United States v. Riddle

REVISED IN THE UNITED STATES COURT OF APPEALS

FOR THE FIFTH CIRCUIT

No. 95-20251

UNITED STATES OF AMERICA, Plaintiff-Appellee,

versus

JOHN C. RIDDLE, Defendant-Appellant.

Appeal from the United States District Court for the Southern District of Texas

January 7, 1997

Before KING and HIGGINBOTHAM, Circuit Judges, and LAKE,* District Judge.

PATRICK E. HIGGINBOTHAM, Circuit Judge:

John C. Riddle appeals his convictions for bank fraud,

misapplication of bank funds, making false entries, and conspiracy. Although he argues a variety of points of error, we limit our

discussion to the trial court’s evidentiary rulings. We are persuaded that the cumulative and interactive effect of four rulings requires that we reverse the judgment of conviction and remand for a new trial.

* District Judge for the Southern District of Texas, sitting by designation.

I.

Riddle opened Texas National Bank–Post Oak on May 7, 1984. He was chairman of the board and co-trustee of a voting trust that controlled a majority of the shares. Because of unusually high opening day deposits totaling around $38 million, the Office of the Comptroller of the Currency (“OCC”) initiated an examination of TNB–Post Oak only sixty days after the bank was launched. An OCC inspector, Gary Meier, discovered that the bank’s purchases of five $800,000 loan participations violated its legal lending limit. Meier expressed to the bank his concern that it was imprudently relying on repurchase agreements without inspecting the creditworthiness of the entities that had promised to repurchase the participations if they went bad. He explained to Riddle and the bank’s board that OCC regulations required banks to review loan participations as thoroughly as if the bank were initiating the loan.

In March of 1985, ten months after opening TNB–Post Oak, Riddle opened a second bank, Texas National Bank–Westheimer ("TNB–W"). The criminal charges at issue in this case arose out of Riddle’s relations with this second bank. As chairman, Riddle held approximately ten percent of the bank’s stock. As with TNB–Post Oak, a voting trust named Riddle as co-trustee. Riddle was not an officer, but he exercised control over various board members. The board declared that Riddle was not an executive officer. But in November of 1985, the OCC concluded that the board’s declaration was ineffective because Riddle in fact controlled the bank’s *3 activities, including the activities of Victor C. Bane, the bank’s president and loan officer.

The OCC inspected TNB–W in September of 1985 and found a number of problems. Its loans-to-debt ratio was an unhealthy 105 percent. The most serious problem concerned loans to insiders. On opening day, it granted a $400,000 unsecured loan to Riddle, which immediately put the bank in violation of banking regulations as well as its own policies. The next month, Riddle had Bane issue a $415,000 letter of credit to Rick Dover, Riddle’s real estate development partner, to satisfy the lender behind one of Riddle’s commercial real estate projects. Dover did not have to post collateral, and in exchange Riddle granted a 15 percent interest in the project to Dover. According to the OCC, the bank failed to keep proper documentation for transactions with businesses owned by bank directors. A full twenty-five percent of the bank’s gross loans went to insiders or companies related to insiders. The OCC inspector discussed the loans-to-insiders violations with Riddle and the board. His report listed loans to Riddle in particular as problematic.

TNB–W lost money during its first six months. To remedy this, the board decided to pursue a strategy suggested by Bane: the bank would raise its interest rate on certificates of deposit to generate short-term assets. The strategy worked as planned, and between September 30, 1985, and December 31, 1985, TNB–W tripled its assets. To pay the interest on these certificates, the board resolved to purchase loan participations. Riddle suggested that *4 TNB–W turn to Vernon Savings & Loan, a thrift operated primarily by Don Dixon, the chairman of Dondi Corporation, Vernon’s controlling shareholder. Riddle had done personal business with Vernon and Dixon in the past. He did not disclose to the board, however, that he had a personal business interest in TNB–W’s purchase of participations from Vernon.

In addition to opening banks, Riddle was involved in real estate development. In September of 1984, Riddle and Dover formed Hickory Creek Joint Venture to purchase a 1230-acre tract west of Houston called Park Green. They signed a note for $40 million. In the late spring of 1985, they bought an adjacent 230 acres with $13 million in financing. Vernon bought 35 percent of the first loan; Western Savings and Loan bought the other 65 percent. Vernon and Western also funded the second loan and took a pro rata profit participation that would take effect upon sale of the property.

