Leonard G. Ambrose III (argued), William P. Weichler, Ambrose, Friedman & Weichler, Erie, PA, for appellee.

Thomas W. Corbett, Jr., U.S. Atty., Bonnie R. Schlueter (argued), Asst. U.S. Atty., Pittsburgh, PA, for appellant.

Before: BECKER, NYGAARD and ALITO, Circuit Judges.

OPINION OF THE COURT

BECKER, Circuit Judge.

John Copple, former President of Mechem Financial, Inc., ("Mechem"), an investment firm specializing in the management of "pre-need" funeral funds, was convicted by a jury of mail fraud, 18 U.S.C. Secs . 1341, 1342, and income tax evasion, 26 U.S.C. Sec . 7201. The district court sentenced him to 71 months imprisonment, a $100,000 fine and three years supervised release, and ordered him to pay over $4 million in restitution. In this appeal, Copple challenges both his conviction and sentence.

Copple challenges his conviction on two principal grounds. First, he argues that the government failed to comply with the requirements of 26 U.S.C. Sec . 6103(h)(5) (which requires the IRS to report whether a prospective juror has been the subject of an audit or other tax investigation) when it limited the scope of the investigation into the jurors' tax records to records since 1986. According to Copple, he is entitled to a new trial because the district court did not strike the entire jury panel after the limitation on the investigation had been disclosed. Reading a reasonableness limitation into the statute, however, we conclude that the requirements of Sec. 6103(h)(5) were met in this case and that the district court did not err when it refused to strike the jury panel. Second, Copple argues that the district court abused its discretion in admitting victim impact testimony that was irrelevant and highly prejudicial. Although we agree with Copple that the admission of the victim impact testimony was error, we believe the error was harmless given the overwhelming evidence of Copple's guilt. We therefore affirm the conviction.

However, we must vacate the judgment of sentence and remand the case for resentencing for two reasons. First, the district court increased Copple's offense level four levels because of the amount of money involved and the large number of victims, which, whether viewed as an enhancement under Sec. 2F1.1(b)(2) or as an upward departure, was improper; second, the court ordered Copple to pay restitution without making the required findings about Copple's ability to pay. On remand, the district court is free to reconsider alternative grounds for upward departure or increase in the offense level mentioned in the original presentence report but not factored into the original sentence. It also must support any order of restitution with factual findings about Copple's ability to pay the order, the financial need of his family, and the relationship between the loss caused and Copple's conduct.

I. BACKGROUND

Over the years a practice has developed in the funeral home business whereby persons who wish to rest assured that their funeral needs are taken care of in the event of a sudden or unexpected death may purchase "pre-need" funeral plans with the funeral director of their choice. In 1986, two Pennsylvania funeral directors, W. James Scott and Michael Orlando, realizing that many funeral directors who had sold "pre-need" funeral plans did not have the time or expertise to manage the plan funds, or to deal with the tax, accounting, and disbursement problems associated with the funds even when they had turned the funds over to conventional trust management plans, conceived of a business idea--a money management firm specializing in "pre-need" funeral accounts. As funeral directors themselves, however, Scott and Orlando lacked the expertise needed to make such a company successful, and hence they sought the aid of someone with considerable experience as an insurance agent and financial planner, defendant Copple.

Copple jumped at the chance to run a money management business like the one Scott and Orlando proposed. He offered to put up $50,000 if Scott and Orlando would contribute their expertise in the funeral business to the venture. They agreed, and Mechem was formed. Copple became president, and Scott and Orlando became "silent partners." Copple promised to oversee the investment decisions himself and to invest the money in "the safest place."1

Copple sold funeral directors on Mechem's services with promises of high yields and low risk. He directed his staff to tell the funeral directors that Mechem invested the "pre-need" funds in high yield, low risk annuities and treasury bonds. Mechem sent letters via the United States mail stating that the money had been invested with reputable insurance companies like John Hancock, Connecticut Mutual, New England Mutual Life, and others. One letter told the funeral directors that "[o]ur investments have been made in insurance companies, annuities, T-bills, long term municipal bond funds, short-term CD's and money markets. Our performance has reflected our excellent investment posture for the last fifteen years."

Copple also had Mechem send out letters representing that it was fidelity bonded. In particular, the letters pointed to a policy issued by an agent named James Domino, a bond issued by the Maryland Casualty Insurance Company, and certain Lloyds of London bond certificates. In addition, Mechem issued quarterly statements to the funeral directors reporting the interest that had accrued on their "pre-need" funds. The sales technique worked. Eventually, about $5 million in "pre-need" account money from Pennsylvania funeral directors made its way into Mechem's coffers. Ohio and Massachusetts funeral directors deposited an additional $7 million.2

Although Mechem and Copple promised the funeral directors high yields and low risk, they gave them neither. Copple did not invest any of the money he received for Mechem in annuities, treasury bills or any similar investments. Mechem had never purchased any fidelity bond. The Maryland Casualty Bond, for example, was actually a general liability policy that a salesman had altered, at Copple's direction, to make it look like a fidelity or surety bond. And the quarterly reports of interest earned were complete fabrications.

Copple actually used most of the money for speculative investments and conspicuous personal consumption. During the three year life of Mechem corporation, Copple bought $5.7 million of rare coins, used $2.8 million to run Mechem, applied $1 million to pay death claims, and spent about $2.5 million on himself. His personal expenses during the time he was running Mechem were lavish to say the least. They included: $228,000 for a building project on his home, $196,334 for furniture, $62,081 for jewelry from Les Crago, $70,279 for jewelry from Fortunoff's, $398,000 for jewelry from Neiman-Marcus, $48,712 for a sable coat, $480,000 for gifts to his family, and a host of similar purchases.

