POSNER, Circuit Judge.
In 1988, Marshall Erdman and Associates, which (among other activities) designs and builds medical buildings, fired Bobby Price, a salesman in its midwest division. Price sued Erdman, and the head of the midwest division, Halverson, under the Age Discrimination in Employment Act, 29 U.S.C. §§ 621 et seq. Price obtained a jury award (against both defendants) of $178,-700 in backpay, consisting of what he would have earned from Erdman between the date of his discharge and the date of trial, had he not been fired, minus what he did earn from others; because the jury found the defendants’ violation of the age discrimination law to have been willful, the judge doubled this award. 29 U.S.C. § 626(b). The jury also awarded Price $750,000 in “front pay,” representing the present value of the future earnings that Price would have had from Erdman, minus the present value of his anticipated future earnings in his current occupation of real estate broker. Front pay is an alternative to reinstatement, which Price had requested but which the judge refused to order because of “mutual dislike and defendants’ continued opinion that plaintiff is incompetent.” The judge nevertheless threw out the jury’s award of front pay on the grounds that the evidence of Price’s lost future earnings was speculative and that an award of front pay would duplicate the award of liquidated damages (Congress’s name for the doubling of damages to which the victim of a willful violation of the age discrimination law is entitled). Price’s cross-appeals challenge both the denial of front pay and the judge’s award of only $82,000 in attorneys’ fees; Price had sought $265,000. He does not appeal from the denial of reinstatement.
The evidence of age discrimination was thin; so thin, the defendants argue, as to entitle them to judgment notwithstanding the verdict. Price was 45 when he became a medical-buildings salesman for Erdman (his previous job had been as a construction supervisor for the company). This was the same age as his boss, codefendant Halverson. Two years after being hired, Price was fired, from the company as well as from the division, after some customers had complained about his inattention to detail and after Halverson had been told by his boss to reduce the sales force in the midwest division by one as part of a company-wide economy drive — a directive that Halverson protested before reluctantly carrying it out by firing Price, his newest salesman. It is true that despite the so-called economy drive, Halverson was permitted to bring in another salesman, but the new man came from another division of Erdman, so there was no additional cost to [323]*323the company. And while Price, although the newest salesman, was not the youngest one and was replaced by a man in his twenties (Foy), a much older salesman was retained. The sale of medical buildings to doctors and hospitals does not seem the type of activity in which a youthful image is valued. Nor was anyone in a senior position at Erdman younger than Price.
These are the facts stressed by Erdman but there are others. Despite the customer complaints, Erdman credited Price with substantially higher sales than the young man who replaced him, even though Price’s territory was reconfigured to make things easier for Foy. (Shades of Shager v. Upjohn Co., 913 F.2d 398 (7th Cir.1990), where we upheld the finding of age discrimination.) And Foy too had been the target of a complaint by a customer — a serious complaint that had required repara-tive action by Halverson. This is not a criticism of Foy but it places' customers’ complaints against Price in perspective. Such complaints are inevitable in a business that involves the sale of a complex product to sophisticated and demanding professionals. Price emphasizes, moreover, that not only did he sell more than Foy, but his sales were more profitable, even though costs incurred in responding to customers’ complaints were subtracted in computing the profitability of a sale. Another bit of evidence — trivial in itself, but part of the mosaic — is that an executive of Erdman said that Price did not fit the desired image of a “consultant for the nineties,” that is, the 1990s (Erdman calls its medical-building salesmen “sales consultants”). When asked at argument what this expression was intended to convey, the defendants’ able lawyer said that Erdman’s salesmen would have to be adaptable in order to flourish in the changed competitive environment anticipated for the then forthcoming decade. Older people are considered less adaptable than younger ones (“you can’t teach an old dog new tricks”). As for the older salesman who was retained when Price was fired, he was nearing retirement age so might be gone soon anyway. Finally, although Halverson was the same age as Price, some supervisors feel uncomfortable having subordinates of their own age or older. They prefer to supervise younger persons, who are apt to be more deferential, because deference is more easily paid by a younger person to an older one than by an older person to a younger person or to a contemporary. So this case fits the general pattern of Shag-er, where a middle-level supervisor, uncomfortable with having a subordinate older than himself, fires the subordinate (or engineers his firing).
