MEMORANDUM AND ORDER

UNDERHILL, District Judge.

The plaintiffs, a group of retired former employees of the Dun & Bradstreet Corporation (“Dun & Bradstreet”), sued Dun & Bradstreet for allegedly underpaying retirement benefits. Remaining in the case is the sole claim that Dun & Bradstreet’s Master Retirement Plan (“the Plan”) used an unreasonably high discount rate to “actuarially reduce” benefits paid to early-retiring former employees. Dun & Bradstreet has moved for summary judgment. The plaintiffs oppose summary judgment and seek leave to amend their complaint to add a new claim regarding the mortality table used by the Plan. For the reasons given below, Dun & Bradstreet’s motion for summary judgment is *696 granted, and the plaintiffs’ motion to amend their complaint is denied, except to the extent it is made on consent.

I. Facts

The following facts are not subject to any genuine dispute.

The Master Retirement Plan is a defined benefit plan; employees receive a fixed amount of benefits based principally on years of service and earnings. Under the Plan, retired employees whose benefits have “vested” which occurs after five years of service — may begin to receive benefits as early as age fifty-five, ten years earlier than the normal retirement age of sixty-five. If such an early-retiring employee is no longer employed by Dun & Bradstreet, he receives the amount he would have received at age sixty-five reduced by 6.75% per year for each pre-age-sixty-five year and additionally reduced by a mortality factor. If the early-retiring employee is still employed by Dun & Bradstreet at the time of retirement, the employee receives her normal retirement benefit reduced by only three percent per year. The intention of the Plan is to pay the early-retiring former employee the “actuarial equivalent” of his age-sixty-five benefit, i.e., the present value of that payment, and to pay the early-retiring current employee a “subsidized” amount, i.e., an amount greater than the present value of her age-sixty-five benefit.

The plaintiffs in this case are all early-retiring former employees who, as such, receive their retirement benefits reduced by 6.75% per year and a mortality factor, i.e., they receive the actuarially reduced amount, not the subsidized amount.

Over the past two years, the Plan assets have earned a rate of return of 9.63%; over the last year, 15.91%; over the past 10 years, 10.78%; and over the past 15 years, 10.27%. The current thirty-year Treasury bill rate is approximately 4.9%.

II. Summary Judgment

A. Statutes and Regulations

The kind of benefit the Plan provides to early-retiring former employees, like the plaintiffs, is mandated by the Employee Retirement Income Security Act (“ERISA”).

In the case of a plan which provides for the payment of an early retirement benefit, such plan shall provide that a participant who satisfied the service requirements for such early retirement benefit, but separated from the service (with any nonforfeitable right to an accrued benefit) before satisfying the age requirement for such early retirement benefit, is entitled upon satisfaction of such age requirement to receive a benefit not less than the benefit to which he would be entitled at the normal retirement age, actuarially reduced under regulations prescribed by the Secretary of the Treasury.

29 U.S.C. § 1056(a). The Secretary of the Treasury has not, however, prescribed any regulations governing actuarial reduction in this context. The only guidance on the matter comes from a Treasury Regulation regarding prohibited forfeitures, which states:

Certain adjustments to plan benefits such as adjustments in excess of reasonable actuarial reductions, can result in rights being forfeitable.

26 C.F.R. § 1.311(a)-4. In other words, a plan is free to use any method to calculate the appropriate “actuarially reduced” benefit to pay early-retiring former employees, so long as that method is reasonable.

B. Actuarial Reduction

There is no dispute about the appropriate general methodology for determining *697 an actuarially reduced benefit. Retirement benefits paid by an ERISA plan typically commence at age sixty-five. When a vested employee retires early, rather than receiving the full amount of his age-sixty-five payment, he receives the present value of that future payment. Accordingly, the process of actuarial reduction is the process of determining the present value of the benefits the early retiring employee would have received starting at age sixty-five.

Determining the appropriate actuarial reduction requires principally two steps. First, the future benefit is reduced by a “discount rate.” 1 This discount rate reduction reflects the time value of money, i.e., that a dollar paid today is more valuable than a dollar paid next year. Second, the future benefit is reduced by a “mortality factor.” The mortality factor reduction reflects the possibility that the retiring employee might not live to age sixty-five (and so, had he not taken his benefits early, might not have received anything) and is calculated based on a table of average life expectancies.