The inevitable occurred in 1989. After rumors surfaced that Mechem could no longer pay the funeral expenses for the "pre-need" accounts and that Copple had invested all of the money in rare coins, the funeral directors made a run on Mechem. Many of the funeral directors asked Mechem to "roll-over" their funds into other money management companies, but Mechem no longer had the money to meet these demands. It filed for bankruptcy in 1990.

While administering the bankruptcy, the trustee discovered Copple's serious mishandling of the Mechem "pre-need" account funds. He discovered that all but $250,000 of the individual "pre-need" account money that had been initially placed in the Mechem investment account had been transferred to the John R. Copple account, from which both business and personal disbursements were made. Some of the assets were still in Copple's hands. Copple turned over to the trustee the rare coins, which were sold at auction for $209,045. The trustee took control of four bank accounts which contained $68,875.73 and an escrow account containing $210,480.78. He also secured the return of a $110,000 deposit that Copple had placed at Neiman-Marcus for the purchase of a 38.33 carat diamond. Most of the money, however, was gone. At last count, the creditors, including the funeral directors, will recoup about twelve cents on the dollar.3 The total amount of money lost by the victims relevant to this criminal case was $4,257,940.45.

On October 17, 1991, a grand jury in the Western District of Pennsylvania returned a 37-count indictment charging Copple and Mechem with mail fraud and tax evasion. Counts 1-16, 18-27, 29-32, and 34 charged Copple and Mechem with mail fraud of the funeral directors. Counts 17, 28, and 33 charged Copple and Mechem with mail fraud of individual investors.4 Counts 35-37 charged Copple with income tax evasion for failing to file returns for 1986, 1987 and 1988.5

After a trial lasting about a month, the jury found Copple guilty on all counts. On December 18, 1992, the district court sentenced Copple on the counts covered by the Sentencing Guidelines (the "Guidelines") to 71 months imprisonment, a $100,000 fine, and three years supervised release. He also ordered Copple to pay $4,257,940.45 in restitution to be made through the bankruptcy trustee. On the counts not covered by the Guidelines, the court sentenced Copple to five years imprisonment to be served concurrently with the Guidelines sentence. Copple has filed a timely appeal challenging both his conviction and sentence.

II. COPPLE'S CHALLENGES TO HIS CONVICTION

A. The 26 U.S.C. Sec . 6103(h)(5) Claim

The first issue we address is Copple's claim that we should reverse his conviction because the requirements of Sec. 6103(h)(5) were not met.6 Section 6103(h)(5) provides in relevant part:

in connection with any judicial proceeding [related to tax administration] to which the United States is a party, the Secretary shall respond to a written inquiry from ... any person (or his legal representative) who is a party to such proceeding as to whether an individual who is a prospective juror in such proceeding has or has not been the subject of any audit or other tax investigation by the Internal Revenue Service. The Secretary shall limit such response to an affirmative or negative reply to such an inquiry.

26 U.S.C. Sec . 6103(h)(5).

On July 27, 1992, about a month before jury selection, Copple moved pursuant to Sec. 6103(h)(5) for disclosure of tax background information of prospective jurors. In its response to Copple's motion, the government agreed to provide the information, but stated that it would be virtually impossible to obtain tax audit information from prior to 1986. The district court granted Copple's motion for the Sec. 6103(h)(5) investigation, but did not mention whether the IRS was to investigate the tax records of the prospective jurors for the years preceding 1986.

The government provided Copple with the IRS review indicating that none of the prospective jurors had been audited or investigated from 1986 to 1991, the years for which the IRS' records were computerized. At a hearing the day before commencement of trial, Copple claimed that he was entitled to the tax information without any limitation as to time period. The government responded that checking records of possible audits occurring before 1986 would require a manual search, which would take weeks, even a month, to complete.

Copple then claimed that he was entitled to ask on voir dire whether any of the prospective jurors had ever been audited by the IRS or whether any of the prospective jurors had ever been the subject of a civil or criminal tax investigation; he also argued that he was entitled to have the IRS verify the answers the jurors gave. The district court granted Copple's request for voir dire but declined to order the IRS to verify the jurors' answers.

On voir dire, the district court asked the prospective jurors: "Have you or any member of your immediate family ever been audited by the Internal Revenue Service?" In response to this question, Juror Number 7, Art Borczon stated that he had been audited during 1988 and 1989 (the audit for 1989 had not yet been resolved), that he had paid a deficiency, and that he had "never [been] satisfied with that." He also represented, however, that he could be a fair and impartial juror. Juror Number 45, James Henderson, also stated that he had been audited about 35 years earlier, but he too testified that he could be fair and impartial. A few other jurors responded that they had not been audited personally, but that members of their families had been.7

Copple did not ask that Borczon, Henderson or the other jurors be dismissed; instead he moved to have the entire panel rejected because the government had failed to comply with Sec. 6103(h)(5). He provided two bases for his motion. First, Copple argued that the investigation was inadequate because it only went back five years. Second, Copple argued that the investigation of all of the jurors was demonstrably unreliable because it had failed to disclose Borczon's audits occurring within the five year period.

The district court, however, denied the motion. Copple submits that this was error and that his conviction therefore should be reversed and the case remanded for a new trial. The government responds that such a ruling was not error because the district court properly found that ordering the IRS to supply such information would have unduly delayed the trial and that, even if it was error, such error was harmless and any prejudice was cured by asking the prospective jurors about their tax histories.

The question whether Copple is entitled to a new trial because the tax investigation was limited to the jurors' records since 1986 requires us to determine the requirements of Sec. 6103(h)(5), something we have not had occasion to do until now. Although the statute explicitly provides a defendant with a right to require that the Treasury provide an affirmative or negative response as to whether a prospective juror has been audited,8 it is silent on the appropriate time period covered by the investigation. In addition, the statute does not specify the procedures that a district court must follow to carry out the purposes of the provision, and it is silent on the consequences, if any, for noncompliance.1. The requirements of Sec. 6103(h)(5).

