This pair of criminal appeals presents an important question concerning the meaning of "loss" in the Sentencing Guidelines. The Guidelines for property crimes such as larceny and fraud make the punishment vary with the loss to the victim of the crime. Secs. 2B1.1(b)(1) (larceny and closely related crimes such as embezzlement), 2F1.1(b)(1) (fraud). In the case of fraud, the loss need not be actual; it is enough if it is probable or intended. Application Note 7 to Sec. 2F1.1; United States v. Haddon, 927 F.2d 942, 951 (7th Cir.1991). In the case of attempted larceny, the Guidelines reach the same result by analogy to other attempts. Application Note 2 to Sec. 2B1.1. The dangerousness of an attempt--and hence the fitness and apt degree of punishment--is a function not only of the likelihood that it would have been completed but for interruption or mischance, but also of the gravity of the completed crime; so attempted murder is a more serious crime than attempted theft. Many fraudulent schemes are interrupted before they reach fruition. From a practical standpoint they are attempts, and their gravity depends in significant degree on the size of the loss that would have been inflicted had the scheme not been interrupted.

The defendants, a husband and wife, were sentenced to 30 and 15 months in prison, respectively, for conspiring to defraud, and defrauding, two federal agencies, in violation of various federal statutes. Mr. Schneider, an experienced building contractor, submitted a bid for $65,900 on a contract let by the General Services Administration for alterations in a federal building. In the papers accompanying the bid, he certified falsely that he had not been charged with a criminal offense within the past three years; in fact a charge of forgery was pending against him in an Illinois state court. He also submitted a fraudulent payment and performance bond. He was the low bidder and won the contract. But before performance began or any payment was made under the contract, the GSA canceled it on the ground of misrepresentation and awarded the job to another--and considerably higher--bidder.

Mrs. Schneider, who was associated with her husband in the building business, submitted to the Defense Department a bid for $76,500 (we are rounding off all dollar figures to the nearest $100), also for building alterations. The bid was accompanied by the same type of fraudulent payment and performance bond. The Department rejected the bid, being unsatisfied with the bond (though it didn't know at first that it was fraudulent), and later accepted a higher bid from another contractor.

Although the two different bids were submitted by two different Schneiders, they were acting in concert and each was convicted of both frauds. But they were treated differently in sentencing with respect to increase in offense level by reason of loss to the victim, that is, the government. In the case of Mr. Schneider, the district judge added together the amounts by which the prices at which the two contracts were ultimately let to other contractors exceeded the prices that the Schneiders had bid--a total of $82,700. In the case of Mrs. Schneider, the judge simply added together the Schneiders' two bids, producing a sum in excess of $100,000. As a result, Mrs. Schneider's offense level rose by six points because of victim loss, whereas her husband's rose by only five points because the method of computing victim loss in his case placed him in a lower bracket. The Guidelines' brackets for translating dollars of such loss into punishment "bonus" points have since been changed, so that the Schneiders might receive a greater or lesser punishment if they committed their crimes today, but the changes have no significance for our analysis.

The judge gave no reason for treating husband and wife differently and the government does not defend the difference in treatment. It could not do so. The difference is irrational. Not only is the loss that the Schneiders caused the government (or rather would have caused it, if his bid had not been cancelled and hers had been accepted) a joint one that cannot be apportioned; not only, if it could be apportioned, would there still be no basis for apportioning a larger share to Mrs. Schneider; but in addition the two methods of calculating loss make opposite assumptions about the significance of the size of the bid. For Mrs. Schneider, the higher the bid, the heavier the punishment, because for her the loss to the victim was equated to the amount of the bid. For Mr. Schneider, the higher the bid, the lighter the punishment, because the higher the Schneiders' bid the smaller the difference between their bid and that at which the contract was ultimately let; and it is this difference, the "excess procurement cost" as the government calls it, that is the measure of victim loss that the judge applied to Mr. Schneider.

Forced, as it believes itself to be, to choose between the two methods used by the district judge, the government says--naturally, for it is the position that yields the heavier punishment--that the proper method for calculating loss to the victim in a case such as this is the one that the judge used on Mrs. Schneider. And so the government proposes that "the calculation of appellant Paul Schneider's total offense level should be raised one level to account for the error committed by the district court." But since the government neglected to appeal, we cannot raise Schneider's sentence.

The government's failure to appeal is academic, since the method of calculating Mrs. Schneider's offense level was incorrect. The amount bid for a contract procured by fraud is not a reasonable estimate of the loss to the other party to the contract in a case such as this where the contract is terminated before that other party--the intended victim of the fraud--has paid a dime. There may be a loss. The victim may have incurred expenses in terminating the contract or in obtaining a substitute contract. One of the latter expenses might be a higher contract price, if the delay occasioned by the first termination forced the victim to recontract in a market that had moved against him. No evidence was presented of either type of expense with respect to either contract involved in this case; and the face amount of the contract would not be a reasonable estimate of either type of expense.

