John R. Parsons was convicted of one count of conspiracy to commit bank fraud, 18 U.S.C. §§ 371, 1344, and 13 counts of bank fraud, 18 U.S.C. § 1344, and now appeals from his conviction and sentence. We set forth a brief summary of the underlying events. Since Parsons challenges the sufficiency of the evidence, we describe the facts in the light most favorable to the verdict. United States v. Bergodere, 40 F.3d 512, 518 (1st Cir.1994), cert. denied, 514 U.S. 1055, 115 S. Ct. 1439, 131 L. Ed. 2d 318 (1995).

In spring 1987, Parsons and Robert Hakala agreed to build an office building on a lot they owned on the Leominster/Lancaster town line in Massachusetts. Parsons planned to handle financial and leasing matters; Hakala, who ran Hakala Construction Corp., was slated to do the construction. In May 1987, Parsons and Hakala formed Quest Realty Trust as a vehicle to hold the land and own the planned building.

Financing was sought from First Service Bank for Savings, a federally insured bank in Leominster headed by Clary William Wester. Between May and September 1987, Quest Realty received interim short-term loans from First Service so that it could acquire the land and undertake the initial steps, pending a full-scale construction loan. In August 1987, Hakala sold his interest in Quest Realty to Parsons but continued as the builder.

On September 30, 1987, the bank agreed to make a full-scale loan to Quest Realty, now fully controlled by Parsons, for $4.8 million. Some of this money was disbursed at once; but most was held back and payments were made over the succeeding months. These subsequent payments were partly based upon "requisitions" from Hakala attesting that work had been done and were partly ad hoc payments directed by Wester. In the government's view, the September 30th loan was procured by Parsons through fraud, and much of the money was diverted by Parsons to improper uses.

In February 1996, a grand jury charged Parsons with the single count of conspiracy and the 13 counts of bank fraud already mentioned; there were other counts in the indictment, but they were later dismissed and so are irrelevant here. Wester, an alleged co-conspirator in count 1, pled guilty in October 1996 to a conspiracy charge substantially identical to the conspiracy charged here. He had earlier been convicted of unrelated frauds in connection with his operation of First Service. United States v. Wester, 90 F.3d 592 (1st Cir.1996). Parsons was tried in a two-week jury trial in November 1996.

At trial the government sought to prove two different schemes. The first scheme (counts 1-8), in which Wester was supposedly a participant, involved the alleged diversion by Parsons of seven specific amounts, totaling about $1.4 million, from the bank's loan reserved for Quest Realty's use in constructing the building. According to the government, the diversions occurred so that Parsons could make either personal expenditures (e.g., purchase of a Florida condo) or other investments not authorized by the loan documents (e.g., in a different development project not approved by the bank).

In the second alleged scheme (counts 9-14), the government sought to show that Parsons, acting alone, had diverted six additional sums, totaling just over $272,000. The government charged that after the bank had paid money over to Quest Realty to satisfy Hakala's requisitions, Parsons had in six cases short-changed Hakala and withheld the balance in Quest Realty for his own benefit. This, said the government, was contrary to the commitment in the loan agreement that the bank's money would be paid out by Quest Realty in accordance with the requisitions.

On November 14, 1996, the jury convicted Parsons on all 14 counts. In March 1997, Parsons was sentenced to 37 months' imprisonment, fined $70,000 and ordered to pay restitution in the amount of the diversions (just over $1.6 million). Parsons's appeal followed, challenging the sufficiency of the evidence on 10 of the 14 counts. Parsons also alleged several trial errors and attacked his sentence.

On this appeal, Parsons's sufficiency-of-the-evidence claims are not backed by detailed analysis of the record evidence on individual counts. Instead, after some limited background on each count in the fact section of his brief, Parsons's argument on the alleged insufficiency of the evidence consists--in total--of two pages on counts 2-8 and one page on counts 9-10. Effectively, this amounts to a few sentences each for nine different, very complex transactions. Parsons's approach frustrates any detailed examination by us and largely forecloses his claims. U.S. Healthcare, Inc. v. Healthsource, Inc., 986 F.2d 589, 595-97 (1st Cir.1993).

