The asset and liability portfolios of financial institutions generate patterns of future cash flows that must conform to many restrictions in order to assure solvency and profitability. Many institutions, including insurance companies and pension funds, have definite and certain future commitments of funds. These institutions may wish to invest funds now so that their cash inflows (investment with accumulated earnings) will match their future commitments. In principle, the simplest way to meet future commitments exactly is to purchase single payment notes (or zero coupon bonds) which mature on the commitment dates. For long-term commitments, such instruments are not readily obtainable, at least in the United States. Most available bonds promise coupon payments over time so that these payments would have to be reinvested at unknown future interest rates in order to realize an accumulated sum at any future date when a commitment must be discharged. Since future interest rates are unknown at the initial moment of investment, it is not certain what accumulated earnings will be at future dates. In the absence of default, the risk of not meeting future commitments may be minimized by adopting investment strategies based on the concept of duration. Duration is a measure of the average maturity of an income stream; it is a weighted average of the dates at which the income payments are received, where the weights add to unity and are related to the present value of the income stream. Dating from the initial work of Macaulay [9] and Hicks [6], duration has been shown to be important in constructing portfolios that are hedged or ‘immunized’ from the possible ravages of interest rate uncertainty.

In recent years, a number of studies have been published evaluating alternative bond portfolio strategies. These studies basically simulate risk-return characteristics for a variety of strategies designed for use by financial institutions. Typical strategies considered include portfolios of bonds that have laddered or barbell (dumbbell) maturity structures. In laddered strategies, bonds are spaced evenly among a number of consecutive maturities, while in barbell strategies, bonds are concentrated in short and long maturities. The results of these studies tend to differ and conflict. For example, in a recent article in this journal, Fogler, Groves, and Richardson (FGR) conclude that “dumbbell portfolio strategies are not as efficient as indicated by previous analyses.” Among the previous studies to which they refer is one by Watson, who concluded that “portfolios split between a spaced group of short maturity bonds and a longer investment security” (barbell portfolios) are most efficient. Similar results are reported by Wolf and by Bradley and Crane.

A number of recent papers have shown that it is possible for an investor to immunize a portfolio of default and option-free coupon bonds so that the return realized over a given planning period will never be less than that promised at the time the bonds were purchased. In this way, a future fixed dollar liability may be discharged with certainty by acquiring an asset portfolio with a market value equal to the present value of the liability and setting its appropriate duration equal to the time remaining to the date of discharge. However, most investors have more than one liability to discharge. In his seminal article in 1952, F. M. Redington showed that a stream of liabilities may be immunized if an asset portfolio having the same present value as the liabilities is selected so that:1. its duration is equal to the duration of the liabilities; and2. “the spread of the value of asset-proceeds about the mean term (duration) should be greater than the spread of the value of the liability” ([16], p. 191).

G. O. Bierwag, George G. Kaufman, Chulsoon Khang, Duration and Bond Portfolio Analysis: An Overview, The Journal of Financial and Quantitative Analysis, Vol. 13, No. 4, Proceedings of Thirteenth Annual Conference of the Western Finance Association, June 20-26, 1978 (Nov., 1978), pp. 671-681