That is not the end of the analysis, however, because "loss" within the meaning of the Guidelines includes intended, probable, or otherwise expected loss, a qualification of vital importance in a case such as this where the fraud is discovered or otherwise interrupted before the victim has been fleeced. United States v. Davis, 922 F.2d 1385, 1391-92 (9th Cir.1991). But it is necessary to distinguish between two types of fraud. One is where the offender--a true con artist (as in Davis )--does not intend to perform his undertaking, the contract or whatever; he means to pocket the entire contract price without rendering any service in return. In such a case the contract price is a reasonable estimate of what we are calling the expected loss, and we repeat that no more than a reasonable estimate is required. Id. at 1392; United States v. Haddon, supra, 927 F.2d at 951-53; Sentencing Guidelines Sec. 2F1.1, Application Note 8. The other type of fraud is committed in order to obtain a contract that the defendant might otherwise not obtain, but he means to perform the contract (and is able to do so) and to pocket, as the profit from the fraud, only the difference between the contract price and his costs. This is such a case. Mr. Schneider had at the time of the offenses for which he was convicted in this case performed some fifty contracts for federal agencies. Unless he is a fraudfeasor of monumental proportions--which is not suggested--he could not have kept getting government contracts if he never performed any of them. And as a matter of fact, so far as appears he performed all of them, and to the perfect satisfaction of the contracting agency. There is no reason to believe that he and his wife would not have performed these two contracts to the equal satisfaction of the contracting agencies. The Schneiders would not have walked away with anything close to (let alone in excess of) $100,000 in net profit and the government would not have incurred a loss of anything close to $100,000. As a matter of fact, the government--which is not famous for being a fast payer of its bills--would not have paid any part of the contract price until the Schneiders had rendered commensurate part performance. Since the willingness and ability of the Schneiders to perform the contracts are not questioned and they were the low bidders, chances are the government would have gained, not lost, by dealing with the Schneiders.

At least in a narrow financial sense; for we do not mean to suggest that the expected loss to the government was necessarily zero when all relevant consequences are brought into view. Satisfaction of the government's desire to avoid contracting with recent felons and to have a valid bond guaranteeing payment and performance must have some value to the contracting agencies, and the Guidelines permit an increase in offense level for nonmonetizable losses. Application Note 9 to Sec. 2F1.1; cf. United States v. Fousek, 912 F.2d 979 (8th Cir.1990) (per curiam). The Schneiders would not be permitted to offset such losses by showing that because they were the low bidders they would have conferred a net gain on the taxpayer even after due consideration of any added risk (which of course might not have materialized) that their failure to obtain a proper bond might have created. Felix culpa is not a doctrine of the criminal law. Rejecting a parallel argument, we have remarked that "just as it is embezzlement if an employee takes money from his employer and replaces it before it is missed, so it is fraud to impose an enormous risk of loss on one's employer through deliberate misrepresentation even when the risk does not materialize." United States v. Dial, 757 F.2d 163, 170 (7th Cir.1985) (citation omitted).

The government might also have incurred expenses of recontracting, as we have said. But no estimate of such expenses was presented, and the government did not argue that there should be an upward adjustment in punishment in recognition of the impairment of any nonquantifiable values embodied in the bidding requirements that the Schneiders flouted.

The government argues that its suggested method of calculating victim loss is simple. Simple it is. But it is irrational. For it would mean that the Schneiders would (other things being equal) be punished as severely as a con artist who intended to winkle $142,400 ($65,900 + $76,500) from a senile old lady.

We can find no cases in point, but United States v. Whitehead, 912 F.2d 448, 452 (10th Cir.1990), is close and it too reversed a sentence because the estimate of the loss bore no relation to economic reality. True, sentences based on victim losses measured by the entire amount of loans that the defendants had obtained by fraud were upheld in United States v. Johnson, 908 F.2d 396, 398 (8th Cir.1990), and United States v. Connor, 920 F.2d 927 (4th Cir.1990) (per curiam), against the argument that since the loans were secured the lenders' loss was less than the full amount of the loans. But the borrower in Johnson had no intention of repaying her loans (Connor simply repeats the reasoning of Johnson ), and the fact that the victims might have been able to recover some of the money taken from them by enforcing their security interests no more reduced the victims' loss in the contemplation of the law than if a pickpocket got cold feet and returned his victim's wallet before the victim discovered it had been missing. The crime is complete when the thief obtains the victim's property. United States v. Kucik, 844 F.2d 493, 497 (7th Cir.1988). What happens later is irrelevant. But here the defendants got (or would have gotten, had their fraud succeeded) nothing from the government but a pair of contracts, and the defendants were experienced contractors and the low bidders to boot. They may have placed the government at risk, but the government has made no effort to quantify that risk, liberal as the requirements are for such quantification.

We add that of course the Schneiders will not go scot free merely because the government failed--failed utterly--to prove any loss to the victim of their fraud. The statutes under which they were convicted do not require a minimum loss to the victim. 18 U.S.C. Secs. 2, 371, 1001, 1341; Sentencing Guidelines Secs. 2B1.1(a), 2F1.1(a); United States v. Gilliland, 312 U.S. 86, 92-93, 61 S. Ct. 518, 521-22, 85 L. Ed. 598 (1941); United States v. Lea, 618 F.2d 426, 429 n. 3 (7th Cir.1980). It is simply that the Guidelines award bonus punishment points for different levels of proven loss beginning with $2,000. The government did not earn a bonus in this case.

The Schneiders have challenged their convictions as well as their sentences, but on grounds that palpably lack merit and do not require discussion (ditto regarding their other grounds for challenging the sentences). Shortly before the argument of this appeal, Mrs. Schneider wrote this court a letter stating that if her lawyer did not argue one of those grounds, then she--Mrs. Schneider--did not want that lawyer to appear for her in this court. Well, her lawyer didn't argue it, but she was wise not to and Mrs. Schneider has no cause to complain. If Mrs. Schneider nevertheless wants to discharge her and request us to appoint another lawyer, she should submit a motion to that effect to us. We shall merely offer her our opinion that she would be ill advised to follow that course. After argument, Mr. Schneider submitted a handwritten petition for remand that is legally frivolous, and is hereby denied.

The conviction is affirmed but the sentences are vacated and the case is remanded to the district court for resentencing in accordance with the principles set forth in this opinion, that is, without an additional punishment based on a proven monetary loss--for none was proved.