By the fall of 1985, Riddle was having cash-flow problems and wanted to sell the Park Green property. Another Houston developer, John Ballis, expressed interest in buying Park Green and proposed swapping Park Green for a piece of land known as the Superior Oil tract, which was located closer to downtown Houston and thus was more desirable for development. Ballis had deposited $1 million to obtain an option to purchase the Superior Oil tract before December 23, 1985. On September 24, Riddle and Dover signed a letter of intent to purchase the Superior Oil tract from Ballis in exchange for the Park Green property.

At the end of October of 1985, Riddle sought funding for the Superior Oil deal from Vernon and Dixon. Dixon explained, however, that regulators had imposed growth restrictions on Vernon and suggested that Riddle find another lender to buy loan participations from Vernon so that Vernon could finance Riddle’s project. In October and November, Dixon and Riddle took a 3-week European vacation, during which they explored ways to structure transactions so that Vernon could fund the Superior Oil deal. At a November meeting at Dixon’s office, Riddle suggested that Vernon buy 30 percent of the $78 million loan that Western would issue for the Superior Oil purchase. He explained that he would have TNB–W buy $8.5 million in loan participations from Vernon. Woody Lemons, Vernon’s president, expressed concern that this would violate bank regulations governing loans to insiders. But Riddle nevertheless went forward with his proposal to the TNB–W board that it buy loan participations from Vernon. Riddle brought Bane to Dixon’s office to work out the details of purchasing the loan participations.

At November and December meetings, Riddle and Bane urged the TNB–W loan committee to purchase participations from Vernon. TNB–W, however, did not have time to review the quality of the loans, and Vernon did not include sufficient documentation to support TNB–W’s purchase. As it turned out, Vernon’s loans were delinquent: Vernon had been paying the interest itself in order to make the loans appear sound.

Dixon and other Vernon officials were aware of Riddle’s scheme and acted to see that TNB–W purchased enough participations to *6 allow Vernon to fund the Superior transaction. On December 4, TNB–W’s loan committee recommended that TNB–W purchase seven participations from Vernon. In connection with the loans, Vernon issued unconditional buyback letters, which guaranteed that Vernon would pay in case the borrowers defaulted. Both Vernon and TNB–W hid these letters from regulators because, as far as the OCC was concerned, they meant that Vernon had not really reduced its loan portfolio after all. TNB–W’s board approved the purchases on December 17. At the insistence of Riddle and Bane, the board also approved the purchase of an additional $6 million in participations from Vernon.

Facing the December 23 deadline, Ballis bought the Superior Oil tract for $64 million on December 17. After forming the Regents Park Limited Partnership, Riddle and Dover agreed to buy the Superior Oil tract from Ballis for $63.1 million. The closing took place on December 23, and Ballis bought the Park Green property for $64.4 million. Vernon was to fund $23 million of the $78 million loan that Riddle and Dover needed to finance the Superior Oil purchase. But on December 24, Dixon told Riddle that Vernon would not send the money until TNB–W sent the $6 million that had been due the previous week for loan participation purchases. Bane purchased an additional 21 participations from Vernon on behalf of TNB–W without approval of the TNB–W board and wired over $9 million to Vernon during the last week of December. During that same week, Vernon wired $10.9 million to Western to buy a participation in Western’s loan to Riddle and Dover.

Bill Plyler, a TNB–W executive vice president, was suspicious of Riddle’s behavior and asked the OCC to audit TNB–W in late December of 1985 or January of 1986. The OCC was especially worried about the high concentration of participations purchased from Vernon and TNB–W’s rapid growth from $10 million to $32 million over the course of three months. Several board members confronted Riddle, who denied that his Superior loan from Vernon had been contingent on TNB–W’s purchase of participations from Vernon. In part as a result of the board’s discovery of Riddle’s conduct, Bane resigned in January, and Plyler replaced him as president. In the aftermath of Bane’s resignation, bank officials discovered that Riddle had caused TNB–W to make a number of imprudent loans for his own or his friends’ benefit. Plyler insisted that Riddle write a letter to the OCC to explain TNB–W’s decisions. Riddle complied, but the government ultimately concluded that the letter contained a number of misrepresentations. In April, an OCC examiner found that 76 percent of TNB–W’s outstanding loans had become delinquent and recommended that the bank be declared insolvent.

Riddle resigned as chairman on June 12, 1986. On November 13, the OCC issued a cease-and-desist order for TNB–W to stop its unsafe lending practices. After Meier conducted a May, 1987, examination, the OCC declared insolvency on May 28 and appointed the FDIC as receiver.

II.

A grand jury indicted Riddle and Bane on one count of bank fraud under 18 U.S.C. § 1344, one count of misapplication of bank funds under 18 U.S.C. § 656, three counts of making false entries in violation of 18 U.S.C. § 1005, and one count of conspiracy in violation of 18 U.S.C. § 371.

