In his March 17 Other Views commentary, "The way out of the pension funding crisis," David A. Hershey seriously mischaracterized my position on the appropriate way to allocate the assets of a defined benefit pension fund. He identifies analyses that consider only the risk and return of a plan's assets as employing "the Sharpe model" and advocates instead analyses that consider the risk and return of a plan's surplus, which he terms "the Leibowitz model."
As a matter of fact, I have long advocated surplus optimization (also known as asset-liability optimization) for defined benefit plans. A representative discussion can be found in my 1990 paper, "Asset Allocation," in "Managing Investment Portfolios, A Dynamic Process," John L. Maginn and Donald L. Tuttle, editors (Warren, Gorham & Lamont, 1990). I have also participated in several studies of the asset allocation of large pension funds taking liabilities into account.
On this, as on almost all subjects, I am in complete agreement with Marty Leibowitz.
In the future, I hope that Mr. Hershey will try to avoid associating positions with other authors without reading relevant portions of their work.
William F. Sharpe
STANCO 25 Professor of Finance,
Your Feb. 17 editorial ("Real assumptions") was right on the mark. It's imperative that people adopt more realistic expectations of long-run investment returns, and that these expectations be based on sound economic fundamentals. To start, they should consider the underlying factors that determine returns to less risky assets, like short-term government securities. From there, estimating other asset returns boils down to gauging risk premiums, or how much compensation people require to take on investments with riskier outcomes.
Estimating returns on short-term government securities requires making explicit assumptions about inflation and real interest rates. For our part, we expect the major central banks to remain committed to price stability. This will likely result in measured rates of inflation in the 2% to 21/2% range over the long haul. And we assume that real short-term interest rates will average about 2% for some time. Near their 20th century average, this is consistent with a reasonable rate of long-term technological progress that determines productivity growth and the marginal product of capital. Putting it all together, nominal cash rates will likely gravitate toward 4% to 41/2% in the long term. Although well above current, cyclically depressed levels, 4% to 41/2% is far below the averages of recent decades and more comparable with rates prevailing in the 1950s and 1960s, when inflation was similarly well contained. Government bonds will likely command a small risk premium of about 50 basis points over cash yields, commensurate with the somewhat greater uncertainty of forecasting inflation over longer horizons.
Pinning down the equity risk premium is trickier. Intuitively, the size of the premium investors will require to buy equities will be related to the non-diversifiable risk of holding equities, and to investors' aversion to assuming that risk. In practice, the actual realized equity premium over cash has been very large, especially in the 1980s and 1990s - too large to be reconciled with reasonable degrees of risk aversion given the actual riskiness of equities. This "equity premium puzzle" suggests that investors did not really expect such enormous equity returns and are unlikely to require such outsized returns going forward. In fact, part of the reason realized returns were so big is that the premium required to induce investors into equities was falling (owing to increased information about equities, lower transactions costs, etc.), driving up equity prices in the process. Going forward, we believe that investors will likely require an equity premium over cash of only 2% to 3% in the long haul, which is much more consistent with reasonable degrees of risk aversion and the actual risk characteristics of equities. Adding this 2% to 3% premium to our assumed cash yields gives an expected equity return of 6% to 8% - well below the surprisingly high outcomes of recent decades.
Comparing our estimate of what equities will need to return (to get investors to hold them) with what equities can return (based on the free cash flows they'll be able to generate) provides a check on the appropriateness of current equity valuations. If investors expect that the returns equities can generate are, say, insufficient to meet their required rates of return, they'll bid down equity prices until expected returns from these lower prices match their requirements. The returns equities will be able to generate will be the sum of today's free-cash flow yield (roughly 2% to 3%) plus the growth of those free cash flows (likely 4% to 5% per year, consistent with expected productivity growth plus inflation). All told, equities will likely be able to return 6% to 8% per year, which is consistent with our expectation of the returns investors will require. This suggests that current equity valuations are reasonable and have purged much of the excessive valuation of the late 1990s.
In sum, our analysis suggests that financial asset returns are likely to be much smaller going forward (at least on average) than people grew accustomed to in recent decades. But the true value of this analysis is not the specific forecasts per se, but the methodology, which is robust, compared to alternative assumptions about how events may unfold. A framework based on economic fundamentals, not the latest ups and downs in the markets, is what investors need to condition their expectations of asset returns over the long haul.
Joshua N. Feinman
Deutsche Asset Management
EDITOR'S NOTE: Mr. Feinman is a member of the firm's global asset allocation team.
Pensions & Investments welcomes Letters to the Editor and submissions of commentaries for the Other Views page, and we reserve the right to edit them for style and/or clarity. Letters and other submissions may be sent by mail to the attention of Barry B. Burr, Pensions & Investments, 360 N. Michigan Ave., Chicago, IL 60601; by fax, to (312) 649-5228; or by e-mail to [email protected]