The jury didn’t have to believe all this, and, even if it did, this would still be a weaker case than Shager; but we cannot say that it is so weak that no rational jury could have brought in a verdict of age discrimination. Nor do we think the jury’s conclusion that the violation was willful can be upset. Ordinarily, it is true, it is irrelevant to liability, whether civil or criminal, that the defendant did not know he was violating the law. But sometimes the law is so intricate, or so remote from the moral intuitions of the people made subject to it, that the legislature conditions the imposition of a penalty on proof that the violation was “willful.” The concept of willfulness clearly includes the defendant who knew he was violating the law and clearly excludes the defendant who reasonably believed that he was not violating it, but what about the defendant who was careless or even reckless with respect to the possibility that he might be violating the law? This case falls in that intermediate zone. If, as the jury found, Halver-son fired Price because Halverson was uncomfortable supervising a person of his own age, this was not an accident resulting from the intricacies of the age discrimination statute or the interpretive doctrines that have been grafted on to it, as in Trans World Airlines, Inc. v. Thurston, 469 U.S. 111, 105 S.Ct. 613, 83 L.Ed.2d 523 (1985), and Heiar v. Crawford County, 746 F.2d 1190, 1200-01 (7th Cir.1984). Halverson testified, however, that he had thought the protection of the statute did not begin until a worker reached the age of 50. (Notice how this weakens the defendants’ attempt to get mileage from Halverson’s retention [324]*324of an older salesman — Halverson thought him, but not Price, protected by the statute, thus tying Halverson’s hands.) The jury was not required to believe him — and if he was telling the truth, and Erdman had left him in ignorance of a basic feature of the statute, it was .an extraordinary mistake for Marshall Erdman and Associates to have made. No mom and pop grocery store, Erdman had annual sales the year Price was fired of almost $200 million. But was it a mistake that connotes willfulness?
The cases cut the intermediate area as follows. A defendant’s negligent mistake concerning the lawfulness of his conduct under the age discrimination law will not suffice to make that conduct willful, but a reckless mistake, in the criminal law sense of the defendant’s being “indifferent to whether he was violating [the law] or not,” will. Shager v. Upjohn Co., supra, 913 F.2d at 406; see also Trans World Airlines, Inc. v. Thurston, supra; EEOC v. Century Broadcasting Corp., 957 F.2d 1446, 1457-60 (7th Cir.1992). The jury was entitled to find that Erdman’s conduct fell on the reckless side of the line. Erdman had so far neglected its responsibilities for compliance with the age discrimination law as to allow a supervisory employee to whom it had delegated the power to hire and fire to remain ignorant of one of the most basic features of the law — namely the age at which workers are protected by it.
Here is a bit of a complication, however: Halverson was named as a defendant along with his employer, and both were found to have committed willful violations. The finding is unproblematic with respect to the employer. Halverson’s act in firing Price because of his age (as the jury found) is imputed to his employer, Shager v. Upjohn Co., supra, 913 F.2d at 405, and the employer’s failure to inform Halverson concerning the fundamental requirements of the age discrimination law supplies, as we have just seen, the requisite willfulness. But what was the basis for the jury’s finding that Halverson was willful, if he genuinely didn’t know the requirements of the statute? The jury may not have believed him, but suppose it did? Could it then be said that he was reckless in failing to inform himself of those requirements? Suppose he had been erroneously informed by lawyers employed by Erdman that the protection of the statute doesn’t kick in until the employee reaches the age of 50; could he be deemed willful? Probably not. But we need not pursue the issue. The defendants’ brief, a single brief filed on behalf of both defendants, makes no distinction between Halverson and Erdman so far as the willfulness of their violations is concerned. Nor do the defendants argue that age played at most an “unconscious” role in the decision to fire Price, a possible ground, we have held, for repelling a finding of willfulness. EEOC v. Century Broadcasting Corp., supra, 957 F.2d at 1458-59; id. at 1465-67 (dissenting opinion).
We move to the issue of front pay. Fortino v. Quasar Co., 950 F.2d 389, 398 (7th Cir.1991), itself an age discrimination case on the relevant issue, holds that both entitlement to and amount of front pay are equitable issues to be decided by the judge rather than legal issues and hence for the jury. An equity judge can always submit an issue to a jury for advice, but he is not bound by the advisory verdict. Yatvin v. Madison Metropolitan School District, 840 F.2d 412, 418 (7th Cir.1988); Wilson v. City of Aliceville, 779 F.2d 631, 635-36 (11th Cir.1986); 9 Charles Alan Wright & Arthur R. Miller, Federal Practice and Procedure § 2335, at p. 126 (1971). The contrary is intimated in Doyne v. Union Electric Co., 953 F.2d 447, 450-51 (8th Cir.1992), but its citation to Gibson v. Mohawk Rubber Co., 695 F.2d 1093, 1101 (8th Cir.1982), suggests that the court confused the effect of an advisory jury’s verdict (none) with the effect of a real jury’s finding of fact on issues tried to the judge in. the same case (preclusive). Hunter v. Allis-Chalmers Corp., 797 F.2d 1417, 1421 (7th Cir.1986); EEOC v. Century Broadcasting Corp., supra, 957 F.2d at 1463; Davenport v. DeRobertis, 844 F.2d 1310, 1313-14 (7th Cir.1988); McKnight v. General Motors Corp., 908 F.2d 104, 113 (7th Cir.1990). Judge Shabaz was free to disregard the $750,000 award.