Copple argues that the statute's lack of any limitation on the appropriate time period covered by the investigation implies that the investigation can have no time limitation and that the government violated the statute when it limited the investigation to the preceding five years. In addition to relying on the statutory language, Copple relies on a Ninth Circuit opinion. United States v. Sinigaglio, 925 F.2d 339, amended, 942 F.2d 581 (9th Cir.1991), which held that where the defendant makes a timely motion for a Sec. 6103(h)(5) investigation, the investigation must cover all of the years the prospective jurors paid taxes. Id. at 341. Copple urges this Court to adopt the Sinigaglio view of Sec. 6103(h)(5).

The government counters with United States v. Spine, 945 F.2d 143 (6th Cir.1991), in which the Sixth Circuit stated that Sec. 6103(h)(5) does not require an investigation extending to all of the years the prospective jurors paid taxes. Spine held that the requirements of Sec. 6103(h)(5) are met as long as the court orders an investigation and, if the IRS cannot locate all of jurors' histories from the time they began paying taxes by the time of trial, the district court obtains such information on voir dire. Id. at 148. Spine reached this result by reading a reasonableness limitation into Sec. 6103(h)(5). According to Spine, Sec. 6103(h)(5) merely requires the district court to order an investigation which would be appropriate under the circumstances, one which would take into account both the cost and inconvenience of the investigation and the ability to get the same information on voir dire.

The Sixth Circuit's interpretation of the statutory language is informed by practical considerations. Allowing a defendant to request an investigation of all of the potential jurors' tax information from the time they began paying taxes could take months, indeed, even as much as a year. See United States v. Johnson, 762 F.Supp. 275, 277 & n. 1 (C.D.Cal.1991), rev'd on other grounds, 991 F.2d 569 (9th Cir.1993). Scheduling criminal cases, which is already difficult enough, would be made even more difficult since a trial with a current jury pool would have to be postponed for months while the IRS completed an investigation. See Spine, 945 F.2d at 148.9 This might also cause serious inconvenience to prospective jurors.10 In addition, strict compliance with Sec. 6103(h)(5) would also impose substantial costs on the IRS since defendants would routinely request information requiring the IRS to conduct manual searches of noncomputerized records.

Interpreting Sec. 6103(h)(5) to require tax investigations stretching back twenty or thirty years would transform Sec. 6103(h)(5) into a significant practical bar to tax prosecutions. It potentially would permit a defendant in a tax case to postpone a trial indefinitely by continually requesting potential jurors' tax information. Spine, 945 F.2d at 148. Indeed, interpreting Sec. 6103(h)(5) to require such an extensive search might make tax prosecutions so expensive that the government would be reluctant to bring them. See United States v. Nielsen, 1 F.3d 855, 858 (9th Cir.1993), cert. denied, --- U.S. ----, 114 S.Ct. 1410, 128 L.Ed.2d 82 (1994).

We should not interpret the language of Sec. 6103(h)(5) to create such an absurd result absent a clear direction from Congress, see Griffin v. Oceanic Contractors, Inc., 458 U.S. 564, 575, 102 S.Ct. 3245, 3252, 73 L.Ed.2d 973 (1982), United States v. Schneider, 14 F.3d 876, 880 (3d Cir.1994), and there is no such clear direction in either the language or the legislative history of the provision. As has been mentioned, the language of Sec. 6103(h)(5) is silent on whether the IRS must search the tax records for all of the time the jurors began paying taxes. And the legislative history of Sec. 6103(h)(5) demonstrates that Congress' principal concern in enacting the provision was merely to ensure that the government and the defendant would have access to the same information--not that the information had such intrinsic worth that Congress meant to bring prosecutions to a standstill while the IRS conducted an investigation.11 Spine, 945 F.2d at 147. There is simply no suggestion that Sec. 6103(h)(5) was meant to create the significant practical barrier to tax prosecutions that would result if we were to accept Copple's interpretation. See, United States v. Lussier, 929 F.2d 25, 30 (1st Cir.1991) (per curiam) ("The statute itself makes no provision for such an extreme alteration of normal trial arrangements.").

We therefore adopt the Sixth Circuit's approach and conclude that Sec. 6103(h)(5) requires only that the investigation into the tax records of potential jurors meet the standard of reasonableness.12 Specifically, upon timely request by the defendant, the district court must grant a reasonable period of time for the IRS to complete a search of the potential jurors' tax records for the time period requested by the defendant. If the district court only allows the defendant enough time for the IRS to conduct a search of the computerized records and not a search of the noncomputerized records,13 the grant of time will be reasonable as long as the computer search is made, and the court elicits on voir dire information about the jurors' tax histories for the period of time not covered by the investigation.14 We add only that Congress might be well advised to revisit the provision and specify more clearly the intent behind it and its requirements.

2. Did the district court comply with Sec. 6103(h)(5)?

Under our reading of Sec. 6103(h)(5), the district court's failure to order a complete search of the jurors' past history was not error. First, after Copple requested the Sec. 6103(h)(5) inquiry, the district court ordered the clerk to provide a list of jurors in the case along with other relevant information to the United States Attorney so that the IRS could conduct the investigation. This occurred about twelve days before the trial and it gave the IRS enough time to search the computerized records and get information about the potential jurors' tax histories for the period from 1986 to 1991.

