POSNER, Circuit Judge.

Three professors brought suit under the Age Discrimination in Employment Act, 29 U.S.C. § 626, complaining that the Early Retirement Program of the City Colleges of Chicago, adopted in 1982 as part* of a collective bargaining agreement with a local union of the American Federation of Teachers, discriminates against persons over 65. All three plaintiffs were between *316 65 and 69 when they brought the suit. Two had retired — one at 65, the other at 67. The third, who had been 64 when the Early Retirement Program was adopted, was still employed by the Colleges when the district judge granted summary judgment for the defendants and dismissed the case.

In 1980, two years before the Age Discrimination in Employment Act lifted the minimum mandatory retirement age for college and university teachers from 65 to 70, the Illinois legislature passed a statute making 70 the mandatory retirement age for employees in Illinois. At the time, the City Colleges had a mandatory retirement age of 65 for faculty members. The Colleges contended that the new statute did not apply to them, so they declined to comply with it, precipitating a suit that they lost when the Supreme Court of Illinois decided in 1981 that the statute did apply to them. Even before the decision the Colleges had been concerned with the high average age of the faculty members in many of their departments — the interaction of the tenure system with declining enrollments and budgets had resulted in there being few openings for new teachers — and had been trying to devise an early-retirement program. Four months after the supreme court’s decision, the City Colleges began drafting the Early Retirement Program at issue in this case. In trying to “sell” the Program, the Chancellor of the City Colleges explained that its main purpose was “to realize financial savings by replacing older faculty members earning a higher salary with younger faculty members at the lower end of the salary schedule.” Such replacement would not only save money but also promote the “infusion of new views, ... because you couldn’t "force oldtimers to go out and get more courses in the more recent things in their own field.” There was also talk of bring-ingHn “new blood” (an expression that we suspect will soon disappear from the employers’ lexicon as awareness of the menace of age discrimination litigation seeps through the employer community), and, as part of this process, of making room for women and members of minority groups; most tenured faculty members were white and male.

The Early Retirement Program is open to any faculty member between 55 and 69 years of age who has been continuously employed full-time for at least ten years. The size of the pension that a faculty member who takes early retirement is entitled to is a function of his highest four years’ salary plus his years of service. About this aspect of the Program the plaintiffs have no complaint — for the excellent reason that the size of the pension is keyed not to age but, as we have said, to salary and length of service. The plaintiffs complain about two other features of the Program, which are keyed to age. The first relates to accumulated sick pay. A faculty member who is between the ages of 55 and 58 when he takes early retirement is entitled to receive, in addition to his pension, a lump sum equal to 50 percent of his accumulated sick pay, valued at his base salary rate in his final year of service. The percentage rises to 60 percent for 59-year-olds and 80 percent for 60- to 64-year-olds, then plunges to 45 percent for 65- to 70-year-olds. Retirement is mandatory at 70. Not until 1994 does the Age Discrimination in Employment Act eliminate mandatory retirement for academic employees. See Act of Oct. 31, 1986, Pub.L. 99-592, § 6(b) (repealing 29 U.S.C. § 631(d) as of Dec. 31, 1993).

The second feature of the Early Retirement Program of which the plaintiffs complain is that while faculty members who retire between the ages of 55 and 64 continue to be covered by the Colleges’ group insurance policy (which includes life, health, dental, vision, homeowner’s, and automobile insurance) until they reach the age of 70, those who retire at age 65 or later cease, upon retirement, to be covered. Although they can continue to buy this insurance at the rates in the Colleges’ policy, they are out the premiums, which covered employees do not pay.

The combined effect of the two challenged features of the Early Retirement Program on these plaintiffs is substantial. One of the plaintiffs would have received *317 an extra $23,857 in sick-pay distribution had he retired at 64; and a faculty member who retires at 65 rather than 64 has to pay $22,430 in insurance premiums to retain his coverage until 70. However, the net loss in retirement benefits is lower than the sum of these figures. The later one retires, the higher one’s salary is apt to be in one’s last four years of service and the greater one’s length of service; hence one's pension will be higher. And — though this is a small factor — unused sick leave not credited to the lump-sum distribution is credited to years of service, which also pushes up one’s pension. Moreover, the distribution of sick pay, though it represents a smaller percentage of his accumulated sick leave for the employee retiring later, is computed at the employee’s last salary, which is likely to be higher the longer he stays.

