The premise of the Norris ruling against retroactivity--that equalization of women's benefits requires the employer or the plan to pay out extra sums of money--is inapplicable to the case before us. This is so because of the fundamental difference between the plan in Norris and the TIAA-CREF arrangements. In Norris the Arizona Deferred Compensation Plan provided sufficient certainty concerning the amount of annuity payments to enable the District Court to calculate, long before the plaintiff's retirement, the amount of her monthly annuity. 486 F. Supp. 645, 648. It was this expectation of a determinable benefit that the Supreme Court majority in Norris did not wish to have jeopardized by imposing added financial burdens on the plan. As Justice O'Connor observed, "Many working men and women have based their retirement decisions on expectations of a certain stream of income during retirement. These decisions depend on the existence of adequate reserves to fund these pensions." 103 S. Ct. at 3512 (O'Connor, J., concurring). By contrast, the TIAA-CREF plans do not guarantee retirees "a certain stream of income." As we previously noted, CREF "guarantees plan participants no specified amount of monthly payments. The TIAA-CREF publication which explains the CREF system states that: 'The CREF annuity pays you a retirement income that varies year by year, reflecting primarily the experience of the securities in CREF's portfolio.' " 691 F.2d at 1068-69. Similarly, the TIAA benefit, based on investment results, is not ascertainable. Id. at 1068. Thus, with one possible exception to be noted below, the limited form of retroactive relief in the District Court's judgment need not be rejected because of any fear that either the employer or TIAA-CREF will be burdened with any additional financial obligations.

However, noting the absence of new burdens upon the employer or the plan does not automatically mean that we may reinstate our prior decision. The retroactive aspect of the District Court's judgment will unquestionably have an adverse economic impact. The elimination of gender-distinct mortality tables and the consequent equalization of benefits for similarly situated men and women will mean that male retirees, as a class, will receive less dollars than whatever amounts they would have received if gender-distinct tables continued to be used to determine the portion of benefits derived from pre-judgment contributions. The significance of this reduction depends on whether it is considered in the aggregate or individually. The American Council of Life Insurance and the Health Insurance Association of America, amici curiae, point to an aggregate sum of $2 billion that will be effectively transferred from male to female annuitants by the use of unisex tables. Brief for Amici Curiae at 14. On the other hand, our prior opinion noted that for 60% of the men the use of unisex tables will have no consequence because they elect a joint-survivor option that continues benefits for the lifetime of their surviving wives and that for the remaining 40% of the men use of unisex tables will mean a reduction of between 1% and 8% of whatever benefits they would otherwise receive. 691 F.2d at 1069 n. 12. TIAA-CREF contend that Norris proscribes retroactivity in this case because of the adverse consequence to some of the male annuitants. The plaintiff, the EEOC, and the intervenor AAUP (representing thousands of male and female members) disagree, arguing that reduction in benefits to males is not inequitable in the absence of any settled expectation of the males concerning the level of their future benefits.

After careful consideration of the majority opinion in Norris on the issue of relief, we conclude that the Court did not intend to bar retroactivity in the circumstances of this case. The entire thrust of Justice Powell's opinion focuses on the burdens that retroactivity would impose upon the employer or the plan. The prospect of adverse consequences to male annuitants is not mentioned, presumably because the majority was satisfied that the District Court's judgment in that case would not result in reduction of benefits to any male; equalization was to be achieved solely by "topping up" the benefits to females. The absence from the majority opinion on relief of any discussion of benefit levels of male annuitants is especially striking in view of the explicit consideration of this matter by the Justices who joined that portion of Justice Marshall's opinion dissenting on the issue of relief. 103 S. Ct. at 3502-04. Justice Marshall, for the four-member dissent, expressed the view that the availability of retroactive relief should turn on whether unisex tables could be applied to that portion of benefits attributable to contributions made prior to the Manhart decision "without violating the male employee's contractual rights." Id. at 3504. He would have remanded for determination of whether male participants "had any contractual right to a particular level of benefits that would have been impaired by the application of sex-neutral tables to their pre-Manhart contributions." Id.