The dispute in this case concerns the first step — selection of a discount rate. In general, it is not easy to select a discount rate that appropriately reflects the time-value of money. In part, this is because all people and entities do not equally value money paid over time. In other words, there is not one ideal discount rate. Instead, a particular person’s discount rate depends on what the person expects his money to earn over time, and that expectation will vary according to a number of factors such as how risky an investment the person is willing to make and how soon he expects to need his invested money. For example, if Susan is willing to put her money in a risky investment that she expects will earn 30% over a year, one dollar a year from now is worth only around seventy-seven cents to her today, because that is the amount she believes she could invest today to receive a dollar in a year. By contrast, if Sam is only willing to put his money in a conservative investment that he expects will earn 1% a year, he will, today, value one dollar paid a year from now much higher, at around ninety-nine cents.

Consequently, determining the appropriate discount rate for the purpose of calculating actuarially reduced retirement benefits requires answers to two questions: (1) Whose discount rate is to be used? (2) What is that discount rate?

C. Whose Discount Rate

The first question — whose discount rate should be used — has two possible answers: either the Plan’s or the Plan’s participants. Resolution of this question is ultimately one of statutory interpretation. If ERISA’s intention in requiring payment of actuarially reduced benefits is to make plans indifferent to the timing of then-payment liability, then it makes sense to use the Plan’s discount rate. If, on the other hand, ERISA’s intention is to make it equally desirable for employees to retire at an earlier age as at later age, then it makes sense to use the participants’ discount rate.

The plaintiffs argue that the discount rate of participants must be used; Dun & Bradstreet argues that the Plan’s discount rate should be used. I will assume the plaintiffs are correct because, as I explain *698 below, infra n. 5, it makes no difference in this case.

D. What Discount Rate

To actually use the discount rate of all Plan members to calculate early-retirement benefits would be nearly impossible. It would require the selection of a different discount rate for every retiree depending on their investment preferences, risk profile, goals, etc. Neither side suggests that ERISA sets such a task. ERISA only requires that a plan select a methodology that is reasonable. I read this to mean that a plan has met its ERISA obligations with respect to calculation of early benefit payments if it selects a discount rate that is reasonably calculated to be representative of its participants’ average discount rate. The question, then, is what assumptions can a plan reasonably make regarding the average discount rate of its participants.

The plaintiffs argue that, whatever other assumptions are made about participants’ discount rate, the Plan must assume that its participants have effectively zero tolerance for risk. This is so, the plaintiffs claim, because the Plan itself is a defined benefit plan, which guarantees its participants payment of their benefits. In other words, because the Plan is risk-free, it must be assumed that its participants are, on average, risk-averse. The plaintiffs conclude that a reasonable discount rate must, therefore, be based on the rate of return that a plan participant could expect to receive in the market for an effectively zero-risk investment. Such a rate, according to the plaintiffs, is best given by the rate on thirty-year Treasury bills, which is currently at approximately 4.9%. Consequently, the plaintiffs believe that the rate of 6.75% chosen by the Plan is unreasonably high.

I agree with the plaintiffs that it is reasonable to look to the investment characteristics of the Plan itself in attempting to determine a discount rate representative of the discount rate of the average plan participant. After all, every plan participant has at least the following investment characteristic in common: some of his or her retirement savings are invested in the Plan. I do not, however, agree that this premise requires selection of a zero-risk discount rate.

Although zero-risk is a salient feature of the Plan, an equally salient feature is the rate of return on the Plan’s assets. That rate of return does not directly benefit participants — because their benefit is defined — but it indirectly benefits them by controlling the amount of defined benefit the Plan is able to offer in the first place. Accordingly, it is at least as reasonable to assume that a reasonable discount rate would be based on the Plan’s prospective rate of return as it is to assume that a rate would be based on the Plan’s riskiness.

There is no certain way to predict the Plan’s prospective rate of return; there are, however, some indicators in the record. Historically, the Plan assets have earned a rate of return of approximately 10% (although they earned considerably more in the last year). Moreover, in 2002, the Plan’s actuary estimated, for funding purposes, that the Plan’s projected rate of return was 8.25%. Accordingly, available evidence indicates that it would be reasonable to expect the Plan to earn somewhere between an 8% and 10% return on its assets.

Thus, to the extent the Plan’s investment characteristics are reasonable proxies to use in evaluating the discount rate of its participants, there are at least two reasonable ways to estimate a representative discount rate. If the focus is on the riskiness of the Plan, as the plaintiffs suggest, *699 then an appropriate discount rate would be a risk-free discount rate, which presumably is the rate of the thirty-year Treasury bill, around 4.9%. 2 On the other hand, if the focus is on the performance of the Plan, an equally reasonable assumption, then the appropriate discount rate would be the Plan’s projected rate of return, somewhere between 8% and 10%.

