Research Brief

For many years, the Defense Department funded military retirement on a "pay-as-you-go" basis, estimating how much money was needed to write checks for current retirees and adding that amount to the budget. This system worked well as far as paying retirees went, but it did not hold policymakers fiscally responsible for today's decisions affecting the size of the future retirement bill, e.g., increasing the force size. To promote better management, in 1984, Congress directed a switch to an accrual method of funding retirement. Under this procedure, each year the services transfer into a fund the amount necessary to pay for future retirements. The amount transferred is a percentage of the service's basic pay. Thus, if a service implements policies that affect the future value of retirement benefits, it sees the budgetary consequences of that decision immediately in the form of an increase in the amount transferred to the retirement fund. Analysis by Arroyo Center researchers William Hix and William Taylor, reported in A Policymaker's Guide to Accrual Funding of Military Retirement, suggests that the current procedures do not fully capture the intent of the legislation and that changes could eventually save the Army as much as $5–6 billion annually.[1]

How the Retirement Fund Works

When Congress established the retirement fund, it shifted responsibility for service rendered before October 1, 1984 to the Department of the Treasury; DoD has responsibility to fund service rendered after that date. At the time of the transfer, Treasury accepted an unfunded liability estimated at $529 billion, which was to amortize over 60 years.[2] Annually the services transfer an amount equal to a percentage of their basic pay accounts for the active and reserve components. The percentage differs by component, but it is identical within components for all services. In FY95, fund transfers equaled 33.5 percent of the active duty basic pay and 9.7 percent of the selected reserves. The Board of Actuaries annually calculates the liability for the pre-1984 service, adjusted for changes in assumptions and experience, and transfers an amount equal to one year's amortized payment. The money in the fund is invested in nonnegotiable government securities, and it draws interest.

Transfers into the fund and its investment transactions qualify as intragovernmental transfers (even though they represent an outlay to DoD) and thus have no effect on the deficit. Only payments to retirees from the fund represent outlays to the federal government. Figure 1 shows the operation of the fund and lists important factors in the calculation of the amounts.

Figure 1. Operation of DoD Retirement Fund

How Gains and Losses Occur

The fund's liability is not static, and three things can cause what are called actuarial gains and losses:[3]

  • Funding assumptions can change
  • Benefits can change
  • Experience can differ from assumptions.

Funding Assumptions

To determine how much money DoD has to transfer to the fund, a Board of Actuaries reviews assumptions about economic and noneconomic factors at the beginning of each year. Economic factors include assumptions about pay raises, cost of living allowance (COLA) increases, and interest rates. An assumed pay raise means that the future liability of the fund will increase because retirees will draw more money. Therefore, the amount transferred into the fund has to increase to account for this future liability. An assumption that the interest rate will increase has the opposite effect. The fund earns interest on nonnegotiable government securities. So if interest rates go up, the fund will earn more interest, thus the amount transferred can be less. The so-called noneconomic assumptions include such things as the rates of retirement and the longevity of retirees. If, for example, higher retirement rates or lower death rates are assumed, funding requirements increase.


Benefit changes also affect the size of the contribution. For example, Congress slipped the 1994 and 1995 COLA increases from January 1 to later in the year. These delays reduce the actuarial value of the retirement benefit and, hence, the funding required.


As mentioned, the actuaries review certain economic assumptions at the beginning of the year. Frequently, these differ from what actually happens during the year. For example, if the pay raises or COLAs approved differ from the assumptions, the fund earns more interest than anticipated, or fewer people retire than anticipated, the funding requirements change.

Right now only Treasury benefits from any decreases in fund liability.[4] The Treasury would make annual payments to fund this amount amortized over 50 years. If the liability of the fund goes down because it earns more interest or for some other reason, the size of the Treasury payment goes down. The assumptions made by the Board of Actuarieshave turned out to be conservative; over its life, the fund has never had an actuarial loss. All the net changes in liability have been downward.

These decreases can be substantial. In FY95, for example, the liability was reduced by $48 billion. The Treasury amortized this amount over 30 years, and reduced its annual payment by that amortized amount. Over the first 10 years of the fund's life, the average annual gain has been almost $30 billion, and the annual Treasury payment has shrunk from $25 billion to $11.5 billion.

Different Service Retirement Policies

A second area of interest pertains to service retirement practices. The intention of the annual transfer from the services is to fund the future retirement liability of the individuals represented in those accounts. This procedure assumes that all services retire people at an identical rate. But they do not. A service's level of seniority directly affects the number of people it retires. The higher the level of seniority, the more retirees. For both the officer and the enlisted forces, the Air Force maintains the most seniority, the Marine Corps the least.

Accrual percentages computed with service-specific personnel policies would differ significantly by service. Current policies cause the budgets of the Army, Navy, and Marine Corps to carry several hundred millions of dollars a year of the cost of Air Force personnel policies. Hence, in its retirement budget the Air Force appears several hundred million dollars a year cheaper than its actual cost; the other services appear more expensive.

What Should Happen

To capture the intent underlying the legislation, Arroyo Center researchers suggest two changes to current procedures, one requiring new legislation and one not. First, they recommend that the gains and losses that accrue to the retirement fund be shared between the Defense and Treasury departments. The division should reflect the relative contributions of the populations for which the departments have responsibility. This change would require new legislation. Second, they also recommend that each service contribute to the retirement fund an amount that reflects its retirement liability. This change would not require legislation.

There appears to be a legislative basis for sharing the gains. The clear intent of Congress when it established the fund was to promote better management. The law says that the monthly accrual payments are intended "to permit the military services to recognize the full cost of manpower decisions made in the current year." By making the consequences of decisions affecting retired pay immediately apparent in service budgets, Congress provided strong incentives for better management. But not being able to share in the actuarial gains tends to dissipate the effect of the incentive.

Congress intended for the retirement fund to allow the services to recognize the full cost of their personnel decisions each year. The committee report accompanying the legislation went on to say that "the individual services manage their forces in different ways and different tradeoffs would occur among the services." One of the different ways the services manage their forces is by seniority. If the legislative intent were to be followed, the Air Force would set aside the largest fraction of its base pay to fund retirements, and the Marine Corps the smallest. Yet each service sets aside an identical percentage. This policy in effect causes the Army, Navy, and Marine Corps to fund Air Force retirements. Furthermore, it means that the budgets do not reflect the cost of their personnel decisions. Were these two changes made, Arroyo Center researchers estimate that the Army share of these changes could approach $5–6 billion annually, or about 8 percent of the current budget.


  • [1] The research in this report builds upon a study by Richard Eisenman et al., The Accrual System for Funding Military Retirement: Assessment and Recommended Changes, Santa Monica, CA: RAND.
  • [2] Subsequently, the Board of Actuaries, an independent board that reports to the President on the actuarial status of the fund and advises DoD, reduced this period to 50 years.
  • [3] "Actuarial gains and losses" refer to changes in the predicted liability of the fund and not to gains and losses in the normal sense of debits and credits to cash accounts. Thus, when the fund sustains a gain, it means that the predicted liability of the fund has deceased.
  • [4] In theory, only Treasury would suffer from any increase in liability.

Research conducted by

This report is part of the RAND research brief series. RAND research briefs present policy-oriented summaries of individual published, peer-reviewed documents or of a body of published work.

This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited; linking directly to this product page is encouraged. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial purposes. For information on reprint and reuse permissions, please visit

RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.