Cooper, Kathi v. IBM Personal Pension

457 F.3d 636, 19 A.L.R. Fed. 2d 729, 2006 U.S. App. LEXIS 20128, 104 Fair Empl. Prac. Cas. (BNA) 151, 2006 WL 2243300, 98 A.F.T.R.2d (RIA) 5884, 38 Employee Benefits Cas. (BNA) 1801

Opinion

EASTERBROOK, Circuit Judge.

The IBM Personal Pension Plan is a cash-balance defined-benefit plan. It is almost, but not quite, a defined-contribution plan. Although each employee in a defined-contribution plan has a fully funded individual account, the personal account in a cash-balance plan is not separately funded. Instead IBM imputes value to the account in the form of “credits”: there are pay credits (set at 5% of the employee’s gross taxable income) and interest credits (set at 100 basis points above the rate of interest on one-year Treasury bills). A trust holds assets that may (or may not) be enough to fund all of the individual accounts when workers quit or retire. IBM’s plan permits an employee who quits or retires after working long enough for pension benefits to vest (a maximum of five years) to withdraw the balance in cash or roll it over into a fully funded annuity. During the time before cash-out the employee takes the risk that IBM will suffer business reverses and be unable to pay the full stated value of the account (if the amount already in trust for participants as a group turns out to be insufficient); otherwise IBM’s plan is economically identical to a defined-contribution plan funded the same way and invested in a bond fund that returns 1% above the Treasury rate.

Plaintiffs in this class-action litigation contend that IBM’s plan violates a subsection of ERISA (the Employee Retirement Income Security Act) that prohibits age discrimination. The district court ruled in plaintiffs’ favor, see 274 F.Supp.2d 1010 (S.D.Ill.2003), and proceedings continued as the parties debated how much IBM owes (and how it must change its plan in future years) as a remedy. That subject has been resolved to mutual satisfaction— contingent on the district judge being right on the merits — so this appeal is limited to the question whether the plan is unlawfully discriminatory.

All terms of IBM’s plan are age-neutral. Every covered employee receives the same 5% pay credit and the same interest credit per annum. The basis of the plaintiffs’ challenge — and the district court’s holding — is that younger employees receive interest credits for more years. The language on which plaintiffs rely was added to ERISA in 1986; Congress also enacted a parallel provision covering defined-contribution plans. Pub.L. 99-509, 100 Stat. 1874, 1975, 1976 (1986). We set these out alongside to facilitate comparison:

Defined-benefit plans: erisa § 204(b)(l)(H)(i), 29 U.S.C. § 1054(b)(l)(H)(i) Defined-eontribution plans: erisa § 204(b)(2)(A), 29 U.S.C. § 1054(b)(2)(A)

[A] defined benefit plan shall be treated as not satisfying the requirements of this paragraph if, under the plan, an employee’s benefit accrual is ceased, or the rate of an employee’s benefit accrual is reduced, because of the attainment of any age.

A defined contribution plan satisfies the requirements of this paragraph if, under the plan, allocations to the employee’s account are not ceased, and the rate at which amounts are allocated to the employee’s account is not reduced, because of the attainment of any age._

These appear to say the same thing, except that the rule for defined-benefit plans tells us what is not allowed, while the rule for defined-eontribution plans tells us what works. Either way, the employer can’t stop making allocations (or accruals) to the plan or change their rate on account of age. The IBM plan does neither of these things and therefore, one would suppose, complies with the statute. If this were a real, rather than a phantom, defined-contribution plan, that much would be taken for granted. Yet if the 5%-plus-interest formula is non-discriminatory when used in a defined-eontribution plan, why should it become unlawful because the account balances are book entries rather than cash?

Plaintiffs persuaded the district court, however, that the two subsections are radically different. That difference is attributable to the phrase “benefit accrual,” which appears in the subsection for defined-benefit plans but not the one for defined-eontri-bution plans. Neither ERISA nor any regulation defines this phrase, so the district judge went looking for some equivalent elsewhere in the statute. It found the phrase “accrued benefit,” which is defined in § 3(23)(A), 29 U.S.C. § 1002(23)(A). An “accrued benefit” is an amount “expressed in the form of an annual benefit commencing at normal retirement age.” Plug this back into § 204(b)(l)(H)(i), and the rule against discrimination then refers not to what IBM puts into the plan, but what the employee takes out on retirement. Someone who leaves IBM at age 50, after 20 years of service, will have a larger annual benefit at 65 than someone whose 20 years of service conclude with retirement at age 65. The former receives 15 years’ more interest than the latter — and the judge assumed that this is not counterbalanced by the fact that older workers generally draw higher salaries. Under the district court’s analysis, compound interest becomes a scourge, for the younger the employee when any given year’s salary is earned, the greater the payout “expressed in the form of an annual benefit commencing at normal retirement age.”