[325]*325One reason we concluded in Fortino that front pay is an equitable issue is that it can be awarded only if reinstatement, clearly an equitable remedy, is impracticable. We shall not disturb the judge’s ruling that it was impracticable in this case, though the reasons he gave — “mutual dislike and defendants’ continued opinion that plaintiff is incompetent” — are not satisfactory. The passage we have just quoted implies that it makes no difference whether the employee dislikes the idea of working for the employer, or the employer dislikes the idea of having the employee work for him (“defendants’ continued opinion that plaintiff is incompetent”); either way, or both ways (“mutual dislike”), reinstatement should not be ordered. A more discriminating analysis is necessary. Take first the employee’s disinclination to return to working for his employer. If the disinclination is rational and sincere (rather than a maneuver to get front pay), it is a good reason for allowing the employee to elect his alternative remedy of front pay. Whittlesey v. Union Carbide Corp., 742 F.2d 724, 729 (2d Cir.1984). The employer’s dislike of the employee’s returning is a far more problematic ground for declining to order reinstatement. This would be obvious in a case of racial discrimination in which the employer pleaded hostility to the entire group (blacks, or women, or whomever) to which the plaintiff belonged, as a ground for refusing to order reinstatement. To decline to order reinstatement in such a case would reward the employer for the very attitudes that precipitated his violation of the law, by giving him a choice of remedies.
The intermediate case is where the employer dislikes the employee for reasons independent of the latter’s membership in a protected class, and where the feasibility of awarding front pay in lieu of reinstatement makes the burden on the court of supervising a coerced employment relation between the parties disproportionate to any gains from giving the plaintiff his preferred remedy. In such a case a refusal to order reinstatement would be within the trial judge’s equitable discretion. So a belief by the defendants that Price is incompetent could be a proper reason for denying reinstatement. The problem with using it here is that although the judge was not bound by the. jury’s decision to award front pay, or by the jury’s decision on the appropriate size of the award, a jury’s findings within its jurisdiction — here its findings that the defendants violated the Age Discrimination in Employment Act (and owed so much backpay) — bind the judge, as we noted earlier, when he goes to make findings on his part of the case, here the issue of front pay. (All this assumes of course that the jury’s findings are not set aside. Swentek v. USAIR, Inc., 830 F.2d 552, 559 (4th Cir.1987).) A finding that the defendants did not really believe Price incompetent seems implicit in the jury’s verdict in his favor, since the main strut of his case was that he was more competent than the person who replaced him. If this finding can fairly be deemed entailed by the verdict, it would be binding on the district judge when he came to decide whether to order Price reinstated. EEOC v. Century Broadcasting Corp., supra, 957 F.2d at 1463.
But Price does not appeal from the denial of reinstatement, and we think he is right not to. It is one thing to order the reinstatement of low-level employees performing routine tasks, or higher-level employees after the supervisors involved in the unlawful employment action have left the company or been transferred to another division. But to order reinstatement of a high-level employee performing discretionary functions into the division from which he was fired and which remains under the management of the person who fired him is a formula for continuous judicial intervention in the employment relation, even when as here the plaintiff is a salesman who spends much of his working time away from his office and so is not constantly rubbing shoulders with his enemies. If Price is reinstated, every time he is denied credit for a sale, or denied a raise or a bonus, or has a squabble with Halverson, he will be tempted to run to the district court for further equitable relief ancillary to the reinstatement order or even for a finding of [326]*326contempt of the order. There is an analogy to the common law’s refusal to grant specific performance of a contract of employment. A federal district court is not equipped to be the labor relations equivalent of a domestic relations court. Reinstatement in the circumstances that we have described would be justified only if front pay could not be computed.