Second, the district court conducted an extensive voir dire about the potential jurors' tax histories and experience with the IRS including whether they or any member of their family had been audited. The questioning covered all of the period for which the jurors had paid taxes. Moreover, the voir dire worked. It identified a juror who was outside the scope of the IRS audit (Henderson) and a juror who the IRS for some reason simply missed (Borczon).15 Borczon, for example, indicated that he had been unhappy with the audit results but also stated that he could still be fair and impartial. Apparently, his answers were satisfactory since Copple did not even move to strike him. In a sense, then, Copple had access to more accurate information than he would otherwise have received had the inquiry been limited to a full IRS investigation.

For all the foregoing reasons, we hold that the district court complied with Sec. 6103(h)(5).

B. The Victim Impact Evidence

1. Relevance and prejudice.

During its case in chief, the government called to the stand a number of the funeral directors who had put their money in Copple's hands. They testified about their losses, and about the impact of those losses on their lives. Copple argues that evidence about the victims' losses was irrelevant under Federal Rule of Evidence 401, and that the testimony about the impact of the losses was both irrelevant (or at least of negligible probative worth) and unfairly prejudicial, and hence excludable under Federal Rules of Evidence 401 and 403. Our review of these challenges to the conviction is for abuse of discretion. See United States v. Versaint, 849 F.2d 827, 831 (3d Cir.1988).

With respect to the testimony about the financial losses, Copple properly argues that the government does not have to show that the victims actually suffered a loss to satisfy the elements of the mail fraud statute. The essential elements of the crime of mail fraud are 1) a scheme or artifice to defraud; 2) participation by the defendant with specific intent to defraud; and 3) use of the mail in furtherance of the scheme. 18 U.S.C. Sec . 1341; see United States v. Burks, 867 F.2d 795 (3d Cir.1989). Proof of actual loss by the intended victim is not necessary. See United States v. Kelley, 929 F.2d 582, 585 (10th Cir.), cert. denied, --- U.S. ----, 112 S.Ct. 341, 116 L.Ed.2d 280 (1991); United States v. King, 860 F.2d 54, 55 (2d Cir.1988), cert. denied, 490 U.S. 1065 , 109 S.Ct. 2062, 104 L.Ed.2d 628 (1989).

But that does not mean that evidence of loss was irrelevant. Proving specific intent in mail fraud cases is difficult, and, as a result, a liberal policy has developed to allow the government to introduce evidence that even peripherally bears on the question of intent. See United States v. Foshee, 606 F.2d 111, 113 (5th Cir.1979), cert. denied, 444 U.S. 1082 , 100 S.Ct. 1036, 62 L.Ed.2d 766 (1980).16 Proof that someone was victimized by the fraud is thus treated as some evidence of the schemer's intent. See United States v. Heimann, 705 F.2d 662, 669 (2d Cir.1983) ("While technically the success or failure of a scheme to defraud is irrelevant in a mail fraud case, realistically, when the contested issue is intent, whether or not victims lost money can be a substantial factor in a jury's determination of guilt or innocence." (citation omitted)). Also relevant is the defendant's failure to take any steps to ameliorate the loss. See Anderson v. United States, 369 F.2d 11, 15 (8th Cir.1966), cert. denied, 386 U.S. 976 , 87 S.Ct. 1171, 18 L.Ed.2d 136 (1967). The government submits that the evidence about the victims' losses and Copple's refusal to make good those losses was relevant to show Copple's specific intent to defraud. We agree, with the qualification that district judges should exercise their wise discretion in imposing limits on such testimony. The following discussion is a case in point.

Copple's defense was that he had simply made a bad business decision when he, as trustee, had relied on the advice of experts to invest in rare coins. Yet the funeral directors had to pay for their losses out of their own pockets, while he refused to part with any of the luxuries he had purchased with the Mechem "pre-need" funds. The evidence of Copple's refusal to part with the property under such circumstances was, we believe, evidence of Copple's fraudulent intent. It tended to show that Copple intended to convert the Mechem "pre-need" money to his own personal use, something he had no right to do.

In addition, the testimony about the funeral directors' losses also corroborated the testimony of insurance agent James Domino, who said that he had never issued a surety bond to Mechem despite Copple's requests. Yet a few of the funeral directors testified that they had received a letter on Domino's stationary stating that the surety bond had been issued, and that letter was offered into evidence. Thus the fact that the funeral directors had to pay the losses out of their own pockets corroborated Domino's testimony that no such bond was issued and showed Copple's intent to mislead the funeral directors with the letter. Since the evidence about the extensive losses suffered by the funeral directors was relevant to show Copple's failure to repay and his intent to defraud, the evidence was admissible under the low threshold of Rule 401.

When the district court ruled on the admissibility of the testimony about the funeral directors' losses, however, its ruling encouraged the government to introduce a wide range of victim impact testimony in addition to the testimony about the size of the losses. Some of the victim impact testimony went beyond anything that was reasonable to prove Copple's specific intent to defraud.

A number of the funeral directors testified that the money they had used to pay back the losses came from money they had saved for their children's college educations. Others testified that paying back the money had affected their health, or had been taken from savings dedicated to other special purposes. For example, in response to the question of what effect the loss of money had on his business, Frank Mihalcik answered: "[w]ell, the situation.... has affected my health. I have lost over 60 pounds, and I am currently under a doctor's care." Terry Starr testified that he had to use every bit of personal savings he had in order to retrust the lost money. He then stated that, in order to obtain the money, he and his wife were forced to break a contract to purchase a home and to use the down payment money to retrust the money they had lost.

Testimony such as this had either no, or very little, probative value and was unfairly prejudicial. We believe that it was irrelevant either for the purposes of proving that Copple had failed to make up the loss to the funeral directors or for any other reason. Even if there had been some marginal relevance to the testimony about the particular personal or professional impact the losses had on the funeral directors, its principal effect, by far, was to highlight the personal tragedy they had suffered as victims of the scheme. The testimony was designed to generate feelings of sympathy for the victims and outrage toward Copple for reasons not relevant to the charges Copple faced. It arguably created a significant risk that the jury would be swayed to convict Copple as a way of compensating these victims wholly without regard to evidence of Copple's guilt.

