Summary
affirming when, based on the evidence presented at trial, the court was convinced, beyond a reasonable doubt, that the government's misconduct could not have reasonably affected the jury's verdict
Summary of this case from United States v. PattersonOpinion
No. 18-2896
02-06-2020
UNITED STATES of America Plaintiff - Appellee v. Asmerom KELETA, also known as Ace Defendant - Appellant
Counsel who presented argument on behalf of the appellant and appeared on the brief was Justin Gelfand, of Saint Louis, MO. Counsel who presented argument on behalf of the appellee and appeared on the brief was Dianna Collins, AUSA, of Saint Louis, MO.
Counsel who presented argument on behalf of the appellant and appeared on the brief was Justin Gelfand, of Saint Louis, MO.
Counsel who presented argument on behalf of the appellee and appeared on the brief was Dianna Collins, AUSA, of Saint Louis, MO.
Before KELLY, MELLOY, and STRAS, Circuit Judges.
KELLY, Circuit Judge.
Asmerom Keleta was convicted by a jury of conspiring to defraud the United States and willfully aiding and assisting in the filing of a false tax return. He appeals the district court’s denial of his motion to suppress, its rejection of his challenges to improper statements made by the government, and its application of a four-level role enhancement to his base offense level at sentencing. After careful review, we affirm Keleta’s convictions but vacate his sentence and remand for resentencing.
I. Investigation
Asmerom "Ace" Keleta owned Eriace Enterprise, LLC (Eriace), which operated several tax-preparation businesses in the St. Louis metropolitan area under the names U-City Tax Service and Ace Express Tax Service. In 2012, the IRS’s Scheme Development Center (SDC) forwarded information about Eriace to the IRS’s Criminal Investigation Division. After reviewing the information, investigators noted that a high percentage of tax returns prepared by Eriace claimed certain tax credits. They also found that much of the information used to seek these tax credits "was not verifiable by other information filed with the IRS" and that the unverifiable information was often combined with verifiable information in ways that made the taxpayer eligible for the maximum or near the maximum available tax credit. The personal tax returns for Keleta and U-City Tax Service employees Miyoshi Lewis and Teklom "Tek" Paulos fit this pattern.
On February 27, 2013, IRS Special Agent Danette Coleman conducted an undercover operation at a U-City Tax Service branch in University City, Missouri. Lewis prepared a tax return for Coleman while Paulos helped another customer. Keleta was not present at the time. Lewis initially calculated a refund amount of $44 based on the information Coleman provided. When Coleman asked why the refund was so low, Lewis responded that she could "make it more, but the fee will go up." Lewis then entered false information and calculated a refund of approximately $4,200. She charged Coleman $500 cash for obtaining this increased refund.
The IRS also received three anonymous letters alleging tax fraud at that U-City Tax Service branch. The anonymous informant claimed that Keleta had sold the use of his preparer tax identification number (PTIN) and electronic filing identification number (EFIN) to several individuals, including Paulos, who used them to file tax returns containing fraudulent information. The IRS corroborated that Lewis, Paulos, and several other individuals named in the letters were "friends" on Facebook. The IRS also found that numerous withdrawals from Eriace’s business checking account appeared to be personal in nature.
Based on this information, IRS Special Agent Nicholas Kenney obtained a warrant to search the U-City Tax Service branch and seize records found on the premises. The government executed the warrant on April 13, 2013. It seized computers, cell phones, client files, and other items, including a signature stamp with Keleta’s signature.
II. Trial
Keleta, Lewis, and Paulos were charged with conspiracy and tax-fraud crimes in a superseding indictment. Lewis and Paulos pleaded guilty. Keleta pleaded not guilty and went to trial on three counts: (1) conspiring to defraud the United States in violation of 18 U.S.C. § 371 ; (2) willfully aiding and assisting in the false and fraudulent preparation of C.M.’s 2011 tax return in violation of 26 U.S.C. § 7206(2) ; and (3) willfully aiding and assisting in the false and fraudulent preparation of K.H.’s 2011 tax return in violation of 26 U.S.C. § 7206(2).