Seven hundred and fifty thousand dollars in front pay for Price may well be too high, but zero is too low. We pointed out in Fortino that the award of liquidated damages in an age discrimination case will often have a compensatory component — for while these “liquidated damages” really are punitive damages, just like two-thirds of every antitrust or RICO treble-damages award, one function of punitive damages is to compensate a plaintiff for items of loss that may have been omitted from the calculation of compensatory damages. 950 F.2d at 398. So it is arguable that some front pay is, as it were, smuggled into every award of liquidated damages. We used that observation in Fortino to justify this circuit’s rule that prejudgment interest should not be awarded in age discrimination cases in which liquidated damages are awarded. Id. But to exclude front pay on top of excluding prejudgment interest would make no sense, for with prejudgment interest and front pay both barred an award of double damages might well fall short of compensation and thus contain no punitive component at all (in fact contain a negative punitive component). In such a case the plaintiff might be better off if the violation were adjudged not willful. Certainly Price would be better off with simple damages plus front pay of $750,000 than with just double damages: better off to the tune of $571,300 ($178,700 + $750,000 — (2 X $178,700)). So while previous cases, illustrated by Hybert v. Hearst Cory., 900 F.2d 1050, 1054 (7th Cir.1990), and most recently by EEOC v. Century Broadcasting Cory., supra, 957 F.2d at 1464, suggest that the presence or absence of a liquidated damages award is material in determining entitlement to front pay, we think it should play only a very small role in that determination.
That brings us to the question of the proper calculation of the amount of front pay due Price. After making seemingly strenuous efforts to find comparable employment in the medical-buildings industry, Price obtained a real estate broker’s license, which at the time of trial he was just beginning to use. In projecting Price’s future income from real estate brokerage his expert witness used the average annual income of beginning real estate brokers in Wisconsin and predicted that Price's brokerage income would increase at an annual rate of approximately 6 percent over his remaining working life as he gained experience. To determine Price’s lost income as a result of his discharge by Erdman, the expert witness projected modest annual raises over the income that he received during his two years as a medical-building salesman for the company. The'difference between the two projected income streams (in which fringe benefits as well as wage income were included) was discounted to a series of present values ranging from $1.2 million if Price retired at the age of 65 to $2.1 million if he retired at 75. The jury’s award of $750,000 could be interpreted as a determination that Price would have retired before he reached 65-the defendants had introduced evidence that the average age at which a salesman leaves Erdman’s employ is 61.3-or alternatively may evince some skepticism about the expert witness’s methodology. But the judge as we said was entitled to make up his own mind.
He was right that the front pay award was “speculative,” but that is true whenever lost future earnings have to be estimated, as they are routinely in personal-injury cases. The problem with the amount that the jury awarded is not that it was speculative, Blum v. Witco Chemical Corp., 829 F.2d 367, 374-76 (3d Cir.1987); Castle v. Sangamo Weston, Inc., 837 F.2d 1550, 1562 (11th Cir.1988), but that it rested on a defective methodology. A minor point is that the expert failed to discount (multiply) each year’s projected earnings loss by the probability that Price would have lived long enough to obtain those earnings. The probability was not a hun[327]*327dred percent and the estimate of lost earnings should have been scaled down accordingly. Coston v. Plitt Theatres, Inc., 831 F.2d 1321, 1334-35 (7th Cir.1987), vacated on other grounds, 486 U.S. 1020, 108 S.Ct. 1990, 100 L.Ed.2d 223 (1988).
A bigger problem was the expert’s failure to take into account the high volatility of a medical-buildings salesman’s earnings. The figures the expert projected may be the best possible estimate of Price’s mean expected earnings had he remained with Erdman, but the variance around that mean must be considerable. Risk-averse persons — and most people are assumed to be risk-averse in their serious financial affairs — will pay a premium, often a very large one, to avoid risk. That is the rationale of insurance. The loading charge— the difference between the price of the insurance and its actuarial value (the loss insured against multiplied by the probability that the loss will occur) — is an approximate measure of what people are willing to pay to avoid bearing risk. For it is the part of the insurance premium that compensates the insurance company for its administrative and sales expenses, as distinct from the part which merely translates a possible future loss into a current expense; and therefore a person who did not mind risk would not be willing to pay a loading charge — he would prefer to take his chances on the loss’s occurring or not. An award of front pay, however, like other monetary awards, is a lump sum certain, which the plaintiff can invest in as safe a vehicle as he pleases (short-term Treasury bills have zero default risk and a negligible risk of unanticipated changes in the rate of inflation). The award in effect enabled Price to exchange his risky expectations as a salesman of medical buildings for a risk-free asset having the same expected value but, assuming Price is risk averse, a substantially higher utility. A computation of damages that ignores the difference in risk between earnings in a volatile occupation and a judicial award of a lump sum equal to the present value of those earnings is unsound. Douglass v. Hustler Magazine, Inc., 769 F.2d 1128, 1143 (7th Cir.1985); Nemmers v. United States, 795 F.2d 628, 634 n. 2 (7th Cir.1986).
