The recent debacle in "government-guaranteed" bond mutual funds has raised the subject of bond portfolio risk to the highest of levels.
Regulators, credit rating agencies, mutual fund evaluators and pension plan consultants all are seeking to discover a better method of assessing bond portfolio risk. But there has been a distinct lack of an answer to that difficult question.
Investors by nature are risk averse and, at various points in the history of the bond market, one specific type of risk has taken center stage. In 1987 (from March through mid-October), investors learned the pain of a powerful market selloff and the dangers of too much duration. In 1990, credit risk became the focal point as the economy slipped into recession and corporate bond yield spreads widened and junk bonds plummeted.
In 1993, negative convexity (i.e., call/prepayment risk) became the hot topic related to bond risk. In 1994, product risk (e.g., mortgage derivatives) and liquidity risk dominated the headlines.
It seems appropriate at this juncture to address the issue of bond portfolio risk. Following is a brief review of the typical measures of bond portfolio risk, and a look at several methods to better assess and control the risks inherent in a bond portfolio.
Typical measures of risk
The commonly used measures of bond portfolio risk include average maturity, average duration, average effective duration, average quality and the historical standard deviation of a portfolio's returns.
Average maturity, which looks only at principal repayments, is simply the weighted average maturity of the individual bonds comprising the portfolio. A portfolio with a long average maturity is considered riskier than a portfolio with a short average maturity.
Average duration, which considers both coupon and principal cash flows, is the weighted average Macaulay's duration of the individual bonds in a portfolio. Or, another way of putting it: Average duration is a measure of the average amount of time one must wait to receive a portfolio's cash flow, where the cash flows are expressed in present value terms. A portfolio with a long average duration is deemed riskier than a portfolio with a short average duration. Average effective duration, which incorporates optionality (i.e., puts, calls, prepayments) into its calculation, is the weighted average of the effective durations of the individual bonds in the portfolio. A portfolio with a high effective duration is considered riskier than a portfolio with a low effective duration.
Average quality is the weighted average credit quality of the bonds comprising the portfolio (e.g., Treasury, AA, BBB). A portfolio with low average quality is viewed as riskier than a portfolio with high average quality.
Standard deviation is a measure of the variability in a portfolio's returns over time. A portfolio with a large standard deviation in return is considered riskier than a portfolio with a low standard deviation in return.
Flaws in the process
Average maturity fails to consider coupon payments, often the single largest contributor to a bond's value. Average maturity finds a 30-year zero-coupon bond of equivalent risk to a 30-year, 8% coupon Treasury bond. In fact, however, the zero-coupon bond has approximately 21/2 times the price volatility of the coupon-bearing Treasury.
Average duration is an improvement over average maturity in that average duration takes account of a bond's coupon payments as well as its principal repayments and evaluates all of a bond's cash flows on a present-value basis. A "present value" is the value today of a cash flow that will be received at some point in the future. For example, in an 8% yield environment, a $100 cash flow to be received in one year is worth approximately $92 today (present value) because that $92 can be invested at 8% and "grow" to $100 over the next year.
But average duration is faulty in that it is constantly changing - as interest rates fluctuate, as time passes, as coupons are paid, etc. Average duration assumes all yields change in a parallel fashion (i.e., all maturities along the yield curve experience exactly the same change in yield) - a truly rare occurrence. Average duration also relies on stable yield spreads between U.S. Treasuries and non-Treasury sectors such as corporate bonds and mortgage-backed securities. But sector spreads always change. Individual issues also are expected to experience identical yield changes - another far-fetched assumption.
Average effective duration is a better risk measure than average (Macaulay's) duration in that it considers a bond's implicit options in its calculation. This feature is particularly important in evaluating portfolios that include mortgage-backed securities and callable (or putable) corporate bonds. Unfortunately, average effective duration is inadequate because effective durations change as interest rates shift, as volatility rises or falls, as prepayment models are adjusted/updated, as time passes, and as coupon payments are made. Average effective duration is also dependent on parallel yield curve shifts and stable sector and issue spreads - heroic assumptions indeed.
Average quality is a snapshot of a portfolio's average credit quality at a point in time. It ignores the trends in quality in various market sectors (stable, improving or deteriorating). It can also lead to a false sense of security about the stability of a portfolio's market value. Recall that the recent debacle in bond funds has been in government-guaranteed funds with the highest possible average quality. Average quality fails to consider general market risks associated with bonds. These market risks can overwhelm the impact of credit quality changes.
The standard deviation of a portfolio's total returns has several faults. First, it mixes together all of the influences on a bond portfolio's returns and does not differentiate the impact of each factor. Second, it typically is based on a very short measurement interval (e.g., monthly or quarterly time periods) which can overstate a bond portfolio's longer-term riskiness. Third, the standard deviation often is based on returns over a recent period of time (three years, five years). The representativeness of such a period-specific standard deviation is questionable, particularly in a market that shifts from a bullish trend to a bearish trend or vice-versa.
Misperceptions about risk
There are several common misperceptions about bond portfolio risk.
First, that cash is a risk-free asset. Not true. Were it not for bailouts by several mutual fund organizations (i.e., BankAmerica, PaineWebber), some cash funds would have suffered declines in net asset value. Cash assets such as T-bills can show considerable variability in returns (recall the days of 18% T-bill rates?) and, consequently, they have risk.
For any pension plan with an investment horizon greater than a few months, the reinvestment risk of a T-bill is substantial. The risk stems from the fact that the T-bill (or any cash-type instrument) must be "rolled over" often and the entire principal is subject to the prevailing market interest rate at each maturity da