Nevertheless, there are several themes in Parsons's argument, primarily addressed to counts 2-8, that we will address briefly. The first theme is Parsons's claim--seemingly accurate--that there was very little direct evidence of a conspiracy between Parsons and Wester to defraud the bank. Wester approved certain of the payments charged in the indictment as diversions to Parsons, but the government did not have solid evidence as to any benefit to Wester, promised or received. Needless to say, this gap in proof makes it less likely that Wester conspired.

But the government did prove that Parsons and Wester had an ongoing relationship embracing many loans; both had various ventures outside the bank; and there was some admittedly vague evidence of business associations between them extrinsic to the bank. Since Wester did approve improper diversions and Parsons took the money, it was not irrational for a jury--given the relationship between them--to infer that they had agreed to the diversions. The jury was not required to draw the inference but was entitled to do so. Glasser v. United States, 315 U.S. 60, 80, 62 S. Ct. 457, 469-70, 86 L. Ed. 680 (1942).

A second related theme in Parsons's argument is that as to those diversions known to Wester, Wester's "approval" of them insulates Parsons, since Wester was in charge of the bank. The "permission" of a co-conspirator is not much of a defense. See United States v. Sheahan, 31 F.3d 595, 600 (8th Cir.1994). Further, the diversions were not debatable expenditures but patently contrary to the Parsons's obligations under the loan. Lastly, Parsons's own attempts to conceal his misconduct make nonsense of any suggestion that he believed that Wester had somehow modified Parsons's obligations to the bank.

The third and fourth themes in Parsons's argument are that Parsons was unaware of his obligations under the loan documents he signed and that any diversions to other uses were inadvertent. These are improbable claims, in light of Parsons's numerous diversions, their size, and his various efforts to conceal his behavior and to shift blame to others (e.g., his accountant). In any event, these evidentiary arguments are not seriously developed through discussions of the evidence, and need not be pursued further, U.S. Healthcare, 986 F.2d at 595-97, with one exception.

Only with respect to count 9 does Parsons's brief offer anything close to a significant attack on the evidence. Count 9 charged that Parsons diverted $50,000 from a May 1987 loan to Quest Realty intended to reimburse Hakala Corp. for construction expenses incurred, and that in papers connected to the September 30th construction loan, Parsons represented that all money previously distributed by the bank had gone to pay construction expenses. One element of Parsons's argument is that Hakala was in charge of the Quest Realty checkbook in May 1987 and therefore necessarily approved the diversion. This is no answer because Parsons benefitted from the diversion and later represented to the bank that the money had been used properly.

Parsons's strongest point is his argument that $25,000 of the $50,000 allegedly diverted was indeed paid by Hakala to Parsons out of the Quest account, but for construction plans and an architectural model. Whether this $25,000 was within the conditions of the loan was the subject of conflicting evidence and inferences at the trial. After studying the record, we think that the evidence was barely sufficient as to this $25,000. The conviction on Count 9 is supported, in any event, by the diversion of the other $25,000, for which no adequate explanation has been provided.

Parsons makes a different evidentiary attack on his count 10 conviction. Here, we are given a total of three sentences addressing different claims of error, without citation to the record. To the extent possible, we have considered the arguments on the merits and find them unpersuasive. It is worth noting that neither of the convictions on counts 9 or 10 appear to affect the sentence of imprisonment, although they do add $50 each in special assessments. U.S.S.G. §§ 2F1.1, 5E4.3 (1987).

Parsons's counsel are competent lawyers and possibly intended their short and sporadic attacks on the sufficiency of the evidence merely to awaken some disquiet in the court, perhaps as a backdrop for technical claims of error that are more elaborately developed. This may be sound strategy. But future litigants should understand that underdeveloped or unsupported attacks on the government's evidence in a complex case are almost hopeless. It is counsel's job on appeal to mine the record and prove the alleged error, not to offer suggestive hints and leave the rest of the work to a busy court. United States v. Zannino, 895 F.2d 1, 17 (1st Cir.1990).

