United States Can Company v. National Labor Relations Board, and United Steelworkers of America, Afl-Cio-Clc, Intervening

EASTERBROOK, Circuit Judge. A long time has passed since United States Can Co. v. NLRB, 984 F.2d 864 (7th Cir.1993), enforced the Labor Board’s substantive decision in this case, 305 N.L.R.B. 1127, 1992 WL 10097 (1992). Now we’re at the remedy stage; the Board has ordered the employer to fork over about $1.5 million in back pay, plus pension credits and other fringe benefits, plus more than a decade’s worth of interest, to 28 employees who between 1987 and 1989 were denied opportunities to transfer to other plants in lieu of layoffs. See 1999 NLRB Lexis 310, slightly modifying the ALJ’s decision at 1998 NLRB Lexis 268. We shall cut to the chase; exhaustive recitations of events may be found in the decisions just cited. The Board concluded in 1992 that, when United States Can Co. acquired four packaging plants, it also acquired a collective bargaining agreement governing labor relations at those plants. As a result, U.S. Can was obliged for the duration of the agreement (which expired early in 1989) to offer employees laid off as a result of a plant closing the opportunity to transfer to vacant positions at other plants under an Inter-Plant Job Opportunity Program (ipjo). The Board’s order left open the calculation of back pay for employees who could have used the ipjo (had U.S. Can honored its obligation) to remain on the payroll for a while. Because U.S. Can eventually closed all four acquired plants, transfers would have postponed but not prevented the employees’ need to find work elsewhere. U.S. Can’s principal contention in this court — that the 28 employees would not have transferred even if offered the chance — goes nowhere given the standard of review. We must enforce the Board’s decision if substantial evidence on the record as a whole supports it. Universal Camera Corp. v. NLRB, 340 U.S. 474, 71 S.Ct. 456, 95 L.Ed. 456 (1951). This decision is supported by the employees’ testimony. Each was asked whether he or she would have transferred if given the opportunity; each answered yes; the Board believed these answers. Such a decision is all but invulnerable. Anderson v. Bessemer City, 470 U.S. 564, 570, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985). U.S. Can grumbles that the Board shifted the burden of persuasion, forcing it to disprove the possibility that the employees would have transferred. That’s not what happened. Once the employees testified that they would have moved, U.S. Can had an opportunity to persuade the Board not to believe that testimony. Extending such an opportunity does not “shift the burden of proof’; the Board chose in the end to credit the employees because they are believable, not because U.S. Can failed to carry some burden. What U.S. Can did in an effort to undercut the employees’ position was offer the testimony of a labor economist, David S. Evans, that the employees did not have much to gain by transferring. Evans began with the salary each could have earned at a new plant following an ipjo transfer. He subtracted what the employee would give up: unemployment compensation, supplemental unemployment benefits (sub) *629 under the collective bargaining agreement, and pension benefits for which many of the laid-off employees were eligible. (Access to ipjo transfers depended on seniority, and it is not surprising that the most senior employees in a closed plant were eligible for either early or regular retirement.) According to Evans, the difference between the wages (and increases in future pension benefits) these 28 employees would have received and the benefits they would have surrendered was just about the minimum wage. What kind of person, U.S. Can asks, will move to another part of the country to take a minimum-wage job when low-paying work is available close to home? One answer might be “the kind who work in packaging plants”; after all, these 28 swore that they would do this. It is not as if Evans calculated that their returns from ipjo transfers would be negative, so that moving would be irrational. And the employees may have been better economists than Evans, for their net wages would have been well above the minimum. Evans assumed that if the employees moved then sub and unemployment benefits would vanish. Not so. These payments would be deferred until the next layoff, when employees would enjoy them (less the time value of the delay) even if they took ipjo transfers. Although a future layoff was not certain, the probability was high given the ongoing consolidation in the industry. Evans, however, implicitly assumed that the probability was zero and that transferring employees never would receive unemployment or sub payments to make up for what they would lose by transferring. Employees themselves would not have made that error. Similarly with pension payments, which not only would be deferred but also would increase as the employees made additional contributions to their pension accounts. For an employee eligible to retire, the decision to work another year at a distant plant is no different from the decision to work another year locally. The employee gives up the pension benefit that could have been received immediately, getting in return a salary plus pension contributions that increase the monthly benefit later; delay in retirement increases the monthly benefit for the further reason that actuarial tables predict that the pension will be paid out for fewer months. Whether