This approach treats the time value of money as age discrimination. Yet the statute does not require that equation. Interest is not treated as age discrimination for a defined-eontribution plan, and the fact that these subsections are so close in both function and expression implies that it should not be treated as discriminatory for a defined-benefit plan either. The phrase “benefit accrual” reads most naturally as a reference to what the employer puts in (either in absolute terms or as a rate of change), while the defined phrase “accrued benefit” refers to outputs after compounding. That’s where this litigation went off the rails: a phrase dealing with inputs was misunderstood to refer to outputs. As long as we think of “benefit accrual” as referring to what the employer imputes to the account — an understanding reinforced by the use of the word “allocation” in the subsection addressing defined-contribution plans — there is no statutory difference between the treatment of economically equivalent defined-benefit and defined-contribution plans. For defined-benefit plans, where the account is an accounting entry rather than cash, “benefit accrual” matches the money “allocated” to a defined-contribution plan.

Nothing in the language or background of § 204(b)(l)(H)(i) suggests that Congress set out to legislate against the fact that younger workers have (statistically) more time left before retirement, and thus a greater opportunity to earn interest on each year’s retirement savings. Treating the time value of money as a form of discrimination is not sensible. Cf. Hazen Paper Co. v. Biggins, 507 U.S. 604, 611, 113 S.Ct. 1701, 123 L.Ed.2d 338 (1993) (variables correlated with age must be kept “analytically distinct” from age when searching for discrimination). The parallel between the subsections for defined-benefit and defined-contribution plans shows as much. All sorts of things go wrong unless we treat both § 204(b)(l)(H)(i) and § 204(b)(2)(A) as addressing the rate at which value is added (or imputed) to an account, rather than the annual pension at retirement age. Consider a formula that adds $500 to annual pension benefits for every year worked. This setup economically favors older workers (as does a top-five-year-salary multiplier) because it ignores the time value of money and treats the last year of work as contributing equally with the first. Yet on the plaintiffs’ understanding it would be a form of discrimination against the old, because the $500 is a smaller “rate of benefit accrual” as time passes. Someone who worked two years would get a “100% rate of increase” in his second year (his annual benefits on retirement would jump from $500 to $1,000), while someone who works 20 years receives only a 2.56% increase (from $19,500 to $20,000) in Year 20.

Our conclusion that “benefit accrual” (for defined-benefit plans) and “allocation” (for defined-contribution plans) both refer to the employer’s contribution rather than the time value of money between contribution and retirement has the support of regulations that the Treasury Department proposed. (Appropriations riders have prevented the Treasury from taking final action on the draft regulations, but they still help to inform our understanding of the statute.) The draft regulations treat the “rate' of benefit accrual” as “the increase in the participant’s accrued normal retirement benefit for the year.” 67 Fed.Reg. 76123, 76125 (Dec. 11, 2002). For cash-balance plans in particular the phrase would have been defined as “the additions to the participant’s hypothetical account for the plan year.” 67 Fed.Reg. at 76126. The draft specifies that interest credits must accrue “at a reasonable rate of interest that does not decrease because of the attainment of age” and be provided “for all future periods, including after normal retirement age”. Ibid.

Thus the Treasury’s view, like our independent reading, looks at the rate of contribution (what goes into the account) rath er than the annual rate of withdrawal at retirement. The Treasury asks: “if this employee were younger, would the hypothetical balance have grown more this year?” For IBM’s plan, the answer is no. When an employee is 20, his account is increased monthly by 5% of his salary plus credits based on the account’s balance multiplied by a prescribed interest rate. The same is true when the employee turns 50, 60, or 70. Neither the contribution rate nor the interest rate changes with age. (No one contests the rate at which interest is credited — though 100 basis points over the Treasury rate looks low compared with what investments could earn in the private market and therefore could be understood to disfavor younger employees.)