The government called seven U-City Tax Service customers at trial, five of whom testified that Keleta reported false information on their tax returns. Lewis also testified at trial. Paulos did not. Before the jury was sworn, the district court granted the parties’ joint motion to exclude Paulos from the courtroom. The parties agree that Paulos was only in the courtroom for "a matter of minutes" during jury selection before he was excluded and that there were approximately 40 potential jurors in the jury box and gallery at the time.
Lewis testified that she had known Keleta since 2009 and began working for him during the 2011 tax year. She identified Keleta in court and stated that he was at the U-City Tax Service branch "a hundred percent of the time" during the 2011 tax year. She testified that she did not have any tax-preparation experience when she began working at U-City Tax Service and that Keleta helped her prepare tax returns in her first few weeks on the job. During this time, he trained her to use false information to obtain larger tax returns while avoiding detection by the IRS. He also reviewed and transmitted the tax returns she prepared during the 2011 tax year. Additionally, Lewis testified that Keleta prepared her personal 2011 tax return, which included false information.
Lewis stated that, at one point, Keleta advised her on how much to charge for obtaining fraudulently increased tax refunds. She explained that, to collect this fee, the customers’ tax refunds would be mailed to U-City Tax Service. When a check arrived, Paulos would walk the customer to a currency exchange across the street. The customer would cash the check and pay the fee out of their refund.
Lewis testified that Keleta was "in the office less the second year" and that Paulos was in charge of reviewing and submitting tax returns for the 2012 tax year. She stated that Keleta purchased the signature stamp for her and Paulos to use while he was out of the office, but she never used it without Keleta’s knowledge. She also stated that Keleta and Paulos looked alike, that customers would sometimes confuse them, and that Paulos did not correct customers when they referred to him as Keleta. The testimony of the government’s first taxpayer witness, C.M., was consistent with the scheme Lewis outlined. C.M. stated that she went to have her taxes prepared at U-City Tax Service because a friend told her they would give her a "hook-up." She identified Keleta in a photographic lineup at trial as "Ace," the person who prepared her 2011 tax return. She testified that Keleta originally calculated a $40 refund. When she said that "wasn’t enough," Keleta falsely reported additional income and calculated a larger refund. C.M. explained that "we both kn[e]w we were doing it illegally." She asked Keleta whether she needed to falsify additional documents to facilitate the fraud but he said that was unnecessary. When her return arrived, she went to the currency exchange with Paulos, cashed the check, and paid $400 for the increased refund. C.M. testified that Paulos prepared her 2012 tax return using false information that was similar to the information Keleta had used the year before.
Keleta impeached Lewis’s testimony by noting that she hoped to obtain a substantially reduced sentence by testifying, had minimized her role at U-City Tax Service during interviews with investigators, and was formerly in a personal relationship with Paulos.
The remaining U-City Tax Service customers also testified that Lewis, Paulos, and Keleta reported false information on their tax returns. But, unlike C.M., none of them indicated that they were aware that false information was being used to increase their tax refunds. And each of them testified that they did not provide Keleta with the false information that appeared on their tax returns.
Keleta’s primary defense was misidentification. He relied on documentary evidence showing that the signatures on several of the customers’ tax returns did not match his signature stamp or signature cards and deposited checks associated with his bank accounts. He anticipated Lewis’s testimony that customers would sometimes confuse him and Paulos in his opening statement and drew out on cross-examination that he and Paulos looked alike. And he attempted to cast doubt on the testimony of the five customers who identified him as the person who prepared their tax returns.
One of those customers, K.H., testified that her 2011 tax return was prepared by a "young black male." When she was asked to identify the tax preparer in a photographic lineup before trial, she tentatively selected Keleta’s photograph and said, "It looked like this guy but I’m not sure." The government did not ask K.H. to identify Keleta at trial. On cross-examination, the defense asked whether the person who prepared her tax return was in the courtroom. She looked around the courtroom and said, "No. I don’t see him." Keleta was seated in the defendant’s chair.
The defense also attempted to argue misidentification as to C.M.’s testimony. It noted that, before the grand jury, C.M. remembered Paulos’s name and that he prepared her 2012 tax return, but she could not recall who prepared her 2011 tax return. At trial, C.M. testified that she later remembered that "Ace ran the company. That came back to me. He ran it. So he was the first one to do my taxes. Tek was the second one." The defense attempted to impeach this testimony by noting that C.M. only remembered Keleta’s name after reviewing her tax return. It suggested that the tax return itself was an unreliable indicator of who prepared it.