the exchange is sensible depends on what the person will earn in the interim, on the rate of interest implicit in the pension plan (that is, how fast do monthly benefits increase if the employee postpones their commencement?), and on how long the employee expects to live after retirement. U.S. Can did not demonstrate that none of these 28 employees could have deemed it sensible to work another year. U.S. Can would have done better to examine its predecessor’s experience in operating the ipjo program. When Continental Can, whose business U.S. Can acquired, closed a plant and offered transfers, what happened? If no one accepted, then U.S. Can would have had powerful support for the proposition that the testimony of these 28 employees should not be believed. The Board should place what people do above what they say, for talk is cheap (and employees who know that a “yes” answer to the question “would you have moved?” will bring a large financial reward, without any risk, will find that this sways their beliefs about what they would have done a decade earlier, even if they are trying to be candid). Whether these 28 would have declined transfers is not dispositive; if they had said no, then the opportunity would have fallen through the seniority table until all eligible employees had been offered the chance to transfer. Only if vacancies would have remained at open plants after all eligible employees at the closed plants had a chance to move *630 would U.S. Can be off the hook. A look at Continental Can’s experience would have revealed whether that happened. But at oral argument counsel for U.S. Can replied that it had not examined Continental Can’s experience (or at least had not put this experience in the record). What is in the record, Evans’ flawed calculation, fell well short of the evidence needed to disable the Board from crediting the employees’ testimony. On to the calculation of back pay. According to U.S. Can, the back-pay period should not start until the union filed its unfair-labor-practice charge, or perhaps until the General Counsel of the Board commenced the unfair labor practice proceeding. U.S. Can concedes that both the union’s charge and the General Counsel’s complaint were timely but contends that it is inequitable to require it to afford back pay for earlier conduct. The idea that judicial notions of equity trump the statutory timeliness rules in cases commenced by the United States (or one of its agencies) runs headlong into Occidental Life Insurance Co. v. EEOC, 432 U.S. 355, 97 S.Ct. 2447, 53 L.Ed.2d 402 (1977), and Costello v. United States, 365 U.S. 265, 281, 81 S.Ct. 534, 5 L.Ed.2d 551 (1961), which hold that equitable principles such as laches play little if any role in the federal government’s litigation. See also NLRB v. P*I*E Nationwide, Inc., 894 F.2d 887, 893 (7th Cir.1990). Cf. NLRB v. Ironworkers, 466 U.S. 720, 104 S.Ct. 2081, 80 L.Ed.2d 715 (1984) (long administrative delay, while back pay accumulates, does not prevent enforcement of the Board’s award); NLRB v. J.H. Rutter-Rex Manufacturing Co., 396 U.S. 258, 90 S.Ct. 417, 24 L.Ed.2d 405 (1969) (same); Reich v. Sea Sprite Boat Co., 50 F.3d 413 (7th Cir.1995) (reserving question whether laches ever applies to the federal government’s cases). There’s an additional problem here: U.S. Can did not establish prejudice from the delay. If it had allowed the employees to make ipjo transfers as soon as the union complained, then it might have had a good claim that delay in making the complaint caused injury by requiring it to pay twice for the same work (one payment to the person actually hired to do the job, the other in back pay to the employee who might have transferred). But U.S. Can dug in its heels and refused to transfer the employees. Its back-pay obligation therefore would have been the same had the union and the General Counsel acted immediately. The Board awarded as back pay the gross wages the 28 employees would have earned at other plants, had they transferred and worked until those plants too were closed. It refused to give U.S. Can credit against this award for supplemental unemployment benefits and pension benefits that the employees received in lieu of wages. Whether to net out sub and pension benefits divided the Board: Chairman Truesdale and Member Fox were the majority; Member (now Chairman) Hurtgen dissented. Member Hurtgen concluded that sub and pension benefits, unlike unemployment benefits, see NLRB v. Gullett Gin Co., 340 U.S. 361, 71 S.Ct. 337, 95 L.Ed. 337 (1951), fall outside the collateral-source doctrine because they are provided by the employer rather than a third party. (These are noncontributory, defined-benefit pension plans.) He concluded: [The] benefits at issue are not part of a governmental program that is broadly applicable to employees generally and aimed at addressing a general “policy of social betterment.” [Language quoted from Gullett Gin.] Rather, the pension and sub payments came from employer-sponsored private benefit programs. These benefit programs, which inure solely to the benefit of [U.S. Can’s] employees, are sponsored and funded by *631 [U.S. Can], and [U.S. Can] is responsible for administering the payments to its employees. Thus, there are no overriding social policy considerations that support the discriminatees’ retention of these payments. To the contrary, the relevant “policy” consideration is that the discriminatees should not receive a double-recovery windfall. [A footnote here continues: “In F&W