In short, we believe that the probative value of the victim impact testimony was outweighed by unfair prejudice, and that such testimony should have been excluded under Federal Rule of Evidence 403.

2. Harmless error.

Although the district court abused its discretion by allowing all of the victim impact testimony into evidence, we need not reverse if that error was harmless. Trial error is harmless if it is highly probable that the error did not affect the judgment. United States v. Simon, 995 F.2d 1236, 1244 (3d Cir.1993). High probability exists if the court has a "sure conviction that the error did not prejudice the defendant." Id. at 1244 (internal quotations omitted). There is no need to disprove "every reasonable possibility of prejudice." Id. at 1244 (internal quotations omitted). We believe that the error of admitting the victim impact statements was harmless because of the overwhelming evidence of both the scheme to defraud and Copple's specific intent.

First, Copple's claim that he had bought the rare coins in order to get a higher return for Mechem was refuted by the bankruptcy trustee's testimony that Copple had sold $877,000 worth of the coins just before declaring bankruptcy and had made the checks from the coin companies payable to himself--not Mechem. Although $450,000 of that money was eventually transferred to Mechem, Copple could not remember where the other $427,000 of the proceeds from the sale went.

Second, Copple spent immense amounts of money for personal use while he was drawing money from the Mechem account. Copple's personal spending during the three-year life of Mechem included the following purchases:

$228,000 Home improvement, Sesler Builders 196,334 Furniture, Russell's Country Manor 67,694 Home improvement, Kitchens by Meade 70,000 Home improvement, architects and contractors. 62,081 Jewelry, Les Crago 70,279 Jewelry, Fortunoff's 398,000 Jewelry, Neiman-Marcus 48,712 Sable coat 11,000 Other fur coats 480,000 Gifts for family members 230,000 Payments to other Copple-owned businesses 61,000 Automobiles 6,000 Country club fees 3,000 Gambling, Caesar's Palace At the time of the bankruptcy, Copple had also just put a $110,000 deposit down on a $450,000 diamond from Neiman-Marcus that weighed 38.33 carats. Nearly all of the money for these expenditures came from money in the "pre-need" accounts that Copple had transferred to himself.

Third, the evidence was overwhelming that Copple had prepared wholly fictitious reports about how Mechem was investing the money, about the interest earned on the investments, and about fidelity or surety bonding. Particularly incriminating is the false letter Copple caused to be sent from Mechem to its investors six months after it was incorporated stating that Mechem's "investments have been made in insurance companies, annuities, T-bills, long-term municipal bond funds, short-term CD's and money markets," and that Mechem's "performance has reflected our excellent investment posture for the last fifteen years." Also highly probative of both the scheme to defraud and Copple's fraudulent intent was the evidence that funeral directors were sent fabricated quarterly reports showing the interest that had accrued on the trust investments, and the evidence that Copple had ordered a salesman to alter a general liability policy to make it look like a fidelity or surety bond.

We believe that all of this evidence overwhelmingly indicates that Copple knowingly devised or participated in a scheme to defraud and did so with the specific intent to defraud the funeral directors and others who had invested in Mechem. We are satisfied that it was highly probable Copple would have been convicted for violating the mail fraud statute even if the victim impact testimony had been excluded. For these reasons, we hold that the error was harmless.17

III. SENTENCING ISSUES

A. The Upward Departure

At the sentencing hearing the district court stated that "an upward departure of two levels is appropriate based on the large number of victims and the amount of monetary loss involved as provided in section 2F1.1(b)(2)(B) of the Guidelines." Copple contends that this departure was improper because the Guidelines adequately take into consideration both the amount of money involved in the offense (in Sec. 2F1.1(b)(1)) and the number of victims of the fraud (in Sec. 2F1.1(b)(2)(B)).18 We must therefore consider whether Copple's crime fell outside the "heartland" of cases which are described in the Guidelines because either the amount of monetary loss or the number of victims swindled was so high that the guidelines which "linguistically apply" significantly understate Copple's culpability. See U.S.S.G. Ch. 1 Pt. A(4)(b) (policy statement).19 We believe that neither ground articulated by the district court is a valid basis for departure in this case.

To begin with, the amount of monetary loss falls well within the range of monetary loss explicitly considered in the Guidelines. Section 2F1.1(b)(1) gives the district court authority to increase the offense level incrementally according to the amount of loss. See U.S.S.G. Sec. 2F1.1(b)(1)(A)-(S). The $4.9 million loss in this case fits squarely within the loss table's range of $2000 to $80 million. See U.S.S.G. Sec. 2F1.1(b)(1)(N). This is not a case in which the amount of loss exceeds the highest amount accounted for in the loss table. Since the Guidelines appear to take into account adequately the amount of monetary loss in this case, this factor was an invalid basis for departure. See United States v. Davidson, 984 F.2d 651, 654 (5th Cir.1993) (upward departure on the basis of amount of money in a fraud impermissible where $800,000 fraud fell within range of former Guidelines which had a ceiling of $5 million).