Are the deficiencies in the expert’s estimation of damages so serious as to vitiate the jury’s award? That is not the question. The jury was not authorized to compute front pay. It was the judge’s responsibility. He did not fulfill it, because he thought the plaintiff entitled to no front pay at all. The case must be remanded for a determination of the amount of front pay due Price in accordance with the principles set forth in this opiiiion. The judge may, if he thinks it useful, receive additional evidence. Circuit Rule 36 shall not apply on remand.
Although the case is not over, we might as well consider Price’s challenge to the award of attorneys’ fees. Because the award was premised on zero front pay, the remand will entitle Price to an additional award of fees assuming that he will obtain a front pay award of more than zero — as he is entitled to do. There is no way, however, that the remand can moot Price’s challenge to the fee award already made by the district judge..
Although it is a small point, we agree with Price that the judge should not have disallowed 26 hours of billable time of Price’s senior lawyer, Michael Fox, on the ground that the hours were spent taking depositions, which could just as well have been done by one of the associates in the firm at a lower hourly rate. This is cutting things too fine. Fox & Fox is a tiny firm, with just two partners and two associates, and the maintenance of a rigid bulkhead between partner and associate work is not feasible. If the occasional use of Fox to take depositions raised the cost of the plaintiff’s legal services, the occasional use of an associate to do work that in a more structured firm would be done by a partner doubtless lowered that cost. We also find no basis in the record for the judge’s action in reducing the billing rates of the associates.
We agree with the judge’s other rulings on attorneys’ fees. Only one requires discussion (or did, before City of Burlington v. Dague, - U.S. -, 112 S.Ct. 2638, 120 L.Ed.2d 449 (1992), decided [328]*328while our opinion was in production, anticipated our conclusion on this issue), his refusal to multiply the hourly rates of the plaintiffs lawyers to reflect their risk of losing the case. Multipliers are appropriate, to reflect the risk of losing the case and not being paid at all, but only in cases in which a contingent fee is infeasible. Maybe it is a class action, so there is no client with whom to negotiate a contingent fee (at least a contingent fee that would bind the class). In re Continental Illinois Securities Litigation, 962 F.2d 566 (7th Cir.1992). Maybe it is an equitable suit, so that no money award on which to base a contingent fee is anticipated. But in a substantial money case on behalf of a client with whom a lawyer can negotiate a contingent fee, the lawyer can protect himself against the risk of coming up empty without the judge’s having to increase the court-awarded fee to which the plaintiff will become entitled if he wins; and therefore the plaintiff does not require the prospect of a multiplier in order to obtain representation. Price has already won almost $400,000 and will win more. We do not know the terms of his contract with Fox & Fox, but we do know that it is a contingent-fee contract. It might have provided — for all we know, it did provide — that the firm would get a fixed percentage of the sum of the monetary award plus the court-awarded attorneys’ fees. Suppose the percentage specified was (in the event of an appeal) 40 percent. Then Fox & Fox will already have earned approximately the amount that it says the judge should have awarded in attorneys’ fees by the application of a multiplier. When the lawyer can protect himself through the market against the risk of not being able to collect a fee, there is no need for the court to protect him against that risk as well in order to make sure that civil rights plaintiffs can obtain adequate representation. Id., 962 F.2d at 572-73; Burdett v. Miller, 957 F.2d 1375, 1384 (7th Cir.1992).
This of course assumes that the purpose of a judicial award of fees is to enable plaintiffs to obtain competent counsel rather than to make the cost of litigation to prevailing plaintiffs zero. For if the contingent fee negotiated by counsel exceeds a reasonable attorney’s fee as adjudged by the court without jacking up the lawyer’s ordinary hourly rate, some part of the expense of representation will be borne by the client. The cases, however, assume that the goal of at'orney-fee shifting is to obtain competent counsel rather than to make litigation costless to prevailing plaintiffs. In re Continental Illinois Securities Litigation, supra, 962 F.2d at 572-73. The assumption seems right to us. We are speaking, recall, of substantial money cases, where even after paying a contingent fee (reduced in effect by the judicial award of fees, which goes to the plaintiff himself), the plaintiff nets a substantial award. The prospect of such an award should be enough to induce the plaintiff to sue, and the prospect of the contingent fee should be enough to induce competent counsel to represent him. No moré is necessary to achieve the essential goals of the civil rights laws. The nation is awash in litigation. We have no reason to think that more generous rules for calculating attorneys’ fees in fee-shifting cases are necessary in order to ensure that plaintiffs in general, and civil rights plaintiffs in particular, will be able to vindicate their legal rights.
Affirmed in Part, Reversed in Part, and Remanded.