I would have a difficult time deciding which of the proposed methodologies is more reasonable. 3 • The truth is that neither is very precise. There is no reason to think that all plan participants are low risk or that all the Plan participants are interested in the specific rate of return achieved by the Plan assets. More likely, some participants like the risk profile of the Plan, some like its return, arid some like both characteristics to varying degrees. In any event, there is no need to decide which of the two proposed methods for choosing a discount rate is the more economically reasonable, because ERISA does not require a plan to pick the most reasonable method of actuarial reduction. 4 ERISA only requires a- plan to pick a reduction method that is not unreasonable. The Plan’s selected rate of 6.75% unquestionably meets this , standard. Although, the rate is above the proposed zero-risk rate of 4.9%, the rate is well, below the 8-10% rate that is the Plan’s approximate projected rate of return. 5 In short, the rate chosen by the Plan is one that no reasonable juror could find unreasonable, and summary judgment for the defendants is appropriate.

III. Motion to Amend the Complaint

A. Procedural History

The original complaint in this case was filed over two years ago, on March 12, 2003. An amended complaint was filed on July 9, 2003. On September 19, 2003, the defendants moved to dismiss counts one, three, and four of the four counts in the *700 amended complaint. At oral argument on this motion, held on February 27, 2004, I asked that the motion to dismiss with respect to count four — the discount rate claim — be taken up on summary judgment after a somewhat expedited discovery schedule. Pursuant to this request, the parties disclosed, and allowed the depositions of, their respective experts, and, on November 19, 2004, the defendants filed for summary judgment on count four, the discount rate claim.

On November 30, 2004, I issued an order granting the defendants’ motion to dismiss counts one and three of the amended complaint. Around that time, the parties informed me that the plaintiffs had agreed to withdraw count two, leaving only count four — the discount rate claim— remaining in the case.

On December 21, 2004, the plaintiffs moved to amend their complaint to add additional parties, dismiss counts one, two and three, correct typographical errors, and “clarify the allegations of the remaining count.” Believing it unopposed, I granted the motion without allowing the defendants time to respond. On January 11, 2005, the defendants moved to vacate my order granting the motion to amend and moved for reconsideration of that order. In their motion, the defendants say that, although they do not oppose most of the proposed amendments, they do oppose one amendment that appears to add a new claim concerning the reasonableness of the mortality table used in the Plan’s actuarial reduction. The plaintiffs do not object to reconsideration of my prior order allowing amendment, but argue that on reconsideration I should still grant them leave to amend. Specifically, the plaintiffs contend that they are not raising a new claim, but simply clarifying their discount rate claim.

B. Law Governing Motions to Amend

Rule 15(a) of the Federal Rules of Civil Procedural instructs courts that leave to amend must be freely given. “If the underlying facts or circumstances relied upon by a plaintiff may be a proper subject of relief, he ought to be afforded an opportunity to test his claim on the merits.” Foman v. Davis, 371 U.S. 178, 182, 83 S.Ct. 227, 9 L.Ed.2d 222 (1962). That does not mean that amendment should always be permitted.

The court plainly has discretion, however, to deny leave to amend where the motion is made after an inordinate delay, no satisfactory explanation is offered for the delay, and the amendment would prejudice the defendant. The burden is on the party who wishes to amend to provide a satisfactory explanation for the delay ....

Cresswell v. Sullivan & Cromwell, 922 F.2d 60, 72 (2d Cir.1990). “[A] proposed amendment ... [is] especially prejudicial ... [when] discovery had already been completed and [non-movant] had already filed a motion for summary judgment.” Krumme v. WestPoint Stevens Inc., 143 F.3d 71, 88 (2d Cir.1998). Moreover, Rule 15(a) is not to be used as a means of avoiding summary judgment on an otherwise non-meritorious claim. See, e.g., Messier v. Southbury Training School, 1999 WL 20907 (D.Conn. Jan. 5, 1999).

C. Discussion

The plaintiffs’ motion to amend seeks to add a new claim, namely, a claim that the mortality table used to calculate benefits paid to early-retiring former employees is unreasonable. The defendants argue that the motion should be denied because it is unfair and prejudicial to allow a new claim to be brought so late in the litigation.