We also believe that the Guidelines adequately calibrate the offense level to take account of the number of victims in this case. The loss fell directly on thirty-one funeral directors, who had to make good for their "pre-need" customers by agreeing to retrust the money out of their own pockets and to perform "pre-need" funerals without compensation. Although thirty-one victims is far more than necessary to trigger the two level enhancement pursuant to Sec. 2F1.1(b)(2)(B) for conducting a scheme to defraud more than one victim, thirty-one is not so extraordinarily large a number in a case of this type that it falls outside the heartland of the fraud provisions. See, e.g., United States v. Boula, 932 F.2d 651, 656-57 (7th Cir.1991) (upward departure due to the number of victims in case involving 3000 victims was not permissible because such a scheme was not outside the heartland of the number of victims contemplated by the Guidelines). Cf. United States v. Benskin, 926 F.2d 562, 564-65 (6th Cir.1991) (relying on the pre-November 1989 Guidelines, the court upheld an upward departure involving 600 victims); Davidson, 984 F.2d at 654 (extraordinarily large number of victims required for upward departure). In our view, the Guidelines, through both the loss table in Sec. 2F1.1(b)(1) and Sec. 2F1.1(b)(2)(B), adequately account for frauds involving the number of victims in this case.

Schemes involving large numbers of victims raise two principal considerations. First, schemes involving large numbers of victims tend to impose much greater losses. Second, schemes involving numerous victims tend to be more systematic, and losses actually caused by such schemes may underrepresent the amount of losses the defendant intended. See U.S.S.G. Sec. 2F1.1, background (justifying the enhancement for schemes involving more than one victim in Sec. 2F1.1(b)(2)(B)). In this case, the first consideration seems to be adequately taken into account by the loss table in Sec. 2F1.1(b)(1), for Copple's offense level was increased to reflect the total loss in increments that the Sentencing Commission deemed appropriate. Similarly, the second consideration generally is adequately taken into account by giving Copple a one-time enhancement pursuant to Sec. 2F1.1(b)(2)(B) for involving more than one victim. Although there may be cases in which the loss table in Sec. 2F1.1(b)(1) disproportionately underrepresents the amount of intended loss that does not appear to be the case here.

Moreover, Ponzi schemes, major securities frauds and other similar frauds involving thousands of victims have been around since the early twentieth century, and the Commission was certainly aware of them when drafting the Guidelines. Indeed such awareness seems implicit in the $80 million ceiling in the loss table. Frauds of $80 million will almost certainly involve numbers of victims far in excess of the thirty-one involved here. Given the structure of the Guidelines and the interplay between the loss table and the "more than one victim" enhancement, the fraud in this case does not appear to involve a number of victims that is outside the "heartland," and hence a departure was impermissible.20 Accordingly the sentence must be vacated and reconsidered.

Nevertheless, our holding does not preclude the district court from making an upward departure. At the time it granted the departure in this case, the district court did not take cognizance of two bases for increasing the offense level suggested in the presentence report.21 Although the district court implicitly rejected these grounds for departure by imposing the departure in the way it did, nothing prevents the district court from reconsidering them on remand.22

B. The Restitution Order

We must also vacate the portion of the judgment of sentence ordering Copple to pay $4,257,940.45 in restitution.23 The Victim and Witness Protection Act ("VWPA") allows the district court to order restitution as part of a sentence. 18 U.S.C. Sec . 3663(a)(1). Section 3664(a) of the VWPA provides the procedures the district court must employ in ordering the restitution:

The court, in determining whether to order restitution under section 3663 of this title and the amount of such restitution, shall consider the amount of the loss sustained by any victim as a result of the offense, the financial resources of the defendant, the financial needs and earning ability of the defendant and the defendant's dependents, and such other factors as the court deems appropriate.

18 U.S.C. Sec . 3664(a) (emphasis supplied). This Court has required the district courts " 'to make specific findings as to the factual issues that are relevant to the application of the restitution provisions of the VWPA.' " United States v. Logar, 975 F.2d 958, 961 (3d Cir.1992) (quoting United States v. Palma, 760 F.2d 475, 480 (3d Cir.1985)).

In Logar, we identified the factual findings the district court must make before ordering restitution: 1) the amount of loss, 2) the defendant's ability to pay and the financial need of the defendant and the defendant's dependents, and 3) the relationship between the restitution imposed and the loss caused by the defendant's conduct. 975 F.2d at 961. We also held that, notwithstanding estimates of loss in a presentence report, the district judge must point to the evidence in the record supporting the calculation of loss to the victims. Id. at 961-62. The district court in this case failed to follow these procedures.

At the sentencing hearing, the district court ordered restitution on the following basis:

We accept as factual the report in the presentence report concerning money due victims, and this amount of money, of course, is difficult to ascertain without having a hearing that might go on for days, but we do accept as fact those findings in the presentence report and order that the defendant shall make restitution in the amount of $4,257,940.45 through the Trustee of the United States Bankruptcy Court for the Western District of Pennsylvania who will make the distribution to the victims listed in the indictment. It is further ordered that the defendant shall pay to the United States a fine of $100,000 and the costs of prosecution. This fine shall be subject to the rights of creditors.

(emphasis in the original). The district court also stated that the bankruptcy court should monitor the restitution ("we feel that the bankruptcy court is better able than this Court to determine who owes what to whom").

The district court made no findings about Copple's ability to pay the restitution. The court also made no findings about Copple's financial needs, or his ability to support himself and his wife and two children (after his release from jail). We will therefore remand for the district court to make the factual findings necessary to support such order of restitution as it may make. We note in this regard that the district court is not at liberty to delegate its role with respect to restitution to the bankruptcy court or the bankruptcy trustee.

IV. CONCLUSION

The judgment of the district court with respect to the conviction will be affirmed. However, the judgment with respect to the sentence will be reversed and the case remanded for further proceedings consistent with this opinion.