However, Keleta was not able to raise a misidentification defense as to E.G., who positively identified him in court and testified that he had known Keleta for his entire life. E.G. testified that Keleta prepared his 2011 tax return and reported false information without E.G.’s knowledge. Keleta continued to assert his misidentification defense during his closing argument. He again highlighted Lewis’s testimony that he and Paulos looked alike. The government responded to this argument in rebuttal, in part, as follows:
Oh, another thing that the defense counsel just said was that this idea that Tek and Ace look alike and that possibly the clients got them confused because Tek would introduce himself as "Ace." Ladies and Gentlemen, Tek was in this courtroom this week. And if I had known that this was going to be a general theme of the defense counsel, I would have asked Tek to stand up. If you’ve been paying attention to who comes and goes out of this room, ---
THE COURT: Hold on.
[Defense Counsel]: Your Honor, I would object.
THE COURT: What’s your objection, Mr. Gelfand?
[Defense Counsel]: This is improper, Your Honor.
THE COURT: Overruled. Proceed.
[Government Counsel]: If you’ve been paying attention to who comes and goes outside of this courtroom, I ask you: Have you seen Ace’s twin in this courtroom this weekend or this week? Have you seen his twin?
The jury found Keleta guilty of conspiring to defraud the United States as charged in count one. It found two overt acts in furtherance of the conspiracy: Keleta’s preparation of E.G. and C.M.’s 2011 tax returns. The jury also found Keleta guilty of willfully aiding and assisting in the preparation of C.M.’s false 2011 tax return as charged in count two, but not guilty of willfully aiding and assisting in the preparation of K.H.’s false 2011 tax return as charged in count three.
Keleta appeals, arguing that the district court erred in denying his motion to suppress, that the government’s misleading statements in closing argument deprived him of a fair trial, and that he was improperly assessed a four-level enhancement in his offense level at sentencing. We address each argument in turn.
III. Motion to Suppress
The district court denied Keleta’s motion to suppress the evidence seized at the U-City Tax Service branch. It concluded that there was probable cause to issue the warrant and, even if there was not, the good-faith exception to the exclusionary rule applied. See United States v. Leon, 468 U.S. 897, 926, 104 S.Ct. 3405, 82 L.Ed.2d 677 (1984). "When reviewing the denial of a motion to suppress, we review the district court’s legal conclusions de novo but its factual findings for clear error." United States v. Burston, 806 F.3d 1123, 1126 (8th Cir. 2015). "We will affirm the district court’s denial of a motion to suppress evidence unless it is unsupported by substantial evidence, based on an erroneous interpretation of applicable law, or, based on the entire record, it is clear a mistake was made." Id. (cleaned up).
Keleta argues that false statements in Special Agent Kenney’s warrant affidavit misled the issuing magistrate judge into believing that there was probable cause to search the U-City Tax Service branch. In particular, Keleta notes that: (1) a chart in the affidavit contained mathematical errors; and (2) the SDC did not expressly find that Eriace’s tax returns were "suspicious," as the affidavit indicated. He also asserts that the affidavit did not provide context for Kenney’s claim that an unusually high percentage of Eriace’s tax returns claimed certain tax credits. Because the statements were made in "reckless disregard of the truth," Keleta argues, the good-faith exception does not apply. Leon, 468 U.S. at 923, 104 S.Ct. 3405. And he contends that, without the misleading statements in the affidavit and without the proper context, there was not probable cause to issue the search warrant.
Special Agent Kenney’s affidavit stated that, for tax year 2011, 86% of Eriace’s tax returns claimed an Earned Income Credit (EIC), 89% claimed Schedule C income, 11% claimed household help income, and 56% claimed the maximum EIC. In fact, 77% of Eriace’s tax returns claimed EIC, 51% claimed Schedule C income, 14% claimed household help income, and 51% claimed the maximum EIC. Kenney noticed the errors before the suppression hearing and notified the lead prosecutor, who, in turn, notified the defense and the court. At the hearing, Kenney testified that the errors did not affect his assessment that the returns were suspicious.