Parsons's next argument, after his challenge to the sufficiency of the evidence, attacks the district judge's decision to exclude proffered testimony. On the last day of trial, the defense called Richard Clark, an attorney who had worked for both Parsons and First Service. Clark had helped Parsons purchase a Florida condominium for Parsons's personal use, using funds from First Service. The government charged in count 8 that the funds were diverted from the Quest Realty construction loans.

According to a proffer, Clark would have testified, if allowed to do so, (first) that Wester had told Clark that the funds for the condo were coming from a line of credit at the bank available to Parsons, and (second) that Clark had then so advised Parsons. The government objected that this was hearsay, that it was intended as backdoor proof of the alleged line of credit, and that Parsons had conspicuously failed to call Wester to confirm the alleged line of credit. The district court heard argument, took a proffer, and then excluded the testimony in question, although Clark was allowed to testify on other aspects of the condo purchase.

In the calm of appellate review, it is clear to us that Clark's second statement--what he said to Parsons--was not hearsay if offered only to show Parsons's belief that the money came from a line of credit apart from the construction loan. Fed.R.Evid. 801(c). (Clark's first statement--what Wester told Clark--would still be hearsay as to the existence of a line of credit and would not itself show Parsons's state of mind). And Parsons did say to the district court several times that he was offering the evidence to show Parsons's state of mind.

No doubt the government believed that Parsons was trying through the back door to show that he had a separate line of credit. But the normal answer to that concern is a limiting instruction, United States v. Bartelho, 129 F.3d 663, 668-69 (1st Cir.1997), unless it can be shown that the benefit of the testimony is substantially outweighed by the risk of jury confusion. Fed.R.Evid. 403. The latter claim would itself be a stretch and, in any case, was not made at trial.

The government's last ditch argument is that if error, the exclusion was harmless. Here we agree. At first blush, the excluded testimony, although brief, appears highly relevant. But a moment's reflection suggests that it would not have been very persuasive as to Parsons's state of mind, unless the jury also believed that Parsons in fact had a line of credit or good reason to believe he did--independent of Clark's statement. Substantial lines of credit do not materialize out of thin air: there had to be some groundwork laid to secure such credit.

Parsons did offer evidence that he had other loans from the bank. But we have examined what is cited in the record; and the evidence that Parsons had an outstanding line of credit, in a large amount, at the time of the diversion, is almost invisible. It is not surprising that the jury was unimpressed. Parsons chose not to testify and did not call Wester, thus depriving himself of the two witnesses best situated to say that such a line of credit existed and was used for the loan.

As to Clark, he was far from a reliable witness for the defense on account of other links with Parsons that tended to make Clark's testimony suspect. In addition, Parsons's accountant testified that when he later asked Parsons about the use of Quest Realty loan proceeds for the condo purchase, Parsons made no mention of any line of credit but provided a quite different explanation for use of the loan funds. All said and done, we deem the exclusion error, but harmless.

We turn next to two claims of error based upon refusals to give jury instructions sought by Parsons. The first refused instruction was meant to follow after an instruction by the court advising that the complicity of bank insiders and their knowledge of a fraud is not a defense to bank fraud and does not absolve the defendant. Parsons does not object to this statement but to the refusal to append to it the following:

On the other hand, the mere fact that a First Service Bank officer or employee granted or facilitated loans does not, in and of itself, constitute bank fraud. If you find that First Service Bank officers or employees approved of the defendant's use of the proceeds in question without acting fraudulently, you may consider their approval in determining whether or not the defendant defrauded First Service Bank.

Except for confusing innuendo ("facilitate," "in and of itself"), the first sentence in the rejected instruction is a pointless banality: making loans is what banks do. The second sentence is misleading: a non-fraudulent approval by a bank officer or employee could easily be based on inadequate disclosure by Parsons. Despite the "may consider" qualification, the language tended to equate approval with exculpation, was misleading, and was properly refused.