Gary Meier, the OCC examiner who prepared reports on TNB–Post Oak in 1984 and on TNB–W in 1987, was one of the prosecution’s primary witnesses at trial. Although he testified as a lay witness, the prosecution elicited opinions that drew on his expertise as a bank examiner. The court admitted all four bank examinations under Fed. R. Evid. 803(8)(B) as public reports prepared as required by law from observations of officials other than law enforcement personnel. Meier read from these examinations in spite of the fact that he was not involved in the 1985 and 1986 examinations. The court explained that Meier had personal knowledge of the reports because he relied on them when he conducted the 1987 examination. The court disagreed with defense counsel’s contention that Meier was testifying as an expert. According to the district court, Meier was a “hybrid lay witness” who could offer explanations of the four relevant bank examinations under the guise of lay opinion testimony. The court admitted his statements about matters such as OCC policy and sound banking practices under the rubric of Fed. R. Evid. 701.

To counter Meier’s testimony, the defense offered the testimony of Stephen Huber, an expert in banking regulation who *9 teaches law at the University of Houston. After holding a proffer hearing, however, the court barred Huber from testifying “based primarily on Rule 403.” Huber had no personal contact with Riddle or Bane, and the court found that his testimony consisted largely of legal conclusions and duplicative explanations of banking practices.

The court made a variety of other contested evidentiary rulings. It admitted as exhibits a proffer letter and sworn statement given by Dixon, who had been convicted and imprisoned for bank fraud in his dealings with Vernon. The defense read portions of these documents during cross-examination, and the court granted the government’s request to admit them in their entirety as prior consistent statements under Fed. R. Evid. 801(d)(1)(B). The court also allowed the government to present evidence that Riddle and Bane violated civil banking regulations in order to show that they possessed criminal intent when they urged TNB–W to buy participations from Vernon. At trial, the defense objected to the reports of bank examiners, to the cease-and-desist order, to much of Meier’s testimony, and to portions of the testimony of Woody Lemons and Bill Plyler, who served as officers at Vernon and TNB–W respectively, on the grounds that the government improperly used civil violations to establish criminal violations. Finally, the court allowed the government to introduce evidence of four unrelated loan transactions in which Riddle used his power at TNB–W for his own personal gain. The court admitted the evidence as probative of a plan or motive under Fed. R. Evid. 404(b).

After seventeen days of trial, the jury returned verdicts of guilty on all counts as to both defendants. The court sentenced Riddle to a total of ten years imprisonment and ordered him to pay four million dollars in restitution.

III.

This was a difficult trial. The government chose methods of proof that forced difficult trial rulings. We are persuaded that the trial tactics resulted in an unfair trial, despite the hard work of the able trial judge to assure the fairness our courts must deliver.

A.

Before Meier began his testimony, the parties and the court agreed that the prosecution had not designated him as an expert and that he would not be offering expert testimony. Counsel for the government told the court that “what I want this witness to talk about are the specific facts that he observed.” This would include such things as accounts of Meier’s interaction with bank officials during his examinations and personal observations of bank records and practices.

With this assurance, the trial court allowed the government to proceed. However, with each new trial day the government pushed to squeeze as much as possible from this “lay witness.” The result is clear, certainly now, that during Meier’s two-and-a-half days on the stand, he wielded his expertise as a bank examiner in a way that is incompatible with a lay witness. In connection with his *11 examination of TNB–Post Oak, Meier explained that “[a]ccording to 12 C.F.R. 32.5, when repayment is expected from only one source, then all of the advances must be combined, again, coming from that one source.” Over the defense’s objections, Meier expressed his opinion that it was not “prudent” for a bank to rely on repurchase agreements issued by banks selling participations rather than on the creditworthiness of borrowers. The next day, Meier expressed his view that bank officers should discuss OCC circulars when the bank receives them and that the OCC expects officers such as Riddle to know the contents of circulars. The defense objected at length to Meier’s testimony about the OCC’s position on whether a bank director may bring loans to his bank. In response, the court reminded that Meier was not an expert, but that his reports had been available for some time and that his testimony should come as no surprise to the defense. “Even if you do consider him an expert,” the court noted, “it seems to me that we have satisfied the requirements of the rule.”

Meier continued to draw on his specialized knowledge as a bank examiner. He testified that it was imprudent “to have the buyback letter stand separate and apart from the participation certificate itself with neither referencing the other.” He asserted that TNB–W violated OCC regulations when it failed to record the fact that Riddle received proceeds from its purchase of participations. He even speculated that unsafe and unsound lending practices, including loans to insiders, caused TNB–W’s failure.