*701 The plaintiffs’ initial response is that this is not a new claim, but merely an elaboration of their claim regarding appropriate actuarial reduction. I do not see it that way. The plaintiffs’ complaint did not allege, in general, an improper actuarial reduction, which might encompass a number of factors, including the mortality table used. It specifically alleged an unreasonable interest rate: “The application of an unreasonable rate of interest works a prohibited forfeiture of benefits under ERISA Section 203(a).” Amended Complaint ¶ 84. It did not claim that the “application of an unreasonable actuarial reduction” worked a forfeiture, and it certainly did not claim that the “application of an unreasonable mortality table” worked a forfeiture. Accordingly, the defendants are correct that a claim about the unreasonableness of the Plan’s mortality table is an entirely new claim. 6

The mere fact that the mortality rate claim is new is not, in and of itself, fatal to the plaintiffs’ motion. On the contrary, courts routinely allow complaints to be amended to state new claims. Nevertheless, to allow an entirely new claim at this point in the litigation would be unfair to the defendants. The plaintiffs did not seek the proposed amendment until after the close of merits discovery 7 and filing of a summary judgment motion. If the amendment is allowed, merits discovery will need to be reopened and the litigation will, in essence, start over — the same experts will likely need to produce new reports and be re-deposed. In the face of this prejudice to the defendants, the plaintiffs have not offered a sufficient explanation for why this amendment could not have been made at least prior to the close of discovery, let alone at the time they brought their initial or first-amended complaint. 8 In other words, the plaintiffs, faced with a strong summary judgment motion, seek to add an entirely new claim with insufficient explanation for the delay in raising that claim. I decline to permit the amendment to the extent it seeks to raise a claim concerning the Plan’s mortality table. The amendment is permitted with respect to those parts to which the defendants have consented.

IY. Conclusion

The defendants’ motion for summary judgment (doc. # 51) is GRANTED. The defendants’ motion to vacate and reconsider my order granting the plaintiffs’ leave to amend their complaint (doc. # 65) is GRANTED. On reconsideration, the plaintiffs’ motion to amend their complaint (doc. # 59) is GRANTED in part and DENIED in part.

*702 The clerk shall close the file.

It is so ordered.

. In their papers, the parties often use the term “interest rate” to describe this rate. Although interest rates and discount rates are related, because the issue here is the rate at which to reduce a given amount of money, not increase it, I think the term "discount rate” is more precise.

. Even if the Plan was required to pick a discount rate based on a zero-risk rate of return, it is not entirely clear to me that today’s thirty-year Treasury rate — which is unusually low — is the only reasonable rate. The Plan was created approximately ten years ago and it is possible that it would be reasonable for the Plan to simply have picked the Treasury rate at that time. In other words, it is not apparent that ERISA requires a plan to continually update its actuarial assumptions. Alternatively, it may be that it would be reasonable for a plan to use some average of historical Treasury rates to come up with a reasonable approximation of what the average rate of return will be for employees retiring at any point during the life of the plan. Quite possibly any of these methodologies would count as "reasonable” for ERISA purposes (and it appears that under either of these two methodologies the Plan’s rate of 6.75% would be appropriate), but there is no need to reach the question because I hold that ERISA does not mandate the use of a zero-risk discount rate.

. The plaintiffs suggest that the Treasury rate should be preferred because that is the rate mandated- in other contexts, such as calculation of "lump sum” benefits. See Internal Revenue Code § 417(e)(3). That argument just as easily cuts the other way; the fact that Congress chose to mandate that rate in some cases, but did not mandate it here, could be read as expressing a desire not to require that rate in this context.

. It should also be apparent that I am not holding that these are the only two methodologies for selecting a reasonable discount rate. Others simply have not presentéd themselves in this case.

. Because the Plan’s discount rate falls below its projected rate of return, that discount rate is certainly reasonable if the appropriate discount rate is to be selected from the perspective of the Plan, as suggested by the defendants. By definition the Plan’s discount rate' is equal to its projected rate of return, and, therefore, if anything, a rate of 6.75.% is, from the Plan’s perspective, generous to the plaintiffs.

. If all the plaintiffs desire is that I consider their expert’s evidence concerning the Plan’s mortality tables in deciding the issue of the discount rate, then the issue is easily resolved. I have considered the evidence concerning the mortality tables and find it to be irrelevant to the question whether the discount rate selected by the Plan is reasonable, a question which I believe is adequately addressed in the first half of this opinion.

. It is true that discovery was only conducted on count four of the complaint, but that is because the remaining counts were either dismissed or withdrawn. Additionally, there is a motion pending for certification of a class, which might have required additional discovery. That motion is moot, however, in light of this order.

.The plaintiffs argue that their expert’s report and some of the questions asked by their counsel at the deposition of the defendants’ expert should have put the defendants on notice of this claim. I disagree. The plaintiffs’ amended complaint was very specific in only challenging the discount rate used. When a complaint accurately and specifically sets out its claim, I do not think a defendant is required, based on the plaintiff's conduct in litigation, to guess at what other claims the plaintiff might intend to bring in the future.