Admittedly, the absence from the majority opinion of any mention of the dissent's preferred resolution of the case leaves us somewhat uncertain as to the meaning of the Norris decision. It is possible that the majority considered the dissent to be in error in thinking that the record was uncertain as to whether male annuitants had contractual rights to specified benefit levels. Perhaps the majority thought it absolutely clear that the male annuitants had such rights. That seems to be the more plausible interpretation; otherwise, it is difficult to understand why the opinion of Justice Powell explicitly assumed that "topping up" was a necessary consequence of the District Court's judgment. On the other hand, it is also possible that the majority considered the dissent to be irrelevant, that the majority intended to bar any form of retroactivity whether or not contractual rights of the male annuitants would be defeated by use of unisex tables. That seems to be the less plausible interpretation; it is difficult to imagine why Justice Powell's opinion was so emphatic in cautioning against the imposition of financial burdens on employers and plans if in Norris it was contractually possible to make retroactive use of unisex tables and thereby equalize benefits without imposing any financial burdens on the employer or the plan.

If, as TIAA-CREF contends, the Norris majority intended to bar retroactivity in all annuity cases simply because equalization of benefits must inevitably burden someone, we would have expected to see some intimation of that view in Justice Powell's opinion. Instead, we see only a prohibition of relief provisions that impose added financial burdens on employers or plans. If the Supreme Court were confronted with a relief provision that reduced the benefits of male annuitants from specified benefit levels (or levels readily calculable, for example, from an annuitant's recent salary), it might well consider such provisions inequitable. And inequity might also arise if a group were so female dominated that retroactive use of unisex tables would cause a substantial reduction even of unspecified benefit levels of males. But we see nothing in Norris that proscribes retroactive provisions simply because unspecified benefit levels for some male annuitants will be slightly lower than whatever they would have been under gender-distinct tables.

In our prior decision we explicitly considered whether the retroactive aspect of the District Court's judgment would be inequitable to those male annuitants who would receive lower benefits. We agreed with the District Court that it would not because "no employees could have had settled expectations as to the amount of monthly benefits" and the references in TIAA-CREF literature to the impact of gender upon benefit levels "are neither so clear, nor are they highlighted in such a way, that it is plausible to think that male plan participants relied upon them in any meaningful way." 691 F.2d at 1069. We also noted that any reliance on gender-distinct tables was "unjustified in light of the visibility of the issue of the legality of using such tables, at least since it was brought to the attention of all participants in 1973." Id. This assessment of the equities remains unaltered by the decision in Norris.

In one minor respect, however, we think it would be prudent to modify our prior decision. Though the benefit level for TIAA participants depends primarily on the investment success of the TIAA portfolio, TIAA guarantees that benefit levels will reflect at least a 2 1/2% return on investments. In our prior decision, we noted that this minimum guarantee was so low "that there is no danger that the male participants' expectation that they will receive the minimal guaranteed benefit will be jeopardized by the relatively minor changes ... necessitated by the relief ordered." 691 F.2d at 1068. We still consider the likelihood that TIAA will fail to earn 2 1/2% on its investment to be an insignificant risk. However, since Norris appears to foreclose any possibility of the retroactive imposition of added financial burdens upon employers or plans, we will direct the District Court to modify its judgment to provide that unisex tables need not be used in calculating the portion of benefits attributable to pre-judgment contributions to whatever extent may be necessary in any year to ensure that the use of such tables will not impose added financial burdens upon the employer or TIAA beyond those resulting from the obligation to pay benefits reflecting a 2 1/2% return on investment. We leave it to the parties to suggest to the District Court appropriate language for precluding this unlikely eventuality.

The judgment of this Court rendered September 29, 1982, is reinstated, subject only to the modification set forth in the preceding paragraph. As thus modified, the judgment of the District Court is affirmed.