Plaintiffs propose a different definition of “benefit accrual”: annual pension (at retirement age) divided by salary. Contributions at age 25 produce an annual pension benefit that equals about 3% of each year’s income in nominal dollars, while contributions at 65 produce annual benefits only 0.4% of the year’s income. Not surprising, and hardly illuminating; this is just another illustration of the power of compound interest. The example depends on (a) allowing the pension to be increased by compounding, while (b) failing to discount the pension to present value. Stated another way, this example compares 1966 dollars (when the 25-year-old earned the salary) with 2006 dollars (when that person turns 65 and starts to draw a pension); of course the return on investment looks good, but much of it is inflation and the rest is real interest rates. The person who earns a salary in 2005 and retires in 2006 is (materially) unaffected by both inflation and compound interest. Nothing depends on age. Someone who earns a salary at age 65 and waits 40 years to start drawing a pension would receive the same 3%-of-salary-per-retirement-year as the illustrative 25-year-old who retires at 65; and someone who earns wages at 25 and retires the next year gets the same 0.4%-of-salary annual pension, if not less (because of discounting).

Much of the plaintiffs’ argument rests on the idea that the account of a 25-year-old worker does not get 5% plus periodic interest, but instead is immediately credited with 5% of salary plus 40 years’ interest. That makes the contributions look discriminatory: the 25-year-old worker’s account receives 40 times as much interest credit in the year the contributions accrue as a 65-year-old worker’s account. Once again, however, this perspective misunderstands both the statute and the time value of money.

Nothing in either ERISA or the IBM plan requires 40 years of interest to be credited to the account as soon as the young worker earns wages. What plaintiffs have in mind is the rule that, when any beneficiary (young or old) elects to take a cash distribution or roll the account into an annuity before reaching age 65, the plan must distribute a lump sum calculated to be the “actuarial equivalent” of the annuity that would be available at normal retirement age. ERISA § 204(c)(3), 29 U.S.C. § 1054(c)(3). To derive the “actuarial equivalent” of a pension at age 65, a plan must (a) add all interest that would accrue through age 65, then (b) discount the resulting sum to its present value. Berger v. Xerox Corp. Retirement Income Guarantee Plan, 338 F.3d 755, 762-63 (7th Cir.2003). Plaintiffs characterize step (a) as extra interest credits for the young, but they ignore step (b). The discount rate may be close to (if it does not exceed) the rate at which interest is imputed, so the amount paid out in cash may be close to if not below the nominal balance in the account. Berger did not consider § 204(b)(1)(H)(i) and does not hold that the process of grossing up the balance and then discounting is a form of age discrimination. To the contrary: Berger described this process as the means to avoid age discrimination.

As far as we can see, ours is the first appellate decision to address the status of cash-balance plans under § 204(b)(l)(H)(i). The class directs our attention to two decisions from other circuits that it says supply helpful analysis. Miller v. Xerox Corp. Retirement Income Guarantee Plan, 447 F.3d 728 (9th Cir.2006); Esden v. Bank of Boston, 229 F.3d 154 (2d Cir.2000). As the class reads them, these opinions stand for two important propositions. First, that an “accrued benefit” in a cash-balance plan is an annuity at normal retirement age. Second, that there is a “fundamental” distinction between defined-contribution and defined-benefit plans. Both of these propositions are correct, and both of them are irrelevant.

Start with the first proposition. What the true meaning of “accrued benefit” may be is not controlling; § 204(b)(1)(H)® does not use that phrase, and we have explained why “benefit accrual” means something other than “accrued benefit.” Once we start to calculate accrued benefits for people who quit or retire early, it is necessary to impute extra interest and discount. Berger describes how this is done. But “benefit accrual” refers to the annual addition to the pot, not to the final payout. The holding of Esden, see 229 F.3d at 168, mirrors that of Berger and requires no further comment. Miller extends the framework of Esden and Berger to evaluate reductions in notional account balances caused by previous lump-sum distributions. 447 F.3d at 733-34. See 26 C.F.R. § 1.411(a) — 7(d)(6) (offset may equal the “accrued benefit attributable to the distribution”).

As for the second proposition, plaintiffs cite language describing the regulatory framework as “rigidly binary” (Es-den, 229 F.3d 154,159 n. 6). True enough. A defined-contribution plan entails fully funded individual accounts; everything else is a defined-benefit plan. 29 U.S.C. § 1002(25). Miller adds that “interest credits are defined benefit entitlements ... and are not analogous to the investment growth of a defined contribution plan”. 447 F.3d at 735. Maybe so; that depends on what the court meant by “analogous.” Interest in a defined-contribution plan is real, while interest credits in a defined-benefit plan are bookkeeping entries (effectively debt obligations of the employer to the extent that the pension trust fund does not cover them). But so what? IBM does not contend that its plan is governed by § 204(b)(2)(A), just that § 204(b)(1)(H)® does not whimsically require a court to find age discrimination for a defined-benefit plan when materially identical statutory language allows functionally identical defined-contribution plans to operate without any taint of discrimination. To say that defined-benefit and defined-contribution plans are governed by different subsections of ERISA is not to say that what is lawful for one must be forbidden to the other. We conclude that § 204(b)(1)(H)® and § 204(b)(2)(A) indeed provide similar treatment with respect to claims of age discrimination.