The Fourth Amendment provides that "no Warrants shall issue, but upon probable cause ...." A magistrate judge evaluating a warrant affidavit must make "a practical, common-sense decision whether, given all the circumstances set forth in the affidavit before him, including the ‘veracity’ and ‘basis of knowledge’ of persons supplying hearsay information, there is a fair probability that contraband or evidence of a crime will be found in a particular place." Illinois v. Gates, 462 U.S. 213, 238, 103 S.Ct. 2317, 76 L.Ed.2d 527 (1983). "If a warrant affidavit contains a false statement made deliberately or out of reckless disregard for the truth, and the false statement was necessary for the finding of probable cause, then the warrant is invalid." United States v. Wold, 979 F.2d 632, 634 (8th Cir. 1992). But if the affidavit would establish probable cause even "if the allegedly false information is ignored or the omitted information is supplemented," the evidence seized pursuant to the warrant need not be suppressed. See United States v. Snyder, 511 F.3d 813, 816 (8th Cir. 2008).
Solely for purposes of addressing Keleta’s argument, we assume that the warrant affidavit contained false statements that were made "out of reckless disregard for the truth." But even without the statements that Keleta disputes, the affidavit still established probable cause to search the U-City Tax Service branch. Kenney began his twenty-five page affidavit by explaining his professional background and experience investigating violations of the Internal Revenue Code and related offenses. Kenney relied on several types of information to support his warrant application. He explained that the SDC noted particular characteristics of the returns prepared by Eriace. These included a large percentage of returns claiming tax credits, which led to an unusually high percentage of tax payers receiving refunds. Even without the specific percentages that Kenney had listed, the affidavit expressly described the percentage of returns that caught the eye of the SDC as "large," "high," or "unusually high." The refunds themselves were also "suspiciously large."
The district court credited Kenney’s testimony that, "at the time he signed the affidavit, he believed it was true and accurate," and found that there was "no evidence that would support a finding that Agent Kenn[e]y was being dishonest or acted with reckless disregard for the truth" when he made the disputed statements.
Keleta contends that Kenney’s statement that "[t]he SDC deemed the returns prepared by [Eriace] to be suspicious" is false. But removing it—as we do for purposes of this analysis—does not eliminate the suspicious nature of the pattern of tax returns prepared by Eriace as described in the affidavit.
Kenney further noted that many of the Eriace returns included information that was not verifiable. A "high percentage" of the returns reported income that could not be easily confirmed, including self-employment income and "household help" income. And most of the returns claiming education-related deductions and credits did not include supporting documentation from any educational institution.
"Another suspicious characteristic" of the returns was the way in which "unsubstantiated income" was combined with verifiable income to qualify taxpayers for the maximum or near-maximum Earned Income Credit amounts. Kenney explained that this pattern repeated itself "on approximately 125 returns for tax year 2011." Even the personal tax returns for Keleta, Paulos, and Lewis fit this unusual pattern.
Also included in the affidavit was a robust description of the undercover investigation. The IRS sent an agent, posing as a customer, to U-City Tax Service. The agent met with Lewis, who said that she could increase the agent’s refund, "but the fee will go up." Lewis then "fabricated" self-employment income on the agent’s tax return, which resulted in a significantly higher refund. When the agent expressed concern about being "asked questions" regarding the information Lewis had added to her return, Lewis assured her that she would make a "notation" that the agent had provided the proper documentation to verify the self-employment income.
The IRS also had received three anonymous letters alleging, in a fair amount of detail, that Eriace was preparing fraudulent tax returns. These allegations included that Eriace claimed "false education and fuel credits," "false dependents," and falsified information "to claim inflated earned income tax credits." Finally, a review of Eriace’s checking account indicated that there were numerous withdrawals that appeared to be for personal, rather than business-related, expenditures. This further underscored the suspicious nature of Eriace’s business dealings.
