EASTERBROOK, Chief Judge.

The Exelon Corporation Cash Balance Pension Plan is a defined-benefit plan that works like a defined-contribution plan, except that the individual accounts are virtual. All of the Plan's assets are held in a single trust; the Plan does not have a separate pot of assets to match each employee's account. Cooper v. IBM Personal Pension Plan, 457 F.3d 636 (7th Cir. 2006), and Berger v. Xerox Corp. Retirement Income Guarantee Plan, 338 F.3d 755 (7th Cir. 2003), discuss more fully the nature of a cash-balance plan.

Many pension plans, including Exelon's, give workers the option of taking a lump-sum distribution when they quit or retire. A defined-contribution plan just turns over the balance of the account. 29 U.S.C. § 1002(23)(B). A defined-benefit plan operates under different rules — or did until 2006, when the addition to ERISA of 29 U.S.C. § 1053(f) brought the treatment of lump-sum distributions into harmony. Our plaintiff, Thomas Fry, left Exelon in 2003, so we describe the former approach, which required pension plans to start with the current balance and add any contractually promised interest (or any other form of guaranteed increase in benefits) through the employee's "normal retirement age." The plan then discounted the resulting number to present value using the "annual rate of interest on 30-year Treasury securities for the month before the date of distribution." 29 U.S.C. § 1055(g)(3), incorporated by 29 U.S.C. § 1053(e)(2).

This process was designed to ensure the actuarial equivalence of the lump-sum payment and the pension available at retirement. But, if the Treasury rate does not match the market return, the process misfires. Berger describes the mechanics. If the Treasury rate is less than a plan's annual guarantee — as it normally will be, because Treasury bonds have very little risk, and a correspondingly low rate of return — the lump sum balloons (a 3.5% difference in the rates doubles the cash paid out to someone who leaves at 45 and does not plan to retire until 65). If the Treasury rate exceeds the plan's guarantee, as it may during a time when the stock market is in decline, the lump sum shrinks accordingly. For most of the 1990s and 2000s, the Treasury rate was below the guarantees offered by cash-balance plans. This gave employees a big incentive to quit early and claim lump-sum distributions; it also encouraged pension plans to reduce their promised annual returns, which hurt all employees (not just those who planned a strategic early departure).

The 2006 amendment fixed the problem for all cash-balance plans. It also avoided the uncertainty inherent in a need to estimate what rates of return lie in the future. (Did anyone in 2003 predict accurately that the stock market as a whole would rise from 2004 through 2007 but plummet in 2008?) Many plans, of which Exelon's was an example, had applied a self-help fix. When it was established in 2002, Exelon's Plan provided that each employee's "normal retirement age" arrived after five years on the job. This was also the Plan's vesting date, and thus the first opportunity to demand a lump-sum distribution when leaving for other employment. Because ERISA required the addition of interest (and discounting at the Treasury rate) only through each participant's "normal retirement age," this enabled Exelon's Plan to avoid the entire adjustment process and distribute the balance of the worker's virtual account just as a defined-contribution plan would distribute the balance of an actual account.

Thomas Fry opted into the Exelon cash-balance Plan when it was created in 2002. His virtual account was funded initially with the actuarial value of his traditional defined-benefit pension. Exelon contributes to the Plan 5.75% of each participant's annual compensation, and it adds annual interest (called "investment credits") at the greater of 4% or an average of the 30-year Treasury bond rate and the average return on the Standard Poors 500 index. Fry quit in 2003, at age 55, after working more than five years at Exelon. He asked for and received the value of his account, more than $500,000, and filed this suit because the Plan gave him just the balance — rather than the balance plus "investment credits" through 2013 (when he will turn 65), discounted to present value at the Treasury rate (which was 5.16% in October 2003, the month before Fry retired). His suit contends that the Plan's definition of "normal retirement age" is invalid and that he is entitled to credits through age 65; the district court held, however, that the Plan satisfies ERISA's requirements. 2007 U.S. Dist. LEXIS 65355, 2007 WL 2608524 (N.D.Ill. Aug. 31, 2007).

Fry makes much of the fact that the Plan's definition of "normal retirement age" is designed to work around the augment-and-discount process required by the pre-2006 version of § 1053(e)(2)(B). He calls this an "evasion"; the Plan calls it careful design. No matter. Names do not decide concrete cases. Employers are entitled to vary by contract those aspects of pension plans ERISA makes variable, and they may act in their own interest when doing so, see Lockheed Corp. v. Spink, 517 U.S. 882, 116 S.Ct. 1783, 135 L.Ed.2d 153 (1996), just as participants are entitled to the benefit of terms (such as vesting rules) that the law makes immutable. Lowering the normal retirement age means that lump-sum distributions may be smaller, but it has benefits for the workers, such as accelerating the application of the anti-forfeiture clause, which like § 1053(e)(2)(B) before 2006 is keyed to the plan's "normal retirement age". See Contilli v. Teamsters Local 705 Pension Fund, 559 F.3d 720 (7th Cir. 2009).