Under Fed. R. Evid. 701, a lay opinion must be based on personal perception, must “be one that a normal person would form from those perceptions,” and must be helpful to the jury. Soden v. Freightliner Corp., 714 F.2d 498, 510-12 (5th Cir. 1983) (quoting Lubbock Feed Lots, Inc. v. Iowa Beef Processors, 630 F.2d 250, 263 (5th Cir. 1980)). We have allowed lay witnesses to express opinions that required specialized knowledge. In Soden, a witness in charge of truck maintenance testified that, based on his experience, step brackets caused the punctures in a fuel tank that had been brought into his repair yard. We held that the district court did not abuse its discretion when it allowed the plaintiff to introduce such lay opinion testimony. “No great leap of logic or expertise was necessary for one in Lasere’s position to move from his observation of holes in Freightliner fuel tanks at the location of the step brackets, and presumably caused by them, to his opinion that the situation was dangerous.” Id. at 512. Other circuits have construed Rule 701 even more broadly. See Wactor v. Spartan Transp. Corp., 27 F.3d 347, 351 (8th Cir. 1994) (admitting under Fed. R. Evid. 701 the opinions of lockmen, “based as they were upon their years of personal experience, their personal inspection of the lockline, their participation with Wactor in the stoppage of the barges, and their positions as the sole eyewitnesses to the wrapping, fouling, and breaking of the line”); Williams Enterprises v. Sherman R. Smoot Co., 938 F.2d 230, 233-34 (D.C. Cir. 1991) (allowing an insurance broker who had personal knowledge of an insured’s business to offer lay opinion testimony on the cause of *13 an increase in the insured’s premiums); United States v. Fowler, 932 F.2d 306, 312 (4th Cir. 1991) (admitting lay opinion evidence as to whether a certain government official would know whether classified budget documents were available to contractors).

Meier, however, went beyond the lay testimony in Soden, as well as the testimony in cases from other circuits. He did not merely draw straightforward conclusions from observations informed by his own experience. Instead, he purported to describe sound banking practices in the abstract. He told the jury how the OCC viewed certain complex transactions. And he asserted a causal relationship between Riddle’s alleged wrongdoing and the ultimate failure of TNB–W. He functioned not as a witness relaying his own observations so much as a knowledgeable bank examiner who could provide the jury with an overview of banking regulations and practices and who could authoritatively condemn Riddle’s actions. He did not offer testimony that a lay person would have been able to offer after conducting the examinations. The district court erred in allowing Meier’s testimony under Fed. R. Evid. 701.

The government insists that Meier was nothing more than a fact witness because his review of TNB–W files and the 1985 and 1986 examinations gave him personal knowledge of their contents. It is true that “[t]he modern trend favors the admission of opinion testimony, provided that it is well founded on personal knowledge and susceptible to specific cross-examination.” Teen-Ed, Inc. v. Kimball Int’l, Inc., 620 F.2d 399, 403 (3d Cir. 1980). Based on this rule, Meier could draw specific conclusions from his work on *14 the 1984 and 1987 examinations, such as that Riddle did not heed Meier’s 1984 advice on self-dealing. See United States v. Leo, 941 F.2d 181, 192-93 (3d Cir. 1991) (allowing an auditor to relate the basis for his opinion that the defendant had altered purchase order dates in a government contract); United States v. Grote, 632 F.2d 387, 390 (5th Cir. 1980) (allowing an IRS official to compare a defendant’s tax returns by characterizing some as “acceptable” and some as “unacceptable”), cert. denied, 454 U.S. 819 (1981). But latitude under Rule 701 does not extend to general claims about how banks should conduct their affairs. Meier’s opinions that TNB–W operated imprudently and that its imprudence caused it to fail depend on an expert’s understanding of the banking industry.

The government also contends that Meier’s opinions were admissible because the prosecution identified him as a witness long before trial and provided his reports to the defense. At trial, however, the government made a point of presenting Meier as a fact witness rather than as an expert. “[A] party cannot seek to have a witness certified as an expert on appeal when the party did not seek to have the witness qualified as an expert before the district court.” Leo, 941 F.2d at 192 (citing United States v. Hoffner, 777 F.2d 1423, 1425 n.1 (10th Cir. 1985)).

B.

The defense proposed to offer Stephen Huber, a professor at the University of Houston Law Center, to show that banking regulations did not require Riddle to disclose his interest in the Vernon participations and that Riddle and TNB–W adhered to industry *15 standards when it purchased loans from other banks. The court ultimately acknowledged that Meier was mistaken when he stated that 12 C.F.R. pt. 31 required Riddle to report that the purchase of participations from Vernon would allow Vernon to finance the Superior Oil deal. As Professor Huber explained at the hearing, the regulations do not apply when a bank buys participations from a thrift.