An affidavit in support of a search warrant must provide sufficient information for a magistrate judge to assess whether there is a "fair probability" that evidence of a crime will be found at the location to be searched. Gates, 462 U.S. at 238–39, 103 S.Ct. 2317. When an affidavit relies on statistical information, it may be necessary to provide context for the magistrate judge to understand whether the data is indicative of unlawful activity. See, e.g., United States v. Singh, 973 F. Supp. 7, 11 (D.D.C. 1997). Here, the affidavit provided a detailed description of the unusual pattern of returns prepared by Eriace and the undercover transaction with Lewis. Even without the statements that Keleta disputes, this provided sufficient context for the issuing magistrate judge to evaluate whether there was a fair probability that evidence of illegal activity would be found at this particular U-City Tax Service branch. And the record supports the magistrate judge’s conclusion that, "taken together," the circumstances described in the affidavit were sufficient to establish probable cause. We affirm the district court’s denial of Keleta’s motion to suppress.
IV. Prosecutorial Misconduct
The government concedes that it was at least misleading to tell the jury that Paulos was in the courtroom during trial and that it could reject Keleta’s misidentification argument if it did not see his "twin" in the courtroom that week. Paulos was only in the courtroom for "a matter of minutes" during jury selection, when approximately 40 potential jurors were also in the courtroom. Before the jury was sworn, he was excluded on a motion that the government joined. He was not in the courtroom for any part of the four-day trial that followed.
Keleta timely objected when the government began to argue that Paulos was in the courtroom during trial, but his objection was overruled. He subsequently filed a motion for a new trial asserting that the government’s argument was "unambiguously improper" because it "invited the jury to consider information not presented at trial" and was "plainly untrue." The district court denied the motion, reasoning that "Keleta opened the door to this argument when he presented a mistaken identity defense during trial and closing argument."
"This court has set forth a two-part test for reversible prosecutorial misconduct: (1) the prosecutor’s remarks or conduct must in fact have been improper, and (2) such remarks or conduct must have prejudicially affected the defendant’s substantial rights so as to deprive the defendant of a fair trial." United States v. Johnson, 968 F.2d 768, 770 (8th Cir. 1992) (citation omitted). Where, as here, the defendant objects to the alleged misconduct during closing arguments, we review for an abuse of discretion. United States v. Alaboudi, 786 F.3d 1136, 1141 (8th Cir. 2015).
"[T]he arguments of counsel must be confined to the issues of the case, the applicable law, the pertinent evidence, and such legitimate inferences as may properly be drawn therefrom." London Guarantee & Accident Co. v. Woelfle, 83 F.2d 325, 342 (8th Cir. 1936) ; see also United States v. Guerra, 113 F.3d 809, 817 (8th Cir. 1997) ("A closing argument should be based upon the facts in evidence and reasonable inferences therefrom and should not assert factual propositions for which there are no evidentiary support."). "To the extent an attorney’s closing argument ranges beyond these boundaries it is improper." United States v. Segal, 649 F.2d 599, 604 (8th Cir. 1981) (citation omitted).
The government’s argument that Paulos was in the courtroom during trial was improper. Keleta laid out his misidentification defense in his opening argument. Nothing prevented the government from responding to this defense based on admissible evidence introduced at trial. It was inappropriate to instead respond by referring to facts not in evidence. See United States v. Harper, 466 F.3d 634, 645 (8th Cir. 2006) ("A defendant is entitled to have his or her guilt determined solely based on the evidence introduced at trial ....").
Moreover, the government’s representation was highly misleading. "In closing argument, a prosecutor is entitled to make a fair response and rebuttal when the defense attacks the government’s case." United States v. Grauer, 701 F.3d 318, 323 (8th Cir. 2012). "But, while he may strike hard blows, he is not at liberty to strike foul ones." Berger v. United States, 295 U.S. 78, 88, 55 S.Ct. 629, 79 L.Ed. 1314 (1935). The government’s assertion that a person it had moved to exclude from the courtroom was present during trial was a foul blow, not a fair one. It was not within the district court’s discretion to conclude otherwise.