Parsons's second refused instruction stated, in relevant part, that in assessing Parsons's state of mind the jury could consider "whether Mr. Parsons ultimately paid the contractor for the work performed on the building." The instruction was apparently aimed at Parsons's belated payment--after Hakala had discovered the scheme--of loan proceeds due to Hakala for work on the building but withheld and diverted by Parsons. It is enough of an answer to say that the district court properly charged the jury on intent to defraud; and for obvious reasons the court is not required to frame an instruction for each piece of evidence that one side or the other would like to have emphasized. United States v. Camuti, 78 F.3d 738, 744 (1st Cir.1996).

Finally, Parsons says that the district court erred in calculating his prison sentence and in ordering restitution; the arguments involve a common premise--that Parsons should have been given credit for a settlement that he reached with the FDIC as receiver for First Service. We begin with the sentencing issue, first describing the district court's calculation of the offense level that determined the sentencing range.

Parsons was sentenced under section 2F1.1 of the 1987 edition of the guidelines covering crimes of fraud.1 It provides a base offense level of 6 with additional levels added (pursuant to a table) to reflect the amount of "loss" inflicted by the crime. In this case, the probation officer reported that the loss to First Service was a figure just over $3.2 million, reflecting Quest Realty's unpaid loan balance less the amount recovered on the sale of collateral.

Favoring Parsons, the district court instead chose to use $1.6 million as the "loss," reflecting the amounts that Parsons had unlawfully diverted from the Quest Realty loan. Where justified, this is expressly permitted by the guidelines: "The offender's gross gain from committing the fraud is an alternative minimum estimate of the loss." U.S.S.G. § 2F1.1, commentary. At sentencing, Parsons accepted this lesser figure as a starting point but then asked for a further reduction on grounds of his settlement with the FDIC.

The district court refused the further reduction and instead added to the base level the additional 9 levels that the table prescribes for a $1.6 million loss. The court added 2 more levels for more than minimal planning and 2 more for obstruction of justice, adjustments not here challenged. U.S.S.G. §§ 2F1.1(b)(2)(A), 3C1.1. This produced a total offense level of 19, and Parsons was sentenced within the resulting range to 37 months' imprisonment, as well as to a fine and restitution.

On appeal, Parsons argues that the district court had to use loss rather than gain and that loss to the bank was zero because it settled its claims with Parsons. The settlement was reached in a civil suit between Parsons and the FDIC as receiver for the bank. Parsons agreed to transfer the deed on the Quest building to the FDIC, and each side gave up its own claims--the bank for all balances due on the Quest Realty loan as well as other loans made by First Service to Parsons (in all, exceeding $8 million) and Parsons for the bank's breach of its lease (made between Wester and Parsons) on the new building constructed for Quest Realty.

Parsons's argument begins with an application note describing "loss" for a fraudulently obtained loan as "the amount of the loan not repaid at the time the offense is discovered, reduced by the amount the lending institution has recovered (or can expect to recover) from any assets pledged to secure the loan." The application note was added in 1991, application note 7(b), added by Amend. 393, U.S.S.G.App. C, at 211 (1997), but may arguably be considered as clarification. He refers also to cases implementing this formula, including a few precedents saying that gain should not be used if loss is zero. United States v. Chatterji, 46 F.3d 1336, 1342 (4th Cir.1995); United States v. Andersen, 45 F.3d 217, 221 (7th Cir.1995).

Loss is a proxy for the seriousness of the offense. A loss of zero is presumptively wrong in this case, since it does not even remotely approximate Parsons's wrongdoing. Further parsing of the application-note formula Parsons quotes is unnecessary, because the fraud guideline commentary has always provided that where the intended loss is greater than the actual loss, the intended loss is to be used. U.S.S.G. § 2F1.1. On its face, Parsons's initial diversion of $1.6 million in loan proceeds to personal use is an intended loss.