The government, then, had to prove that Riddle knew that he was doing something wrong even though he was committing no regulatory violation. Much of Meier’s testimony was an effort to convince the jury that Riddle encouraged the TNB–W board to engage in unsafe lending practices when he encouraged it to buy participations from Vernon. Had he been allowed to testify, Huber would have told the jury that TNB–W handled the Vernon participations in the way that any other bank would have handled them. According to Huber, it is difficult for banks buying small loan participations to acquire documentation from the borrower, so they routinely rely on buyback letters issued by the selling bank.

After more than two weeks of testimony, it is understandable that the court would be wary of allowing the defense to present an expert witness to testify about the proper interpretation of regulations and to make general statements about banking practices. But after giving Meier extensive leeway, the court abused its discretion in refusing to allow Huber to testify regarding Riddle’s knowledge that he was doing something wrong in not making the disclaimer to the board. Testimony that other banks would have *16 made the same decision to buy the Vernon participations would have supported Riddle’s contention that any failure to disclose his interest was not deceitful or even intentional. With Huber’s testimony, the defense could have countered some of the damaging opinions offered by Meier and contained in the OCC examinations. The loss of Huber’s testimony handicapped the ability of the defense to tell the jury its own version of how banks operate and what precautions bankers such as Riddle know they should take.

C.

According to Riddle, the district court admitted into evidence a variety of documents that prejudiced the jury. Among other things, Riddle objects to the admission of portions of the four bank examinations, a proffer letter from Dixon’s attorney, and transcripts of Dixon’s proffer statements.

During Meier’s testimony, the court ruled that it would admit those portions of the OCC reports that Meier read from or discussed during his direct examination. Riddle argues that the reports are inadmissible hearsay. We assume for the sake of argument that the reports were “matters observed pursuant to duty imposed by law as to which matters there was a duty to report” and that bank examiners are not “police officers” or “other law enforcement personnel.” Fed. R. Evid. 803(8)(B). See United States v. Copple, 827 F.2d 1182, 1189 (8th Cir. 1987) (“[Admitting an FDIC] investigation is not improper merely because it seeks evidence that by its nature could be relevant to a civil as well as to a potential criminal proceeding.”), cert. denied, 484 U.S. 1073 *17 (1988); United States v. Quezada, 754 F.2d 1190, 1194 (5th Cir. 1985) (admitting a warrant of deportation under Rule 803(8)(B) because it was “prepared in a routine, non-adversarial setting” rather than “from the arguably more subjective endeavor of investigating a crime and evaluating the results of that investigation”).

But even if the reports fall under a hearsay exception, Riddle has a strong argument that their contents prejudiced him and that some of the reports were not relevant. The government’s asserted purpose in offering the reports was to show that investigators told Riddle in 1984 that banks should not rely on repurchase agreements to assure the creditworthiness of loan participations and told him in 1985 that bank officers may not take advantage of their positions to obtain loans. We agree that these facts are relevant to Riddle’s knowledge that his plan to use TNB–W to free up funds for the Superior Oil project was a violation of law or that he had some other duty to disclose his project. Unfortunately, the reports conveyed statements and implications that conveyed much more information to the jury.

All four reports include sections entitled “Violations of Law and Regulation.” The 1984 report limits its criticism of TNB–Post Oak to its violation of lending limits when it purchased five $800,000 participations from the same bank. Later reports, however, contain longer lists of violations and more pointed criticism of TNB–W. Among other regulatory and statutory violations, the 1985 report cites the bank for violation of 12 *18 C.F.R. pt. 215.5(c)(3) for the $400,000 loan to Riddle. The cover letter to the 1986 report makes an ominous diagnosis:

Significant law violations were disclosed involving insiders. These included infractions of the legal lending limits and repeat violations involving loans to insiders, Regulation O. Directors are reminded of their potential liability for losses sustained on credits exceeding legal limitations. Satisfactory procedures must be implemented to prevent future violations.

The report mentions Riddle in connection with violations of 12 U.S.C. § 375a and 12 C.F.R. pt. 215.5(c)(3). And the cover letter to the 1987 report announces that

[t]he condition of the bank was found to be extremely critical. . . . The extremely critical condition of the loan portfolio is the direct result of the poor credit underwriting standards of the previous management in conjunction with insider abuse, suspected fraudulent loan transfers, and the rapid and extended deterioration in the Houston area economy.

That report included descriptions of violations of TNB–W’s legal lending limits that contributed in part to the appointment of an FDIC receiver.