Next, we must decide "whether the conduct, viewed in the context of the entire trial, was so offensive that it deprived the defendant of a fair trial." Johnson, 968 F.2d at 771. In conducting this analysis, "this court considers three factors: (1) the cumulative effect of such misconduct; (2) the strength of the properly admitted evidence of the defendant’s guilt; and (3) the curative actions taken by the trial court." Id. (citation omitted). To affirm, we must be able to declare a belief that the government’s misconduct "was harmless beyond a reasonable doubt." Id. at 772 (quoting Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967) ).
Although the prosecutor’s misconduct was brief, "a single misstep on the part of the prosecutor may be so destructive of the right to a fair trial that reversal is mandated." Id. at 771. Several features of the prosecutor’s misconduct here are troubling. First, the prosecutor’s statement that "Tek was in the courtroom this week. And if I had known that this was going to be a general theme of the defense counsel, I would have asked Tek to stand up," implied that Tek was in the courtroom during the trial itself, when only eight to ten people were present. Given the limited audience, the jury could have understood this as casting serious doubt on Keleta’s defense. Second, the government’s argument did not go to a collateral issue at trial. The government acknowledged in its brief that "[t]hroughout trial, defense presented a false identification theme as its primary defense." The government’s representation thus went to the heart of Keleta’s case. Cf., e.g., United States v. Hale, 1 F.3d 691, 694 (8th Cir. 1993) (improper remarks that "concerned a collateral matter" did not significantly affect the fairness of trial). Third, although Keleta raised his misidentification defense in his opening argument, on cross-examination, and in his closing argument, the government wrongly suggested to the jury that it had insufficient notice "that this was going to be a general theme of the defense counsel." See United States v. Holmes, 413 F.3d 770, 775 (8th Cir. 2005) (it is improper to "encourage the jury to focus on the conduct and role of [the defense] attorney"). And fourth, the jury was unable to verify the government’s misleading representation because it was based on facts not in evidence, and defense counsel was unable to respond to it because it was made during the rebuttal phase of closing arguments. See id. at 776 ("The potential for prejudice is great during closing arguments, especially when the defense has no opportunity for rebuttal.").
The district court could have mitigated the impact of the government’s improper argument by sustaining Keleta’s timely objection and giving a curative instruction to the jury. See, e.g., United States v. Singer, 660 F.2d 1295, 1305 (8th Cir. 1981) (finding improper statements harmless, in part, because of "the prompt cautionary actions taken by the district judge"). But the district court did not take any curative action. Instead, "the district court overruled defense counsel’s timely objection to the improper argument, which told the jury they could appropriately consider the argument in the deliberations they were about to begin." Gilster v. Primebank, 747 F.3d 1007, 1012 (8th Cir. 2014). And the government continued to make its improper argument even after defense counsel objected by asking the jury whether it had seen Keleta’s "twin" in the courtroom that week.
The evidence against Keleta was not, in all respects, overwhelming. Count three charged Keleta with willfully aiding and assisting in the preparation of K.H.’s 2011 tax return. But K.H. could not say with confidence who prepared her 2011 tax return and failed to identify Keleta in court. The government’s improper argument could have reasonably affected the jury’s verdict on this count by removing one reason to doubt K.H.’s shaky identification of Keleta. But the jury was apparently not swayed by the government’s "twin" argument, as it acquitted Keleta of this count.
Where the identification evidence was stronger, the jury’s verdict was different. Count one charged Keleta with conspiring to defraud the United States. Unlike count three, several of the witnesses who implicated Keleta in count one identified him at trial. E.G. had known Keleta for his entire life and identified him as the person in the defendant’s chair. Lewis had known Keleta since 2009, worked with him every day during the 2011 tax year, and similarly identified him at trial. And unlike K.H., who only tentatively identified Keleta in a pre-trial photographic lineup but was "not sure" about her identification, C.M. positively identified Keleta in a photographic lineup both before and during trial. Given the cumulative strength of this identification evidence, we conclude that the government’s improper argument that Paulos was in the courtroom during trial could not have "reasonably affected" the jury’s verdict on this count. Johnson, 968 F.2d at 770.