Oldsmobile, [272 N.L.R.B. 1150, 1984 WL 36963 (1984)], the judge, affirmed by the Board, spoke, generieally of benefits, without distinguishing between governmental and private benefits. Indeed, it does not appear from the decision in that case that the distinction was even argued. Since the distinction is made by the Supreme Court in Gullett, I shall follow the Court’s teaching and not that of F&W."] Faced with this challenge based on the Supreme Court’s reasoning in Gullett Gin, the majority of the Board did not respond in the text of its opinion but did drop this footnote: In accord with Board and judicial precedent cited and fully discussed in the judge’s decision, we affirm the judge’s conclusion that [U.S. Can] is not entitled to an offset for retirement benefits and supplemental unemployment benefits paid to discriminatees during the back-pay period. Although funded as part of the benefits earned by employees, the trusts are, as the judge correctly found, separate and distinct entities from [U.S. Can], The dissent would effectively overrule the precedent cited by the judge. We decline to do so. That the majority relegated to a footnote the only significant issue in the case is bad; that it gave no reasons is worse. Member Hurtgen made a serious argument; the majority did not give a serious answer. Nor did the ALJ’s opinion tackle this subject, except to claim that cases such as F&W Oldsmobile support the outcome. We have no idea why the majority thought it important that the payments come from “separate and distinct entities” that are funded 100% by the employer. If an employer arranged for wages to be paid by a parent, subsidiary, or corporate affiliate — “separate and distinct entities” all— would the Board conclude that these payments should be ignored when computing back pay? Surely not. Why, then, does it matter that fringe benefits such as sub and pension payments are funneled through separate entities? Why not treat wages and fringe benefits identically for this purpose (especially when the employer is liable if the “separate and distinct entity” does not pay in full)? The Board did not say. The idea behind the collateral-benefits doctrine, which originated in tort law, is that damages measured by the injury are essential to deterrence. See Steven Sha-vell, Economic Analysis of Accident Law §§ 6.6.7, 10.3 (1987). Suppose A negligently trips B, who incurs $10,000 in medical expenses. Suppose further that B had medical insurance covering these costs. To deduct the insurance proceeds from A’s damages not only would give the tortfeasor the benefit of the victim’s expense on insurance but also would curtail deterrence. Why should A take care in the future if his victims bear their own loss? If the government rather than B supplies the insurance, the analysis is the same. B paid for the coverage through taxes, and giving A the benefit would reduce if not eliminate A’s incentive to take care. That’s a central point of Gullett Gin, which held that unemployment insurance benefits are within the scope of the collateral-source doctrine. Although employers pay a tax that funds some of these benefits, some come from general revenues and all are portable; it *632 would be a mistake to say that the employer that laid off a worker fully pays for the resulting benefits and therefore takes the employee’s whole loss into account when deciding what to do. Contrast unemployment benefits, whose level and eligibility criteria are out of the employer’s hands, with severance benefits. Think of an employer that promises its workers that, if they are discharged, it will pay six months’ wages in a lump sum. An employer discharges a worker and pays the promised benefit. Six months later the Labor Board concludes that the discharge was unlawful and orders the worker reinstated. How much back pay should the employer be required to tender to make the employee whole? — for make-whole remedies are the NLRB’s financial tool. 29 U.S.C. § 160(c); Sure-Tan, Inc. v. NLRB, 467 U.S. 883, 900-01, 104 S.Ct. 2803, 81 L.Ed.2d 732 (1984); NLRB v. Strong, 393 U.S. 357, 359, 89 S.Ct. 541, 21 L.Ed.2d 546 (1969). The answer is zero. The employer already has paid six months’ wages, and once restored to the payroll the employee will not have missed a dime. The employee loses nothing tomorrow either, because the employer’s promise to pay severance benefits remains; if the employer makes a lawful discharge in the future, the employee gets another check for six months’ wages. (This restoration proviso is important, lest the employer be able to appropriate the value of a fringe benefit promised to its workers. We return to this subject toward the end of the opinion.) Now suppose that instead of paying six months’ wages as a flat benefit, the employer coordinates severance benefits with unemployment compensation (as many pensions are coordinated with Social Security retirement benefits) and pays only the difference between six months’ wages and any unemployment benefit the former employee receives. Per Gullett Gin, the unemployment benefits — which come from the public fisc- — are not deducted from a back-pay award. But the severance benefit surely would be deducted. And that still would be true if the employer’s severance package lasted not a flat six months, but until the employee found a new job. The amount of back pay the employee would receive would remain the wages lost, less other payments (including the severance benefit from the employer). One more step and the hypothetical becomes