP regulations, the prosecutor had to prove that Michael violated the regulations. But, as the specific-intent aspect of the instruction made clear, that violation alone was not enough; the jury had to find that Michael specifically intended to act unlawfully. The instruction given was correct and sufficient. We cannot agree that the jury was confused or misled, or that it convicted Michael improperly. A further limiting instruction was not necessary. *19 Case Nos. 20-4065/4066/4067, United States v. Busch, et al.

IV. Prosecutorial Misconduct

The Busches moved the district court for a new trial based on their claims of prosecutorial misconduct. The district court denied the motion. They repeat those claims here.

Whether a prosecutor’s actions or statements at trial amount to misconduct raises “mixed questions of law and fact, which we review de novo.” United States v. Carson, 560 F.3d 566, 574 (6th Cir. 2009). We use a two-step approach, in which we first decide whether the challenged conduct was improper. United States v. Warshak, 631 F.3d 266, 302 (6th Cir. 2010). “[I]t is not enough that the prosecutors’ remarks were undesirable or even universally condemned. The relevant question is whether the prosecutors’ comments so infected the trial with unfairness as to make the resulting conviction a denial of due process.” Darden v. Wainwright, 477 U.S. 168, 181 (1986) (quotation marks and citation omitted). Thus, we consider the prosecutor’s conduct within the context of the entire trial. United States v. Francis, 170 F.3d 546, 552 (6th Cir. 1999).

If we conclude that the conduct was improper, we must then determine whether it was so flagrant as to warrant reversal. Warshak, 631 F.3d at 302. For this, we consider whether the prosecutor’s conduct or comments were (1) misleading to the jury or prejudicial to the defendant, (2) extensive or isolated, and (3) deliberate or accidental. Id. And we consider (4) whether the evidence against the defendant was weak or strong. Id.

A. During Trial

The Busches claim that the prosecutor committed misconduct during trial by violating certain of the court’s pre-trial evidentiary rulings when he solicited evidence about BCC’s cash-management practices. This included (1) questioning Amanda about how she recorded cash for bookkeeping purposes; (2) introducing the year-over-year accounting reports with the handwritten notes asking how profits were up when sales were down; (3) questioning the outside accountant *20 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. about Amanda’s cash-management practices, her recordkeeping, and the $191,135 cash hoard; and (4) questioning former BCC employees about their having been paid in cash.

The district court rejected this claim, explaining that its pretrial ruling in limine prohibited the prosecutor from introducing evidence that Amanda “failed to accurately report revenue and income generated from cash sales at BCC to state and federal taxing authorities.” United States v. Busch, No. 3:18-cr-079, 2019 WL 6682157, at *2 (S.D. Ohio, Dec. 6, 2019) (editorial marks omitted). The prosecutor’s questioning and evidence about general cash management did not implicate any tax evasion or tax fraud, and it was relevant to prove Amanda’s knowledge of the SNAP-fraud because that cash flow, questionable cash recording, and overall cash management were central to the prosecutor’s theory of the SNAP-fraud scheme. The court concluded that the evidence had probative value that was not substantially outweighed by the dangers identified in F.R.E. 403. Therefore, the court denied the motion for acquittal or a new trial. Id.

The district court also rejected the Busches’ claim that the jury could not have followed the court’s instructions to disregard certain testimony or evidence. Courts presume that jurors will follow an instruction to disregard inadmissible evidence, unless there is an “overwhelming probability” that they would be unable to do so, Richardson, v. Marsh, 481 U.S. 200, 208 (1987), and a strong likelihood that the evidence would be “devastating” to the defendant, Bruton v. United States, 391 U.S. 123, 136 (1968). The district court found no reason to believe that this jury was incapable of obeying its curative instructions. Busch, 2019 WL 6682157, at *3.

On appeal, the Busches complain that the prosecutor should not have questioned witnesses or proffered evidence about Amanda’s cash management, criticize the court for allowing such questioning and admitting such evidence (e.g., the year-over-year accounting reports with the handwritten notes, testimony about the $191,135 cash hoard, and former employees’ testimony that they had been paid in cash), and insist that this was irreparably prejudicial. But the Busches *21 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. offer nothing to overcome the district court’s explanation or ruling. Moreover, they have offered nothing that would show that the prosecutor’s conduct was even beyond the bounds of normal adversarial advocacy, much less improper—particularly given that the court overruled the defense objections at trial and admitted the evidence about which the Busches now complain.