And given that $1.6 million was intentionally diverted, the later settlement does not reduce Parsons's culpability. It was made well after the fraud was known to persons like Hakala who could report it. Nor does it help Parsons that the FDIC knew little about the criminal behavior when it made the settlement; a defrauder cannot purchase a shorter term by a belated return of the proceeds when the fraud is in the course of unraveling. United States v. Bennett, 37 F.3d 687, 695 (1st Cir.1994). In short, the guideline range was properly computed.

A related but more difficult question concerns restitution. The district court ordered Parsons to make restitution to the government of the same $1.6 million that the court had used as the loss for sentencing purposes. 18 U.S.C. § 3663 permits restitution to compensate a victim for losses. Parsons disputes the order on two different grounds: that it is precluded by the FDIC's civil settlement with Parsons, and that there is no adequate proof of any loss.

Restitution is not generally appropriate when it would represent double recovery by the victim. United States v. Manzer, 69 F.3d 222, 230 (8th Cir.1995).2 However, the burden of establishing this double payment is on the defendant. United States v. Mmahat, 106 F.3d 89, 98 (5th Cir.), cert. denied, --- U.S. ----, 118 S. Ct. 136, 139 L. Ed. 2d 84 (1997). In some cases, the nature of a settlement might show that the victim had been compensated for the loss. Here, however, we have no reason to conclude that the settlement compensated the bank for the $1.6 million loss reflecting the funds diverted from the Quest loan.

It is true that on one side of the equation, the bank's professed loss of all outstanding bank loans involving Parsons, less realized collateral, arguably had to encompass the smaller amount actually diverted. But the $1.45 million that the FDIC realized from the sale of the Quest building left unpaid $3.2 million in loans on the Quest project and about $4 million in loans on other projects. What the FDIC received, over and above the building, was Parsons's surrender of his own claim for $10 million for the bank's failure to honor the lease with Quest Realty.

The FDIC accepted the settlement based on this $10 million claim (and it was only a claim) in ignorance of the fact that Wester and Parsons, the co-signers of the lease, were criminal co-conspirators in defrauding the bank. The settlement may have been wise based on what the FDIC then knew. But a sentencing court, with greater knowledge of what Parsons's lease claim was worth, did not have to treat Parsons's surrender of the claim as meaningful compensation that would make restitution a double recovery.

True, the FDIC itself had no legal claim against Parsons after the settlement, since the FDIC gave him a general release. But this court has already held that a release by the victim does not preclude or cap restitution of losses as part of criminal sentencing in a case where there is no double recovery. United States v. Savoie, 985 F.2d 612, 619 (1st Cir.1993). In United States v. Coleman, 997 F.2d 1101, 1106-07 (1993), the Fifth Circuit took a contrary position and treated a settlement as a bar to a restitution award, but our own view has support in United States v. Cloud, 872 F.2d 846, 853-54 (9th Cir.1989), and United States v. Vetter, 895 F.2d 456, 459 (8th Cir.1990).3

Parsons's alternative objection is more interesting. He says that the diversion of loan proceeds to personal use was not the cause in fact of any loss to the bank. His theory is that his diversion to his own use was primarily of loan funds intended for the contractor, the contractor was later paid off by Parsons and completed the building, and therefore the loan collateral (i.e., the building) was unimpaired.

However, Parsons did not merely divert funds after the master loan; his fraud lay also in misrepresentations to secure the loan. And very large losses to the bank--well in excess of $1.6 million--followed from a loan that likely would not have been made if Parsons had told the truth about diversions of funds from the smaller interim loans by the bank. Contrary to Parsons's position, the link is not too attenuated to permit restitution. See United States v. Vaknin, 112 F.3d 579, 586-90 (1st Cir.1997).

The use of the $1.6 million figure was almost certainly generous to Parsons. At the same time, using funds actually diverted avoided various problems (e.g., Judge Gorton's equitable concern about later declines in property values and thus of the collateral) that would have attended use of the much larger loss figure proposed by the government. We think that the trial judge used common sense in making what is, in a complex set of transactions and possibilities, as much a judgment call as an exercise in mathematics.

Affirmed.