the actual case. Suppose that instead of paying the severance benefit from general corporate funds, the employer sets up a second entity — call it the Severance Benefits Administrator — that disburses the money. Instead of making an unfunded severance-benefit promise, the employer writes a check to the Administrator every month to cover the actuarial cost of its severance promises. Advance funding of promises reduces the cash-flow drain if the employer should decide to close a plant or lay off many employees at once. But from the employee’s perspective it is a detail whether a promise is funded in advance or met on a pay-as-you-go basis, and whether the Administrator is incorporated separately from the employer. It is all the same — the same benefits, the same source, and, one would suppose, the same legal consequences. Yet we have now described the sub program, which is a severance benefit, coordinated with the unemployment insurance system and administered through a separate entity. If ordinary severance benefits are offset against back-pay obligations — something that the Board concedes is proper— then sub payments also must be offset. Pension promises are fundamentally the same; severance benefits with a longer payout. It is impossible to see why unfunded benefits would be offset, while *633 funded benefits administered by a separate entity would not be offset. This distinction is the nub of the Board’s decision, yet the Board offered no logical support. The make-whole goal of putting the employees in the position they would have occupied, but for the employer’s unfair labor practice, is unaffected by whether the benefits are funded or unfunded, paid directly from the employer’s treasury or disbursed by an intermediary. The Board’s majority did refer to “judicial precedent cited and fully discussed in the [ALJ’s] decision”. Here is what the ALJ had to say about pension benefits (adding later that sub payments are identical in principle): [I]t is well-settled that “application of the collateral source rule depends less upon the source of funds than upon the character of the benefits received.” Haughton v. Blackships, Inc., 462 F.2d 788, 790 (5th Cir.1972). Thus, the mere fact that an employer contributes money, either by paying premiums, making contributions, etc., to the fund from which benefits are derived does not establish that the fund is not a collateral source. Id. Also, Blake v. Delaware and Hudson Railway Company, 484 F.2d 204, 206 (2nd Cir.1973); Phillips v. Western Company of North America, 953 F.2d 923, 929 (5th Cir.1992); Molzof v. United States, 6 F.3d 461, 465 (7th Cir.1993); Rtisso v. Matson Navigation Company, 486 F.2d 1018, 1020 (9th Cir.1973). As was more succinctly stated in Folkestad v. Burlington Northern, Inc., 813 F.2d 1377, 1381 (9th Cir.1987), “courts have been virtually unanimous in their refusal to make the source of the premiums the determinative factor in deciding whether the benefits should be regarded as emanating from the employer or from a ‘collateral source.’ ” Respondent’s assertion, therefore, that the benefits should be found not to be collateral because it alone funds the pension trust account is without merit. Further, while Respondent may fund the pension trust, under ERISA the trust remains a separate and distinct entity independent from Respondent. See, NLRB v. Amax Coal Co., 453 U.S. 322, 332-333, 101 S.Ct. 2789, 69 L.Ed.2d 672 (1981); Doyne v. Union Electric Company, 953 F.2d 447, 451 (8th Cir.1992); Garland-Sherman Masonry, 305 NLRB 511, 513, 1991 WL 229938 (1991). As the source of pension benefits received by the dis-criminatees was the trust fund and not the Respondent, said payments were clearly from a collateral source and not deductible from backpay. 1998 NLRB Lexis 268 at *36-37. Most of these cases concern insurance. It is economically irrelevant whether the employer pays an employee $1 per hour extra, and the employee uses the money to buy health insurance, or whether instead the employer offers health insurance as a fringe benefit. (The difference may be vital for tax purposes, but these do not concern us here.) The insurance, like the wages, becomes the employee’s asset. Severance benefits cannot be conceived that way. It is therefore not surprising that, whatever may be the rule in the ninth circuit (the source of the lengthy quotation from Folkestad), this circuit has concluded that severance benefits and them economic equivalents in pension packages must be offset when calculating make-whole awards. See, e.g., Orzel v. Wauwatosa Fire Department, 697 F.2d 743, 756 (7th Cir.1983) (dictum); Kossman v. Calumet County, 849 F.2d 1027, 1032 (7th Cir.1988). Although, as the Board’s appellate brief (though not the Board’s or ALJ’s opinion) notes, we held in EEOC v. O’Grady, 857 F.2d 383, 389-91 (7th Cir.1988), that particular pension benefits should not be deducted from a make-whole award of back *634 pay, we emphasized in O’Grady that the benefits in question came from a state agency other than the employer, making them more like Social Security retirement benefits. The benefits in this case, by contrast, come ultimately from the employer’s resources. What is economically important for deterrence is not whether benefits pass through an intermediary but how the wrongdoer’s actions affect the victim’s well-being, and whether those costs are brought home fully to the wrongdoer. What is important for compensation is not whether benefits pass through a financial intermediary, but whether the