B. During Closing Argument

The Busches claim that the prosecutor committed misconduct during his closing argument in four ways: by (1) vouching for Randall’s credibility; (2) arguing that Randall’s guilt, alone, was enough to prove their guilt; (3) mischaracterizing the evidence; and (4) misstating the law.

1. Vouching

The Busches claim that the prosecutor committed misconduct during his closing argument by vouching for Randall’s credibility as a witness by stating that Randall was “not a liar” and “had no motive to lie” given that he was bound by his plea agreement to testify truthfully.

The district court rejected this claim, explaining that, “[a]s a general matter, it is not improper for the prosecutor to refer to the plea agreements of cooperating witnesses in the expectation that their credibility will be at issue.” Busch, 2019 WL 6682157, at *3 (quotation and editorial marks omitted) (quoting United States v. Reid, 625 F.3d 977, 983 (6th Cir. 2010)). The court also noted that the defense had not objected to this part of the prosecutor’s closing. Id.

On appeal, the Busches fail to acknowledge, much less overcome, the district court’s findings, the foremost being that their counsel did not object to this perceived vouching at trial. The Busches therefore did not preserve this claim for appeal and our review is for plain error. See United States v. Hall, 979 F.3d 1107, 1119 (6th Cir. 2020). “Plain error is (1) error (2) that was obvious or clear, (3) that affected the defendant’s substantial rights and (4) that affected the fairness, integrity, or public reputation of the judicial proceedings.” Id. (quotation and editorial marks omitted). “We will reverse a conviction based on plain error only in exceptional *22 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. circumstances that seriously affect the fairness of the trial.” United States v. Sills, 662 F.3d 415, 417 (6th Cir. 2011).

“Improper vouching occurs when a prosecutor either (1) bluntly states a personal belief in a witness’s credibility, thereby placing the prestige of the office of the United States Attorney behind that witness, or (2) implies that the witness’s testimony is corroborated by evidence known to the government but not known to the jury.” United States v. Henry, 545 F.3d 367, 378-79 (6th Cir. 2008) (citation omitted). In this case, the prosecutor did neither of these two things, but the Busches point to three passages from his closing argument that, they contend, imply the same:

We learned that when Randall Busch came here and testified in this witness box in front of you he had already pled guilty to being engaged in SNAP fraud at the Busch’s Country store. The same case, same facts and circumstances as his brother Michael and his sister-in-law Amanda. He, in fact, had even been sentenced by this Court. So with those facts, irrefutable facts in the open, you have to ask yourself, what on earth motive to lie does Michael -- does Randall Busch have to testify against his older brother, an older brother that he idolized, and his sister-in-law? Does the government have some hammer over his head? That unless he testifies the way we want him to, that somehow we are going to take retribution against him, somehow the sentence is going to be affected? That horse has left the barn. He has no motive to lie. In fact, out of all the witnesses, he had the most motivation to not cooperate, but he did. And you have to yourself judge his credibility. His mannerisms. His tone of voice. His body language. His answers.

The second passage is:

Randall Busch accepted responsibility. He pled guilty. He’s been sentenced. His involvement in this overall SNAP fraud scheme is over. Was he a conspirator with his brother and his sister? Absolutely. I urge you to pay particular attention to the February 27, 2018, transaction which Randall, Mike, and Amanda all have their tentacles in. If we introduced nothing else other than the facts and circumstances of that undercover transaction, that’s enough, that’s sufficient to meet the elements that indeed Randall, Michael, and Amanda had entered into a criminal partnership, an understanding as to what’s going on.

Case Nos. 20-4065/4066/4067, United States v. Busch, et al. The third passage is:

Defense witnesses. With the exception of Ms. Rosfeld, they all had the same last name. Ms. Rosfeld’s obviously married. Her maiden name obviously is Busch. What were the common threads? What was the common theme? All were told not to conduct cash-for-EBT transactions. That’s fine. All knew about pocket, but only Tyler collected pocket. All testified about how the business grew the past eight years. All testified about acquiring some -- acquiring more counter space. I guess it was a 36-foot counter at the market. And all felt betrayed [by Randall]. We talked about that with [Michael] Busch today. When you really think about that term ‘betrayed,’ what are we talking about? Was it betrayed that he got caught? Was it betrayed that he embarrassed the family? I’m sure it was all that. Was it betrayed that he stole? Yes. But no one -- not his mother, not his sister, not his niece, not his nephew, not even his sister-in-law -- said [Randall] lied. There is a difference between being a thief and a liar. Randy is a thief. He’s not a liar. And there is no evidence that Randy stole from the business. He didn’t do it. There is no evidence.