But the relevance of these points is obscure. If an employer had no mandatory retirement age but took away employees’ free parking spaces when they reached 65, this would be age discrimination even if, as a matter of fact, the total remuneration the next year of employees who didn’t take the hint but kept on working rose by more than the value of the parking space that they had lost. And in any event a person who keeps working beyond his preferred retirement age in order to make up the dip in retirement benefits at age 65 is incurring a considerable opportunity cost, whether in leisure forgone or in a forgone opportunity to work for another employer at a salary that would be on top of the retirement benefits from his previous employment. Even if this factor is ignored, it is by no means clear — it is in fact unlikely — that the increment in pension from another year at work would make up for almost a $50,000 drop in retirement benefits. Certainly for purposes of evaluating a motion for summary judgment, the district judge was required to assume that a faculty member who decides not to retire at age 64 incurs a considerable penalty. How large the penally is would normally be a matter for determination at the damages stage of litigation.

The question of the proper treatment of early-retirement programs is the most difficult question under the Age Discrimination in Employment Act. The purpose of such programs often is to ease out older employees, whether because they cost the employer more in salary or fringe benefits, or have gone stale, or are blocking advancement for the ambitious young. Often, therefore, the discussions leading up to the adoption of such programs will generate evidence of age discrimination. Yet the discrimination seems to be in favor of rather than against older employees, by giving them an additional option and one prized by many older employees. Nor can it seriously be argued that the concept of early retirement for workers over a specified age stigmatizes such workers, as would a program designed to change not the age but the racial composition of the work force by allowing blacks but not whites to retire early. Entitlement to early retirement is a valued perquisite of age — an additional option available only to the older worker and only slightly tarnished by the knowledge that sometimes employers offer it because they want to ease out older workers. These considerations led us to hold recently that a worker who elects early retirement cannot turn around and sue his employer unless he can show that he was forced to take early retirement by an explicit or implicit threat to fire (or otherwise punish) him because of his age if he did not. Henn v. National Geographic Society, 819 F.2d 824 (7th Cir.1987).

This case requires us to consider what at first glance may seem a completely different aspect of early retirement: the complaint not of the worker who elected early retirement but of the worker who, having declined it, believes he is being punished for his decision. Actually these two situations are the same. Whether a worker takes early retirement because he fears he will be discriminated against on account of his age if he does not, or refuses to take early retirement and indeed is then discriminated against on account of his age, his rights under the Age Discrimination in Employment Act have been violated.

*318 There is, however, a real and not merely formal difference between this case and Henn. In Henn, all employees over 55 were offered a severance bonus equal to a full-year’s salary if they retired within two months after the offer was made. There was no discrimination within the group eligible for early retirement; the amount of an employee’s early-retirement benefits was not keyed to his age. Of course, everyone in the eligible group who rejected the offer suffered in the sense that if he retired later on he would not get the bonus. But no one suffered because of his age except employees under 55, who were not eligible for the program (more about them shortly). A 56 year old who declined to retire was no better off than a 64 year old who declined to retire; both lost the same thing, a year’s bonus. The offer was more attractive, it is true, the nearer one was to leaving; but ordinarily that would be the older worker. A person who plans to retire at 65, and at age 64 is offered a chance to retire with a full-year salary as severance pay, will leap at the chance. But an early retirement plan that treats you better the older you are is not suspect under the Age Discrimination in Employment Act. Compare Dorsch v. L.B. Foster Co., 782 F.2d 1421, 1427-29 (7th Cir.1986). Title VII protects whites and men as well as blacks and women, but the Age Discrimination in Employment Act does not protect the young as well as the old, or even, we think, the younger against the older. The protected zone begins at age 40, but if on that account workers 40 or older but younger than the age of eligibility for early retirement could complain — workers 40 to 54 in Henn — early retirement plans would effectively be outlawed, and that was not the intent of the framers of the Age Discrimination in Employment Act. The reductio ad absurdum of the reasoning which views early retirement plans with suspicion would be to view permitting workers to retire when they reach the age of 65 as a form of discrimination against workers aged 40 to 64. Then the employer could be confident of escaping liability under the Age Discrimination in Employment Act only by allowing retirement at age 40!