Only a brief mention of Arizona v. Norris, 463 U.S. 1073, 103 S.Ct. 3492, 77 L.Ed.2d 1236 (1983), another mainstay of the class’s briefs, is called for. In Norris women making the same pension contributions as men, all else held equal, had lower annual pension benefits. Id. at 1082, 103 S.Ct. 3492. Under IBM’s plan any differences in pension benefits are a function of differing years of service, salary history, or the years the balance has been allowed to compound; age is not a factor.

Here, as so often, it is essential to separate age discrimination from other characteristics that may be correlated with age. That was the Supreme Court’s point in Hazen Paper: wages rise with seniority (and thus with age) at many employers, but distinctions based on wage levels (in order to reduce a payroll) do not “discriminate” by age. See also, e.g., Achor v. Riverside Golf Club, 117 F.3d 339, 341 (7th Cir.1997) (need to show “the effect of age, isolated from other influences”); Sheehan v. Daily Racing Form, Inc., 104 F.3d 940, 942 (7th Cir.1997) (expert should correct for “potentially explanatory variables other than age” when determining whether discrimination was because of age). While those decisions involved different statutory regimes, the objective is general: a plaintiff alleging age discrimination must demonstrate that the complained-of effect is actually on account of age. One need only look at IBM’s formula to rule out a violation. It is age-neutral.

Now this may give a clue about why plaintiffs (a class of older workers) have sued. The cash-balance plan replaced a more traditional arrangement under which the annual pension on retirement was a function of closing salary (say, an average of the last five years before retirement) multiplied by the number of years of service. Because wages rise with seniority, such a formula favors older workers. Workers who quit or retire early not only miss out on the rising wages that accompany seniority but also don’t receive credit for the time value of money: years of service at age 30 and at age 70 count equally in the traditional formula. Many employers design such back-loaded systems to give the most-experienced members of the labor force a reason (even beyond rising salary) to stay on the job. Like a defined-contribution plan, a cash-balance plan removes the back-loading of the pension formula; older workers (accurately) perceive that they are worse off under a cash-balance approach than under a traditional years-of-serviee-times-final-salary plan. But removing a feature that gave extra benefits to the old differs from discriminating against them. Replacing a plan that discriminates against the young with one that is age-neutral does not discriminate against the old.

There is a transition issue. Wben IBM moved employees from a more traditional plan to a cash-balance system, it gave them the greater of the present value of their pension entitlements as of the transition date or the account balance that they would have had if IBM had a cash-balance plan in effect since the employee came to work. Plaintiffs complain that this gives the younger workers too much credit for interest (because more time passes between their work and retirement dates), but that rehashes arguments that we’ve already rejected. Every employee with the same salary and service record receives the same opening account balance under the new plan. That the change disappointed expectations is not material. An employer is free to move from one legal plan to another legal plan, provided that it does not diminish vested interests— and this transition did not. Cf. Lockheed Corp. v. Spink, 517 U.S. 882, 116 S.Ct. 1783, 135 L.Ed.2d 153 (1996) (employers are not fiduciaries under ERISA when they create or amend pension plans).

Litigation cannot compel an employer to make plans more attractive (employers can achieve equality more cheaply by reducing the highest benefits than by increasing the lower ones). It is possible, though, for litigation about pension plans to make everyone worse off. After the district court’s decision IBM eliminated the cash-balance option for new workers and confined them to pure defined-contribution plans. See Ellen E. Schultz & Theo Fran cis, How Safe Is Your Pension? —Freeze of IBM Plan Leaves Workers Worrying If Their Employer Is Next, Wall St. J., Jan. 12, 2006, at Dl; Ellen E. Schultz, IBM to Exclude New Workers From Its Cash-Balance Pension — Move to Only 401(k) Plans Foliotes Legal Questions, Wall St. J., Dec. 9, 2004, at A2. Whether that is good or bad (for employees or society as a whole) is not for us to say. What we can and do conclude, however, is that the decision may again be made freely, governed by private choice rather than legal constraint.

The judgment of the district court is reversed, and the case is remanded with directions to enter judgment in IBM’s favor.