The Busches contend that the prosecutor vouched for Randall’s credibility as a witness by saying that Randall had “no motive to lie,” concluding that he “is a thief . . . not a liar,” and insinuating that his guilty plea ensured his truthful testimony. But, considering these statements in context, the prosecutor merely urged the jurors—when making their own assessment of Randall’s credibility—to draw certain inferences from the evidence produced at trial. That is not improper vouching. Moreover, considered in context, those statements were not obviously or clearly improper, which likely explains why defense counsel did not object at the time. Nor did they affect the outcome of this trial. This is not an “exceptional circumstance[] that seriously affect[ed] the fairness of the trial.” See Sills, 662 F.3d at 417. We find no plain error.

2. Randall’s Guilty Plea

The Busches claim that the prosecutor engaged in misconduct during his closing argument by referring to Randall’s guilty plea as substantive evidence of their guilt. The Busches refer again to the first of the three passages above, in which the opening paragraph says:

We learned that when Randall Busch came here and testified in this witness box in front of you he had already pled guilty to being engaged in SNAP fraud at [BCC].

Case Nos. 20-4065/4066/4067, United States v. Busch, et al.

The same case, same facts and circumstances as his brother Michael and his sister- in-law Amanda. He, in fact, had even been sentenced by this Court.

In the preceding section, the Busches argued that this passage was improper because it supported the inference that Randall had no motive to lie during his testimony, given that he had already been convicted and sentenced, and was no longer facing any further criminal proceedings. In this section, the Busches argue that this passage was improper because it was a surreptitious means of urging the jury to convict Michael and Amanda on an improper basis, namely that Randall is guilty, so Michael and Amanda must be guilty too.

The district court rejected this claim, as it had the prior claim, explaining that it is generally permissible for the prosecutor to refer to a witness’s plea agreement, and noted that defense counsel had not objected to this part of the prosecutor’s closing. Busch, 2019 WL 6682157, at *3. Therefore, our review is again for plain error. See Hall, 979 F.3d at 1119.

Two oddities about this argument stand out. One, this was a conspiracy case—a conspiracy comprising Randall, Michael, Amanda, and BCC. Randall’s guilt was effectively an element of the conspiracy charge, meaning the prosecutor had to demonstrate Randall’s guilt to the jury. And two, the defense theory was that Randall committed the fraudulent EBT transactions unbeknownst to Michael and Amanda. The entirety of the Busches’ defense was predicated on emphasizing Randall’s guilt. The prosecutor’s reminding the jury that Randall had pled guilty was no worse than the Busches’ counsel telling the jury that Randall was guilty; neither urged the jury to convict Michael and Amanda based solely on Randall’s guilty plea. This was not plain error.

3. Mischaracterizing Evidence

The Busches claim that the prosecutor fabricated and mischaracterized evidence during his closing argument. The Busches point to three specific instances. The first concerns the prosecutor’s reference to: “The second set of books. The mystery set of books. The set of books *25 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. that [the outside accountant] repeatedly asked Amanda to produce.” The Busches contend that this assertion improperly “referenced ‘facts’ not in evidence,” because there was no evidence about any second set of books. In response to the Busches’ post-trial motion raising this claim in the district court, the prosecutor agreed that there was no “second set of books” and explained the statement this way:

The reference to the so-called ‘second set of books’ was nothing more than classic white collar metaphor referencing the glaring absence of any accurate books or records produced or maintained by Amanda Jo concerning [BCC’s] cash transactions. This subject was discussed during Amanda Jo’s testimony and [the outside accountant]’s testimony. A close examination of the argument reveals that the Government specifically indicated no such ‘second set of books’ in reality ever existed.

The district court concluded that “[t]roublesome as this might be, such improprieties are corrected by a court’s instructions to the jury that the attorney’s arguments are not evidence.” Busch, 2019 WL 6682157, at *3.

The Busches do not contest the prosecutor’s explanation, which shows that, even if this statement were improper (as the district court suggested), it did not rise to the level of prosecutorial misconduct because it was not flagrant: it was not necessarily misleading to the jury, particularly when considered in context; it was a passing turn of phrase, neither extensive nor deliberate; and the overall evidence against the Busches was strong. See Warshak, 631 F.3d at 302.