In the present case, however, there is discrimination against the older worker. Everyone between 55 and 69 is eligible for early retirement, but those between 64 and 69 — an older age group — are disfavored relative to the younger employees in the eligible group. If the City Colleges said to their faculty, at age 65 you lose your free parking space (or dental insurance, or any other fringe benefit), they would be guilty, prima facie, of age discrimination. Early-retirement benefits are another fringe benefit — and they plummet at age 65.

There may be good reasons for them to plummet, but the reasons are relevant not to the prima facie case of discrimination but to the defense in section 4(f)(2) of the Act, 29 U.S.C. § 623(f)(2), for “any bona fide employee benefit plan such as a retirement, pension, or insurance plan, which is not a subterfuge to evade the purposes of [the Age Discrimination in Employment Act].” The defense has two elements, and the defendants have the burden of production and of persuasion on both. See, e.g., Cipriano v. Board of Education, 785 F.2d 51, 57-59 (2d Cir.1986) (Friendly, J.). They must show that the features of the Early Retirement Program of which the plaintiffs complain are (1) part of a bona fide employee benefit plan, and (2) not a subterfuge. Some cases, such as EEOC v. Fox Point-Bayside School District, 772 F.2d 1294, 1299 (7th Cir.1985), and EEOC v. Westinghouse Electric Corp., 725 F.2d 211, 223 (3d Cir.1984), mention a third element — that the challenged action by the employer have been in conformity with the plan — and others, such as EEOC v. Borden’s, Inc., 724 F.2d 1390, 1395 and n. 3 (9th Cir.1984), divide part (1) of the required showing into two pieces, plan and bona fide. But these refinements are obvious, and it is desirable to state rules as simply as possible and not make legal doctrine even more prolix and forbidding than it is.

The record compiled in the summary judgment proceeding is adequate to establish (1). The Early Retirement Program, to which the challenged features are integral, is part of the overall retirement plan — a *319 genuine retirement pian — that the City Colleges negotiated with the teachers’ union and embodied in their collective bargaining agreement. The hard question is whether the challenged features constitute a subterfuge, that is, are designed to discriminate against faculty members 65 or older.

Nothing in the Age Discrimination in Employment Act forbids an employer to vary employee benefits according to the cost to the employer; and if, because older workers cost more, the result of the employer’s economizing efforts is disadvantageous to older workers, that is simply how the cookie crumbles. In Metz v. Transit Mix, Inc., 828 F.2d 1202, 1208-11 (7th Cir.1987), this court carved an exception to this principle for the case where the older worker’s higher cost is due solely to his having received pay increases based on seniority. We do not interpret Metz to forbid employers to take into account a higher cost for an older employee that may be due to the higher life-insurance premiums for insuring the lives of older people. But where, as in the present case, the employer uses age — not cost, or years of service, or salary — as the basis for varying retirement benefits, he had better be able to prove a close correlation between age and cost if he wants to shelter in the safe harbor of section 4(f)(2). See 29 C.F.R. § 1625.10(a)(1), (d)(l)-(3) (EEOC regulation interpreting section 4(f)(2)). Otherwise the inference of age discrimination will be strong — certainly strong enough to defeat a motion for summary judgment by the party having the burden of persuasion, that is, the defendant. The City Colleges tried to prove that the age grading in the Early Retirement Program faithfully reflects the pattern of costs; we must decide whether they succeeded to the level required to deprive the plaintiffs of a trial.

Regarding the sick-leave benefit— which, remember, drops from 80 percent at age 64 to 45 percent at age 65 — the Colleges make two arguments. The first is that sick leave not credited to the lump-sum distribution is credited to length of service and therefore goes to increase the pension component of the retirement benefits. The argument is fine in principle, but there are no figures in the record showing the trade-off between the lump-sum distribution of unused sick leave and the crediting of a fraction of that leave to years of service instead. The second argument is that the older a worker is, the more sick leave he will have accumulated and therefore the greater his distribution will be even if it is fixed at a lower percentage of his final base pay (which anyway will ordinarily be higher than that of a younger worker). This argument ignores two points. The first is that older workers may use more of their sick leave than younger ones, because health problems increase with age. The second and more important point is that sick-leave entitlement is a function of years of service — not age. The Colleges reply that age and length of service are highly correlated, but a document put in evidence by the plaintiffs belies this by showing large discrepancies between age and length of service. Moreover, if the purpose of grading the sick-leave distribution by age is to recognize the greater accumulation of unused sick leave by workers long on service, why don’t the Colleges make the percentages vary with length of service, rather than use age as a crude proxy? And why the big drop at age 65? The record does not answer these questions.