victim is put in the same position he would have occupied had his rights been respected. So we set the financing arrangements aside and ask the important questions directly. The employer would indeed gain from its wrong — and the employee would lose out — if the employer were allowed to subtract, from the back-pay obligation, pension and welfare benefits that serve as deferred compensation for work performed. See Hunter v. Allis-Chalmers Corp., 797 F.2d 1417, 1428-29 (7th Cir.1986). That is why health insurance is treated as a collateral source even when the employer provides insurance as a fringe benefit. Likewise with sub and pension benefits. A direct offset would permit the employer to appropriate a portion of the employee’s own economic-benefits package. Deterrence would be reduced, and the employee would be worse off to boot. U.S. Can therefore is wrong to say that it is entitled to a dollar-for-dollar offset. That’s the same mistake David Evans made, a fallacy we discussed above. Think back to our first hypothetical, the lump sum severance payment. Allowing a simple offset would be equivalent to permitting the employer to treat the severance benefit as wages (canceling the back-pay obligation for the six months of unemployment), without restoring that benefit when the employee was lawfully let go. If. the employer is to deduct the benefit from the back-pay obligation, it must restore the benefit for later use; otherwise the employee’s fringe benefit has vanished into the employer’s pocket. Here’s a concrete illustration. Suppose that Alex Baugh, one of the employees not offered the opportunity to transfer, would have moved to a new plant on January 1, 1988, and been laid off on January 1, 1990, if all of his entitlements under the collective bargaining agreement had been honored. Instead Baugh retired at the start of 1988 and began collecting both sub and pension benefits. (These are not the actual details for Baugh; we are simplifying for illustrative purposes.) The Board held that Baugh is entitled to about $90,000, representing the wages he could have earned during 1988 and 1989, plus pension credits and interest, on top of the unemployment, sub, and pension benefits that he actually received. U.S. Can wants to proceed as follows: deduct from the gross wages that Baugh would have received during these years the sub payments that effectively guaranteed him his full pay through July 1, 1988, and then subtract all of the pension benefits he received during those years, cutting the total to about a quarter of the Board’s actual award. Just as the Board erred in saying that sub and pension benefits must be ignored, so U.S. Can errs in saying that they must be deducted dollar for dollar. Suppose that Baugh had worked at another plant through the end of 1989 and then been laid off. He would have been entitled to sub payments for the first six months of 1990, plus higher monthly pension benefits (increased not only by the contributions made to his account during 1988 and 1989, but also because he would have been two years older on retirement). U.S. Can’s approach *635 effectively strips Baugh of the sub payments' — he does not keep what he received in 1988, and he does not get them in 1990 either. Moreover, U.S. Can’s calculation takes away the two years of pension benefits in 1988 and 1989 without higher monthly benefits in 1990, an increase that Baugh would have received had he worked through the end of 1989. (The Board’s position suffers from the same problem in reverse. Under the Board’s award Baugh is entitled to keep the pension benefits he received in 1988 and 1989, plus receive pension credits as if he had worked in those years and thus could retire in 1990 with higher monthly benefits.) Pension and sub offsets are therefore considerably more complex than either the Board or U.S. Can supposes. Everyone’s rights are respected if, and only if, the offset puts each employee in the position he would have occupied had the collective bargaining agreement been respected— and that position is one in which the benefits are deferred but not obliterated. To make this concrete: U.S. Can is entitled to offset the sub payments from its back-pay obligation only if it also restores those benefits as of the time each employee finally would have left its employ. It is entitled, in other words, to the time value of the money: in Baugh’s (hypothetical) case it would make sub payments in 1990 rather than 1988, and thus would gain two years of interest on the obligation. With respect to pension benefits, the employer may deduct them from back-pay obligations during the time they were actually received, only if the employer also recalculates the payments for later months and increases them to the level they would have achieved had retirement been deferred. We suspect that U.S. Can would not reduce its net payouts all that much by these recalculations. Many employers therefore might be content to disregard sub and pension benefits, just as the Board did in this case. But the Board may not deprive an employer of its opportunity to make the more accurate calculation we have described, a calculation that fully respects the employer’s entitlements as well as each employee’s. The Board’s order is enforced, subject to the proviso that if within 30 days U.S. Can informs the Board that it wishes to offer actuarial calculations along the lines we have described, the Board must reopen the record and make new back-pay awards using the methods approved in this opinion.