How much discretion employers enjoy when selecting a "normal retirement age" depends on the language of ERISA, for the phrase is a defined term:

The term "normal retirement age" means the earlier of —

(A) the time a plan participant attains normal retirement age under the plan, or

(B) the later of —

(i) the time a plan participant attains age 65, or

(ii) the 5th anniversary of the time a plan participant commenced participation in the plan.

29 U.S.C. § 1002(24). Exelon says that subsection (A) allows it to define "normal retirement age" as it pleases. Fry insists that the statute implies restrictions: first that the "normal retirement age" be an age rather than a different measure (such as $20,000 or 175 pounds); and second that it be "normal" (which is to say, the mean or median for retirement at the firm, rather than age 25 or the age of the incumbent President).

Fry's first argument flops because the Plan's formula — the participant's age when beginning work, plus five years — is an "age." It is employee specific, to be sure, but "age + 5" remains an age. It is not as if the Plan provided that "an employee reaches normal retirement age when he owns ten umbrellas." The Plan's formula not only specifies an "age" but also is lifted right out of the statute. Subsection (B)(ii) defines as the highest possible "normal retirement age" (for a person hired at 65 or older) "the 5th anniversary of the time a plan participant commenced participation in the plan." Making that statutory definition of "normal retirement age" universally applicable can't be rejected on the ground that the formula does not yield an "age." ERISA does not require the "normal retirement age" to be the same for every employee; § 1002(24)(B)(ii) shows that too.

As for the argument that five years on the job is not the "normal" retirement age: § 1002(24) does not compel a pension plan's retirement age to track the actuarial tables. If it did, then instead of granting discretion to the plan's sponsor the statute would read something like: "The term `normal retirement age' means the median age at which participants in the plan retire." But the statute does not say this, nor does it say that the "normal retirement age" must be at least 62 but cannot exceed 65. Some industries have much younger retirement ages — under 30 for football and under 40 for futures commission merchants. The statutory cap at age 65 itself requires some departure from normal practices at law firms, universities, and other employers where people work past the time when they can start drawing full Social Security benefits (which for those approaching retirement today is 66 rather than 65).

Under § 1002(24)(A) an age is the "normal retirement age" because the plan's text makes it so. The age in the plan is "normal" in the sense that it applies across the board, to every participant in the plan. (It is important to understand that a "normal retirement age" in a pension plan does not control when employees must retire, but only when certain rights vest and how benefits are adjusted. That's why it makes sense to speak of an age being "normal" to the plan's operation rather than to anyone's retirement prospects.)

In 2007 the Treasury Department issued a regulation providing that a plan's "normal retirement age" must be "reasonably representative of the typical retirement age for the industry" (subject to safe harbors for employers with unusually early or late retirement patterns). 72 Fed.Reg. 28604, 28606 (May 22, 2007), amending 26 C.F.R. § 1.401(a)-1(b). Fry contends that this regulation supports his position. It does not, because like almost all regulatory changes it operates only prospectively. See § 1.401(a)-1(b)(4). See also Bowen v. Georgetown University Hospital, 488 U.S. 204, 109 S.Ct. 468, 102 L.Ed.2d 493 (1988). The commentary accompanying this regulation acknowledges that employer choice had been honored in pre-2007 years. 72 Fed.Reg. at 28605. The regulation therefore does not affect the calculation of lump-sum distributions in 2003. (Whether the regulation is within the scope of the agency's interpretive discretion, see Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984), is a question not presented here.)

Fry chastises the district court for disagreeing with Laurent v. Pricewaterhouse-Coopers LLP, 448 F.Supp.2d 537 (S.D.N.Y. 2006). Decisions of district judges lack authoritative force in or outside their districts; they have persuasive weight only. The district judge in Laurent was trying to read the tea leaves in Esden v. Bank of Boston, 229 F.3d 154 (2d Cir. 2000). And Esden itself did not interpret § 1002(24). It dealt with the same subject as Berger: how a cash-balance plan calculates actuarial value when an employee elects a lump-sum distribution before a given plan's "normal retirement age." As far as we can tell, ours is the first appellate opinion on the interaction between § 1002(24) and the pre-2006 requirement of actuarial adjustments to the hypothetical balance. Instead of guessing how other judges might approach the subject, we have analyzed the statutory language directly. And on the understanding of "normal" that we have just explained, the statutory language allows employers to specify a "normal retirement age" that differs from typical retirement patterns.

AFFIRMED