Regarding the cut-off of insurance coverage at age 65, the Colleges are on solider ground in arguing that there is a correlation between age and cost. Insurance premiums are frequently tied to age, life insurance being only the most dramatic example. However, there is no evidence in the record concerning the Colleges’ cost of insurance under its group policies except for life insurance — which indeed is about 50 percent higher for a 69-year-old faculty member than a 64-year-old one, but is only 8 percent higher for a 65 year old than for a 64 year old. And even if it were 50 percent higher for the 65 year old, that would argue for cutting his coverage by 50 percent, not by 100 percent. It is doubtful that the premium costs of the other forms of insurance in the Colleges’ package rise as rapidly with age, and some (e.g., home *320 owner’s) may not rise at all. The cost of health insurance actually falls, because Medicare kicks in at age 65 and the coverage of the group policy is concomitantly reduced.

A feature common to both the sick-pay and insurance components of the Early Retirement Plan is the sharp drop in benefits at age 65. Against the background of the Colleges’ resort to litigation in a futile effort to preserve for a couple of more years its mandatory retirement age of 65, an inference arises that the Colleges were doing everything they could to induce faculty members to retire by 65. The Colleges cannot and do not argue that the drop in benefits at age 65 is justified by the higher cost of benefits to a 65-year-old retiree as compared to a 64-year-old one (because of more accumulated sick leave, valued at a higher base pay, and because of higher insurance costs). They argue that in order to induce early retirement of faculty members in the 65-69 year bracket they have to make 65 a breaking point. They say that if the decline in benefits with age were gradual, as it would have to be to reflect accurately the changing cost of the retirement package, no one would retire before 70. The small annual decline in sick-pay distribution and in insurance coverage would be more than offset by the growth in the pension component of the retirement package as a result of salary raises and additional years of service. So the purpose of the Early Retirement Program — to induce early retirement — would be defeated.

This strikes us as a damaging admission rather than a powerful defense. To withhold benefits from older persons in order to induce them to retire seems precisely the form of discrimination at which the Age Discrimination in Employment Act is aimed. Rather than offering a carrot to all workers 55 years and older, as in the Henn case, the City Colleges are offering the whole carrot to workers 55 to 64 and taking back half for workers 65 to 69. The reason is that the Colleges want to induce workers to retire by 65. In effect they have two early retirement programs; a munificent one for workers 55 to 64 and a chintzy one for workers 65 to 69. It may be that retirees in the second group cost the Colleges more than retirees in the first group, but the difference in cost seems much smaller than the difference in retirement benefits; indeed, there may be no net difference in cost.

As far as the record discloses, the Colleges could achieve all of the lawful objectives of their Early Retirement Program by tying the sick-pay distribution to salary and years of service and the insurance coverage to the actual difference in premium costs that is due to the age of the employee. If so, this is further evidence that the actual Program is a subterfuge, designed to get rid of faculty members when they reach 65.

We sympathize not only with the plight of the City Colleges in an era of declining enrollments and financial stringency, but also with the plight of young academics who cannot find decent teaching jobs because tenured faculty refuse to retire, and with the concerns of those who believe that the quality of American higher education is endangered by the prospect of faculty ger-ontocracies protected by the Age Discrimination in Employment Act. However, the colleges and universities lobbied hard with Congress against the raising of the minimum mandatory retirement age to 70 (and its elimination altogether, effective in 1994), and they are a powerful lobby. They lost, and they cannot be allowed by indirection to reinstitute what was for so long the age-65 mandatory retirement norm.

Whether that is what the City Colleges were trying to do here is a question that must abide the trial that the plaintiffs were erroneously denied. We hold only that the evidence submitted by the defendants in support of their motion for summary judgment does not establish that a reasonable jury could not find that the Early Retirement Program is a subterfuge and therefore violates the Age Discrimination in Employment Act.

Reversed and Remanded.