1 Soon after the company was formed, both Scott's and Orlando's relationship with Copple deteriorated. Copple presented Scott and Orlando with bills for start-up costs. Scott countered by suing Copple. Orlando's response was to give up his stock in the company in return for a fee for consulting services. As a result of these disputes, Scott and Orlando ceased to have meaningful responsibilities at Mechem. In 1987, when an accountant reviewed the Mechem corporate records, the officers were listed as Copple, his brother Charles Copple, and W. Gustave McGeorge. The records made no mention of Scott or Orlando having any formal association with the company

2 Copple did not spend his entire time operating Mechem. He also acted as a financial advisor to a number of individuals. The three relevant to this case include: Audrey Garfield Woo, a widow who gave Copple $55,000 that she had inherited from her husband to invest for her grandchildren's education, Patrick Mastrian, who gave Copple $50,000, and Virginia Sczepanski, who gave him $25,000. All of this money was commingled with the Mechem "pre-need" trust money

3 The Pennsylvania Board of Funeral Directors and the Pennsylvania Attorney General required the funeral directors, as a condition of keeping their licenses, to replace the money that had been taken from the "pre-need" trusts

4 At trial, the district court granted a motion to dismiss Mechem from the indictment because the corporation was defunct

5 During the three years he ran Mechem, Copple did not file any personal tax returns. At trial, the IRS estimated Copple's total tax liability to be $753,323; minus withholdings and credits, the IRS claimed that Copple owed taxes of $665,859 for the three years

6 Our review of this issue is plenary since it requires the interpretation of a federal statute. See Air Courier Conference/Int'l Comm. v. U.S. Postal Serv., 959 F.2d 1213, 1217 (3d Cir.1992)

7 Juror Number 44, Lisa Hayes stated that her father-in-law had been audited. Juror Number 75, John Moore responded that his father had been audited. Juror Number 43, Jim Harry, stated that a company for which he had worked had gotten into trouble with the IRS and had gone out of business. As a result, Harry had lost his job. He said that he could not be impartial and he was dismissed

8 The explanation for this provision apparently is that a juror who has been audited might either be 1) antagonistic to the government or 2) fear retaliation from the government and convict the defendant to curry favor

9 In United States v. Huguenin, 950 F.2d 23, 30 (1st Cir.1991), for example, the IRS took nineteen days to complete a manual search of the jurors' tax histories. All of the jurors in that case had filed their tax returns at a single IRS Regional Service Center for all of the years they paid taxes. Had any of the jurors lived outside the region at any other time they were paying taxes, the search time would have been considerably longer

10 Moreover, it is not clear that all the checks can be completed within a fixed window between identification of the panel and the time for its appearance in court. Further complicating the administration of the jury panel is that when the period between identification and appearance is increased, continuances become more likely

11 Before Congress enacted Sec. 6103(h)(5), the procedure for inquiring into tax records of potential jurors was governed by Treasury Regulation Sec. 301.6103(a)-1(h). This regulation authorized the government to inquire of the IRS whether a prospective juror had been investigated by the IRS. Criminal defendants, however, had no similar right to inquire. When deciding whether to incorporate this regulation into the Tax Reform Act of 1976, Congress decided to allow such disclosures as long as the taxpayer had the same access to the information. Spine, 945 F.2d at 147 (quoting Senate Committee on Finance, 94th Cong., 2d Sess., June 4, 1976, Press Release (1976) reprinted in Tax Magmt. (BNA), Primary Sources, Series II, Sec. 6103 (1976), at 40 (Nov. 1, 1977))

12 Nearly every case considering Sec. 6103(h)(5) has cited Spine favorably and has held that a limitation on the time period covered by the IRS investigation is not reversible error. Nearly all of those cases, however, have analyzed limitations on the time period covered by a Sec. 6103(h)(5) investigation under the rubric of harmless error. See, e.g., United States v. Axmear, 964 F.2d 792, 793 (8th Cir.1992), cert. denied, --- U.S. ----, 113 S.Ct. 963, 122 L.Ed.2d 120 (1993); United States v. Droge, 961 F.2d 1030, 1034 (2d Cir.), cert. denied, --- U.S. ----, 113 S.Ct. 609, 121 L.Ed.2d 544 (1992); Huguenin, 950 F.2d at 29-30; United States v. Schandl, 947 F.2d 462, 469 (11th Cir.1991), cert. denied, --- U.S. ----, 112 S.Ct. 2946, 119 L.Ed.2d 569 (1992); United States v. Hardy, 941 F.2d 893, 896 (9th Cir.1991); United States v. Masat, 896 F.2d 88, 95 (5th Cir.1990), cert. denied, --- U.S. ----, 113 S.Ct. 108, 121 L.Ed.2d 66 (1992). But see Nielsen, 1 F.3d at 858 (where there is substantial disclosure of the information about the persons audited or investigated, voir dire supplements the information, and there is no palpable suggestion of prejudice, Sec. 6103(h)(5) was not violated)

While the harmless error approach has a certain appeal, we believe that its use to cure the perceived deficiency in the statutory language is highly questionable and ultimately flawed. Harmless error analysis is typically a retrospective analysis, one that requires the reviewing court to make a considered judgment about the impact an error had on a particular conviction. See Kotteakos v. United States, 328 U.S. 750, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946) (discussing the theory behind harmless error). Courts that have used harmless error analysis to cure the purported "error" of limiting a tax record investigation to six years have used the doctrine in quite a different sense. Rather than using harmless error to consider the particular circumstances of a given case, these courts have used it to prescribe a course of action a district court may take to insulate its "noncompliance" with the statute from challenge on appeal. E.g., Droge, 961 F.2d at 1032-35. When used this way, harmless error analysis becomes a surrogate for interpretation of the statutory requirements in the first instance, but this is not the function of the harmless error doctrine.