Introducing these documents into evidence did more than provide the jury with evidence that Riddle knew that he should have come clean with the TNB–W board. It provided the jury with a fouryear history of Riddle’s banking endeavors that tied him to dozens of regulatory violations. It gave the jury reason to connect Riddle’s 1985 scheme with two gloomy reports issued after his allegedly criminal conduct was complete. At bottom, it put Riddle at the center of a spectacular bank failure. But Riddle was not on trial for being an ineffective or even a corrupt banker. He was on trial for using his position as a TNB–W officer to convince the *19 TNB–W board to purchase participations from Vernon that would help him personally and disregarding his duty to disclose his personal interest in the deal. The reports — as well as much of Meier’s testimony drawn from them — were of little probative value on that score in comparison to the danger of prejudicing, confusing, or misleading the jury. See Fed. R. Evid. 403. Indeed, it is difficult to understand how the 1987 examination was relevant at all; its primary purpose seems to have been to allow Meier to testify based on the 1985 and 1986 reports, which he relied on in preparing the 1987 report. We said in United States v. Christo, 614 F.2d 486, 492 (5th Cir. 1980), that “[t]he government’s evidence and argument concerning [regulatory] violations . . . impermissibly infected the very purpose for which the trial was being conducted — to determine whether Christo willfully misapplied bank funds with an intent to injure and defraud the bank, not whether Christo violated a regulatory statute prohibiting the bank from extending him credit in excess of $5,000.” The same principle applies in this case. The trial court abused its discretion when it allowed the government to admit extensive evidence about the OCC’s appraisal of TNB–W’s general health and its failure to comply with regulations from its inception to its demise.1

1 Riddle also calls our attention to the court’s decision to admit the cease-and-desist order issued against TNB–W on November 13, 1986. The transcript discloses that the court did admit the order, but the court decided not to send it back with the jury. The record is unclear on whether the jury ever actually saw the order. Nevertheless, Plyler testified that the bank received the order and that it meant that TNB–W had to “stop . . . continuing with unsafe lending practices.” As we indicated in Christo, 614 F.2d at 495, the mention of irrelevant cease-and-desist orders is

Riddle also objects to the admission of two documents in which Dixon connects Riddle with banking violations. The first is a letter and accompanying memorandum that Dixon’s attorney sent to an Assistant United States Attorney in which Dixon offers to provide information about criminal abuses by bank insiders. The unusual offer lists seventeen institutions, twenty-nine individuals, and twenty generic situations involving banking violations. It does not, however, indicate which institution or individual corresponds to which situation. Thus, Riddle’s name appears as an individual who could have been involved in a number of crimes at a number of institutions. The second is Dixon’s sworn statement used in the grand jury investigation of Riddle.

The government moved to admit the letter and memorandum and the sworn statement under Fed. R. Evid. 801(d)(1)(B) as a prior consistent statement to rebut the defense’s suggestion that Dixon fabricated his testimony in order to cut short the prison term he was serving. The defense objected that the statement was not proper rebuttal because Dixon had the same motive to fabricate when he made the statement and sent the letter as he had in court. According to the defense, admission under Rule 801(d)(1)(B) would require a statement made before Dixon had any motive to reduce his time in prison by pleasing the government.

At the time of trial, our law did not impose this requirement. United States v. Parry, 649 F.2d 292, 296 (5th Cir. Unit B 1981).

highly prejudicial. Even if the jury never laid eyes on the order itself, the government’s reference to the order during Plyler’s direct examination was ill-advised and should not have been made.

The Supreme Court, however, has since instructed that Rule 801(d)(1)(B) “permits the introduction of a declarant’s consistent out-of-court statements to rebut a charge of recent fabrication or improper influence or motive only when those statements were made before the charged recent fabrication or improper influence or motive.” Tome v. United States, 115 S. Ct. 696, 705 (1995). Consequently, admitting the statement and the letter was an error.

Apart from the court’s action under Rule 801, admitting the letter and memorandum was inflammatory. According to the memorandum, “many, if not most, of the activities described herein expose these Insiders to significant criminal culpability.” By associating Riddle with more than two dozen alleged white-collar criminals and a score of criminal scenarios, the memorandum could easily suggest that Riddle regularly kept company with, as the memorandum puts it, “good old boys” who make a habit of stealing from banks. The trial court erred in admitting such a suggestive and prejudicial document.

D.