The issue is closer as to count two, which charged Keleta with preparing C.M.’s 2011 tax return. The defense raised a misidentification argument as to this count by noting that C.M. was unable to recall the name of the person who prepared her 2011 tax return before the grand jury and that she had to review her tax return before she was able to testify that it was prepared by Keleta. However, at no point did the defense suggest that C.M. may have mistaken Keleta for Paulos. And C.M. distinguished between Keleta and Paulos at trial, testifying that Keleta was the first one to prepare her taxes whereas Paulos was the second one. She also testified that Paulos took her to the currency exchange to cash her 2011 tax return. This matched Lewis’s testimony that Keleta often prepared tax returns at U-City Tax Service for the 2011 tax year but was absent during the 2012 tax year. It also matched Lewis’s testimony that it was Paulos who took the customers to the currency exchange to cash their refund checks. Based on the evidence presented, we are confident that the government’s improper argument, which rebutted an argument that was not even raised as to this witness, could not have "reasonably affected" the jury’s determination. See id.
We affirm, but only because a careful review of the jury’s verdict and the evidence presented at trial has convinced us, beyond a reasonable doubt, that the government’s misconduct could not have reasonably affected the jury’s verdict on either count of conviction. See id. at 770, 772.
V. Role Enhancement
The Presentence Investigation Report applied a four-level role enhancement to Keleta’s base offense level under United States Sentencing Guidelines § 3B1.1(a). Keleta objected on the basis that the government had not identified five criminally responsible participants as required to apply the enhancement. In response, the government argued that Keleta, Lewis, Paulos, and the "seven customers" who testified at trial each counted as "participants." The district court overruled Keleta’s objection and imposed the enhancement.
"We review a district court’s factual findings regarding whether a leadership enhancement is warranted for clear error and its legal conclusions de novo." United States v. Musa, 830 F.3d 786, 788 (8th Cir. 2016). "It is the government’s burden to prove a leadership enhancement applies by a preponderance of the evidence." Id.
Sentencing Guideline § 3B1.1(a) allows a four-level increase in a defendant’s offense level "[i]f the defendant was an organizer or leader of a criminal activity that involved five or more participants or was otherwise extensive." "A ‘participant’ is a person who is criminally responsible for the commission of the offense, but need not have been convicted." USSG § 3B1.1 cmt. (n.1) (2016). "A person who is not criminally responsible for the commission of the offense ... is not a participant." Id. To be criminally responsible of conspiring to defraud the United States, a person "must have known that an agreement existed to defraud the government and must have knowingly and voluntarily became part of the agreement." United States v. Hammerschmidt, 777 F. App'x 171, 173 (8th Cir. 2019).
Keleta concedes that the government satisfied its burden of proving that he, Lewis, and Paulos were criminally responsible participants. He also concedes that C.M. was likely a criminally responsible participant. But Keleta argues that this only shows that there were at most four participants and the government was required to prove that there were five participants for the enhancement to apply. We agree.
Apart from C.M., none of the U-City Tax Service customers who testified at trial indicated that they were aware that false information was being reported on their tax returns. Although they each acknowledged at trial that the information was false, they denied providing the information. The government emphasized this to the jury, remarking that, "They did not give Ace that information. Ace made it up himself." In presenting the customers as victims of Keleta’s scheme, rather than knowing participants in it, the government failed to provide direct or circumstantial evidence that the customers knew that there was an agreement to defraud the government and knowingly and voluntarily became part of the agreement. Being a customer of a business that prepares false tax returns does not, by itself, render a person criminally culpable. See United States v. Mentzos, 462 F.3d 830, 842 (8th Cir. 2006) (explaining that an individual who is unwittingly involved in a conspiracy does not qualify as a participant under § 3B1.1 ).
At the district court, the government did not identify any potential participants other than the customers who testified at trial. Nor did it do so in its brief on appeal. At oral argument, the government argued, for the first time, that persons other than those identified before the district court were participants. That is too late. See Twin Cities Galleries, LLC v. Media Arts Grp., Inc., 476 F.3d 598, 602 n.1 (8th Cir. 2007) ("Because this point was raised for the first time at oral argument, and has not been briefed, it is waived."); United States v. Hunter, 554 F. App'x 5, 11 (D.C. Cir. 2014) (applying waiver in this context).
Accordingly, we conclude that the government failed to carry its burden of proving that the four-level enhancement applied. We therefore vacate Keleta’s sentence and remand to the district court for resentencing.