13 As has been mentioned, the records are computerized for the years following 1986 and the IRS can retrieve the information without too much difficulty. The likelihood that a juror who had been audited before 1986 would be biased and would refuse to mention that bias on voir dire seems quite small, so that the costs of a manual investigation prior to 1986 might not be justifiable. At all events, as 1986 recedes, the number of jurors who potentially will have had an audit or other investigation which is not recorded on computer will diminish and, eventually, the problem will disappear

14 Although some cases might be read to require that the responses of the jurors on voir dire should be verified, see, e.g., Huguenin, 950 F.2d at 29, and Lussier, 929 F.2d at 30, we believe there is no need to verify the answers given by the jurors on voir dire because jurors are presumed to respond truthfully to such questions, see Masat, 896 F.2d at 95. Although some prospective jurors might be reluctant to answer such questions truthfully, "veniremen are often asked sensitive and potentially embarrassing questions, including inquiries into their involvement in criminal activity or the involvement of family or friends in criminal activity, their religious or philosophical beliefs, and other matters of a personal nature." Id. We see no reason to depart from the presumption that the potential jurors will respond truthfully

15 We do not believe that the fact that the IRS missed Borczon necessarily means that the computerized search was inadequate

16 This policy extends in the other direction as well, and allows the defendant to introduce testimony of collateral transactions that tend to negate the requisite intent. See Id

17 Copple has also argued that the cumulative effect of six trial errors in addition to the admission of the victim impact testimony denied him a fair trial. A new trial is required on this basis only when " 'the[ ] errors, when combined, so infected the jury's deliberations that they had a substantial influence on the outcome of the trial.' " United States v. Thornton, 1 F.3d 149, 156 (3d Cir.), cert. denied, --- U.S. ----, 114 S.Ct. 483, 126 L.Ed.2d 433 (1993) (quoting United States v. Hill, 976 F.2d 132, 145 (3d Cir.1992)). The six asserted errors include a prosecutor's allegedly improper remark during the opening statement, three allegedly improper statements by witnesses, the admission of testimony summarizing the testimony of other witnesses, and the showing of a video tape of Copple's home. We have examined the six asserted grounds for error and believe that they were at most nothing more than minor aberrations in a long trial, and did not consist of cumulative evidence indicating a proceeding dominated by passion and prejudice. To the extent any of the incidents constituted error, we believe that in light of the overwhelming evidence of guilt, the errors were harmless and did not deprive Copple of a fundamentally fair trial

18 Copple argues alternatively that the district court did not really depart upwards but instead gave a four level increase pursuant to Sec. 2F1.1(b)(2) because the offense involved both more than minimal planning and more than one victim. Copple points out that although at sentencing the district court stated that it was departing upward, the court checked the box in the judgment that stated "[t]he sentence is within the guidelines range." Copple correctly argues that Sec. 2F1.1(b)(2) allows for a two level increase if the offense involves either more than minimal planning or more than one victim, and that a district court may not impose a four level increase under Sec. 2F1.1(b)(2) if the offense involved both more than minimal planning and a scheme to defraud more than one victim. United States v. Astorri, 923 F.2d 1052, 1057 (3d Cir.1991). Thus to the extent that the district court did not depart upwards, but rather gave a four level enhancement because the crime involved both more than minimal planning and more than one victim, we hold alternatively that the increase was improper on such a basis

19 Our review of this issue is plenary. See United States v. Kikumura, 918 F.2d 1084, 1098, 1110 (3d Cir.1990) (review of upward departure is plenary on whether the increase was permissible and for abuse of discretion on whether the degree of the increase was reasonable)

20 Indeed two circuits, the Second and the Eleventh, seem to have imposed a categorical bar on departures based on the number of people involved. See United States v. Mandel, 991 F.2d 55, 58 (2d Cir.1993); Alpert, 989 F.2d at 459

21 The presentence report stated:

92 Approximately thirty additional directors and their clients were excluded because the losses were less than $30,000 per funeral director. Also, the mail fraud perpetrated by the defendant involved not only Mechem Financial, Inc., but also Mechem Financial of Ohio as evidenced by the results of the civil investigation conducted by the Ohio Attorney General's Office. If the losses suffered by that affiliate were added to the fraud that occurred in this district, the total loss to the funeral directors would exceed $11 million, and under the provisions of Sec. 2F1.1(b)(1), the offense level would increase by an additional 2 levels

93 Section 4A1.3 provides that the court may consider imposing a sentence departing from the otherwise applicable guideline range if reliable information indicates that the criminal history does not adequately reflect the seriousness of the defendant's past criminal conduct. A factor that may be considered is prior similar adult criminal conduct not resulting in a criminal conviction. (Section 4A1.3(e)). Paragraph 59 of this report deals with fraudulent transactions involving over $100,000 for which the defendant was not criminally prosecuted. This conduct began in December 1984 but was not discovered for several years. The victim was reimbursed by the defendant's former employer, and criminal charges were not pursued. This conduct is similar to what occurred in the cases of Patrick Mastrian, Audrey Garfield Woo, and Virginia Sczepanski

Of course we express no opinion on the appropriateness of these bases for increasing the offense level, preferring to leave that determination in the first instance to the district court.

22 On remand, a district court can consider matters not explicitly or implicitly part of the decision in the appellate court. See United States v. Uccio, 940 F.2d 753, 758-59 (2d Cir.1991). Cf., United States v. Kikumura, 947 F.2d 72, 76 (3d Cir.1991) (law of the case prevented the district court from reconsidering issue explicitly decided by the appellate court). As long as the district court gives Copple adequate notice that it might reconsider these bases for departure and affords an adequate hearing to allow the parties an opportunity to elaborate on their position, see Burns v. United States, 501 U.S. 129, 137-39, 111 S.Ct. 2182, 2187, 115 L.Ed.2d 123 (1991), the court may reconsider them

23 Our review of the restitution issue is plenary. See Air Courier Conference, 959 F.2d at 1217