Nine government witnesses spent at least part of their time on the stand discussing four unrelated TNB–W loans that supposedly showed that Riddle systematically withheld information from the bank in order to direct loan proceeds for his personal benefit. As required by United States v. Robinson, 700 F.2d 205, 213 (5th Cir. 1983), the court conducted bench conferences on whether this evidence had probative value and whether it was unduly prejudicial. The prosecution viewed the loans as evidence of “other crimes, *22 wrongs, or acts” that showed Riddle’s “motive, opportunity, intent, preparation, plan, knowledge, identity, or absence of mistake or accident.” Fed. R. Evid. 404(b). For its part, the court stated that “it’s relevant to motive, to plan, the way he did business, the things he did.”

In the first loan, Jim Hague borrowed $350,000 from TNB–W to buy an apartment complex that Riddle owned and wanted to sell. Regulators viewed the loan as an illegal extension of credit to Riddle, but they also concluded in the 1986 examination report that the violation was a result of a misunderstanding of regulations and thus was not willful or intentional. TNB–W also loaned $300,000 to Architects Alliance, Inc., and took an interest in accounts receivable and inventory as collateral. Riddle was the architectural firm’s primary client and owed it more than $500,000. Riddle also owed thousands of dollars to Richard Weems, who had done extensive mowing and landscaping work at some of Riddle’s properties. At Riddle’s suggestion, Weems borrowed $20,000 from TNB–W to keep up with operating expenses while waiting for Riddle to pay his debt. A fourth extension of credit went to Rick Dover, Riddle’s real estate partner, who obtained a $415,000 letter of credit as collateral on a loan issued by a Florida bank to fund the development of a shopping center.

The government explained during its closing argument that these extraneous loans showed “that this was the way that John Riddle did business. That tells you about his mental state when he was entering into the loan participations for the Superior purchase *23 deal.” In other words, the 404(b) evidence was relevant because it established that Riddle consistently withheld information from the bank that he knew he had an obligation to disclose. The government’s 404(b) theory is in agreement with our analysis in United States v. Beechum, 582 F.2d 898, 911 (5th Cir. 1978) (en banc), cert. denied, 440 U.S. 920 (1979):

Where the issue addressed is the defendant’s intent to commit the offense charged, the relevancy of the extrinsic offense derives from the defendant’s indulging himself in the same state of mind in the perpetration of both the extrinsic and the charged offenses. The reasoning is that because the defendant had unlawful intent in the extrinsic offense, it is less likely that he had lawful intent in the present offense.

The government must present enough evidence to permit a reasonable jury to conclude by a preponderance of the evidence that Riddle’s intent in connection with the four extraneous loans was criminal. United States v. Anderson, 933 F.2d 1261, 1269 (5th Cir. 1991); United States v. Guerrero, 650 F.2d 728, 734 (5th Cir. Unit A 1981). Riddle is charged with bank fraud, misapplication of bank funds, and false entries in the records of a bank, all of which require proof that the defendant knew he was making a misrepresentation. See United States v. Kington, 875 F.2d 1091, 1097 (5th Cir. 1989) (following the rule that in misapplication cases, “the government must prove that the defendant knowingly participated in a deceptive or fraudulent transaction” (emphasis in original) (quoting United States v. Adamson, 700 F.2d 953, 965 (Former 5th Cir. Unit B) (en banc), cert. denied, 464 U.S. 833 (1983))). In this case, that means that the extraneous loans must *24 make it more likely that Riddle intentionally kept the TNB–W board in the dark on his personal interest in having TNB–W extend credit.

Our review of the record convinces us that the government did not meet its burden. At most, the evidence suggests that Riddle took improper advantage of his position to encourage TNB–W to extend credit unwisely and for the benefit of his non-banking endeavors. The OCC itself found that any violation Riddle committed in connection with the Hague loan was unintentional. Walter Beard, a TNB–W director, offered undisputed testimony that the TNB–W board “had to know” that Architects Alliance had done extensive work for Riddle because Architects Alliance designed TNB–W’s building. And the government’s evidence does not reveal any deception in Riddle’s role in generating the loan to Weems or the letter of credit to Dover. As far as the record is concerned, Riddle simply suggested that various business associates apply for loans at TNB–W. Some of Riddle’s associates needed money because he was unable to keep up with his debts, but that does not by itself mean that Riddle intended to deceive or defraud the bank. David Hall, a partner at Architects Alliance, testified that Victor Bane instructed him to submit his firm’s records of accounts receivable in a format that excluded Riddle’s name because “it might reflect poorly” on Riddle if TNB–W knew that Riddle owed so much money to Architects Alliance. The insinuation, of course, is that Bane was acting under Riddle’s instructions. Riddle may also have used Bane to hide his debt to Weems and his involvement in real estate ventures with Dover to convince TNB–W to extend credit. *25 But without evidence from government witnesses, we are not willing to allow the jury to make such an attenuated inference.