The Busches’ second claim of mischaracterized evidence is the prosecutor’s statement that Amanda “acknowledged knowing the SNAP rules and regulations,” whereas the actual testimony was that “she was aware of the rules and regulations,” and “the difference between knowledge and awareness is the difference between guilt and innocence.” Because defense counsel did not object to this at trial, our review is for plain error, Hall, 979 F.3d at 1119, and this comes nowhere close to plain error. Nor would this satisfy the flagrancy step of prosecutorial misconduct. See Warshak, 631 F.3d at 302. The prosecutor produced substantial evidence that Amanda knew the SNAP rules *26 Case Nos. 20-4065/4066/4067, United States v. Busch, et al. and regulations. She testified to having signed the USDA forms for the SNAP-retailer license and several reauthorizations, in which she attested that she knew, understood, and would comply with the rules and regulations. This contention is frivolous.

The Busches’ third claim of mischaracterized evidence is the prosecutor’s reference to Amanda’s utterance about emptying a safety deposit box of cash. The prosecutor said:

The search warrants were executed on that fateful day, May 10, 2018. We know that the search warrants were conducted at the store, at Randy’s house, [and] at Michael and Amanda’s house. . . . Correspondingly, $190,000 in excessive cash was found in Michael and Amanda’s house. What do we know? As far Michael and Amanda’s only source of income was the store. What do we also know? That Amanda, while [speaking with a taskforce agent] out at the house, indicated [that] she just returned from cleaning out a safe deposit box of cash. She had cash all over the place.

Defense counsel objected and the court sustained, saying “That will be stricken. Ladies and gentlemen, you are instructed to disregard that.” Defense counsel did not specify that he was directing his objection to the safe-deposit-box statement and the court did not specify exactly what the jurors were to disregard. Moments later, the prosecutor said:

Amanda’s interview by [a taskforce agent]. She acknowledged knowing the SNAP rules and regulations. She was able to accurately cite the percent of overall store revenue generated by EBT sales. And she admitted to having just removed cash from a bank safety security box. That’s what she told him.

There was no objection to this statement and our review is for plain error. Hall, 979 F.3d at 1119. This does not rise to the level of plain error. This would not even satisfy the flagrancy step of prosecutorial misconduct. See Warshak, 631 F.3d at 302.

4. Misstating the Law

The Busches claim the prosecutor engaged in misconduct during his closing argument by misstating the law in a way that negated an essential element of the charged offense. The prosecutor addressed the illegality of SNAP transactions for credit as follows: *27 Case Nos. 20-4065/4066/4067, United States v. Busch, et al.

And first and foremost, the SNAP laws and regulations prohibit the exchange of EBT benefits for cash, or extending credit or loans. Couldn’t be clearer. The government is not in the business of floating loans. And the fact that certain grocers, certain retailers, certain vendors for whatever reason take it upon themselves that they are going to be the good samaritans or the Robin Hoods of society is not a defense to breaking the law.

The Busches argue that “a ‘good Samaritan’ by definition lacks specific intent to defraud . . . [or] violate the law,” which are “both elements of the underlying offenses charged.” Thus, they claim this was reversible error.

Because defense counsel did not raise this objection at trial, our review is for plain error, Hall, 979 F.3d at 1119, and this comes nowhere close to plain error. This would not even satisfy the flagrancy step of prosecutorial misconduct. See Warshak, 631 F.3d at 302. Perhaps most importantly, the judge instructs the jurors on the law. One of those instructions was that the judge, not the attorneys, say what the law is. Another was that the jury had to find beyond a reasonable doubt that Michael and Amanda had knowingly and intentionally violated the law. And a third was that “good faith” “is a complete defense to all charges . . . because good faith on the part of the defendant is simply inconsistent with the intent to defraud.”

While the term ‘good faith’ has no precise definition, it encompasses, among other things, a belief or opinion honestly held in the absence of malice or ill will, and an intention to avoid taking unfair advantage of another. The burden of proving good faith does not rest with the defendant because the defendant does not have any obligation to prove anything in this case. It is the government’s burden to prove to you beyond a reasonable doubt that the defendant acted with an intent to defraud. If the evidence in this case leaves with you a reasonable doubt as to whether the defendant acted with an intent to defraud or in good faith on any of the charges, you must acquit the defendant of those charges.

Contrary to the Busches’ characterization of it, the prosecutor’s statement was consistent with this instruction: a good faith mistake is excusable; a knowing and intentional violation of the law is not excusable just based on a belief that a Good Samaritan would do so. *28 Case Nos. 20-4065/4066/4067, United States v. Busch, et al.

The Busches have not proven that prosecutorial misconduct occurred at their trial.

V.

We AFFIRM the judgment of the district court.