The government’s 404(b) evidence certainly makes Riddle out to be an irresponsible banker who paid little attention to OCC warnings. But Riddle was not on trial for irresponsibility. He was on trial for bank fraud, misapplication of bank funds, and making false entries. Even if illegal, Riddle’s extraneous selfserving banking practices are irrelevant under Rule 404(b) unless they tend to show that he systematically withheld vital information from the TNB–W board. By allowing the jury to consider these four loans, the trial court made the mistake of treating general evidence of poor banking as if it were evidence of Riddle’s criminal intent to mislead TNB–W’s board of directors.

Even if the extraneous loans were relevant to the government’s case against Riddle, they failed to meet the second prong of the Beechum test: “the evidence must possess probative value that is not substantially outweighed by its undue prejudice and must meet the other requirements of rule 403.” Beechum, 582 F.2d at 911. Unduly prejudicial extraneous evidence often plays on the jury’s emotions unfairly, but “[p]rejudice can result from any of the significant factors set out in Rule 403, of which inflamed passion is only one.” United States v. Zabaneh, 837 F.2d 1249, 1265 (5th Cir. 1988). These other factors include confusion of the issues and misleading the jury, Fed. R. Evid. 403, and they are especially troubling when they take up a significant portion of the government’s case. See Zabaneh, 837 F.2d at 1265-66 (remanding for *26 Robinson findings where a “substantial portion” of the evidence involved extrinsic offenses and registering concern that “[o]ne witness’s testimony pertained in its entirety to such an offense”).

The court’s limiting instructions can “substantially reduce” the danger of prejudice, United States v. Buchanan, 70 F.3d 818, 831-32 (5th Cir. 1995), cert. denied, 116 S.Ct. 1340 (1996), but in this case they did not counteract the prejudicial effect of allowing government witnesses to testify about the extraneous loans for a total of more than a full day. The government called James Hague, David Hall, and Richard Weems for the sole purpose of explaining how Riddle had wronged them in connection with the extraneous loans. In each instance, it was clear that the witness had a gripe against Riddle that had nothing to do with the charged offense. Rick Dover, Walter Beard, William Plyler, and bank examiner Meier also went into the details of the loans. The government analyzed each loan individually during its closing argument to remind the jury that Riddle was “manipulating people and the bank for his personal benefit.” And the 1986 OCC examination contained detailed descriptions and criticisms of the Hague and Architects Alliance loans.

When extraneous activity receives such intense, unfocused attention, it is too likely that the jury will “feel that the defendant should be punished for that activity even if he is not guilty of the offense charged.” Beechum, 582 F.2d at 914. In explaining relevance under the general category of “the way he did *27 business, the things he did,” the court itself demonstrated that the evidence of extraneous loans is powerful not so much because it tends to establish Riddle’s motive or scheme, but because it paints Riddle as lacking the character of an upstanding businessman.2 The government’s extensive and undiscriminating use of the extraneous loans was misleading to the jury, not because the jury was not discerning but because the evidence was offered in such a large and unchecked way that its permissible limited use was overwhelmed.

IV.

We have found four errors: allowing Meier to testify as a lay witness, barring the testimony of Professor Huber, admitting the OCC bank examinations and the Dixon documents, and admitting testimony about four extraneous loans. We ask now whether these errors were so harmful that they mandate reversing the conviction.

Turning these rulings in a different direction would have produced a very different trial. Instead of hours of testimony about extraneous loans, Professor Huber would have given his opinion that Riddle and TNB–W operated in the way that any bank

2 Riddle makes much of the trial court’s statement from the bench that “it’s relevant to the issue of the character, the government has to prove intent, knowledge, motivation, opportunity, all those things, and it’s not inflammatory, it’s not unduly prejudicial if you weigh it in terms of what the government has to prove, so I’m going to let it in.” According to Riddle, this indicates that the court improperly admitted the evidence as general character evidence. We do not base our decision on the use of the word “character.” It appears that the court merely changed its direction of the sentence mid-stride; we will not interpret judges uncharitably when they make extemporaneous remarks from the bench.

would have operated. Instead of reading the OCC’s and Dixon’s claims that Riddle violated banking regulations, the jury would have focused on the narrow question of Riddle’s intent when he kept silent about his interest in the Vernon participations. Looking at the cumulative effect of the errors, we are persuaded that they are not harmless and require a new trial. We express no view as to whether any one of the errors standing alone would be sufficient to justify reversal.

The judgment of conviction of John C. Riddle is reversed, and the case is remanded for a new trial.

REVERSED and REMANDED.