No concept in investing is more important than that of expected returns. Whether one is an individual planning for retirement income, a university endowment establishing a spending policy or an insurance company determining how to price its policies and annuities, assumptions made about returns likely to be generated from financial assets (stocks, bonds) over the long run will have a critical impact on the success or failure of the plan, policy or price.
All too frequently, planners draw their estimates of expected returns from historical returns. But are historical returns helpful? Suppose I’ve just retired and want to determine how much I can withdraw from my savings. Assume I’ve reached the full Social Security retirement age of 66, with a 20-year actuarial life expectancy. To be conservative, assume I might live at least another 30 years.
Since I want to project 30 years into the future, to be safe, I’ll look 60 years into the past. I find that the nominal (before adjusting for inflation) returns for U.S. stocks from 1951 to 2011 averaged 10.4 percent annually, while those for bonds averaged 6 percent. Because the value of the dollar has declined 88 percent over those 60 years, or 3.7 percent annually, I have to temper my return expectations with the assumption that we may see similar inflation over the coming 30 years.
Financial planners and economists calculate “real returns” by adjusting nominal returns for inflation. Subtracting the expected inflation from my 60-year historical averages gives expected real returns (annualized) of 6.7 percent for stocks and 2.4 percent for bonds. These sound like reasonable numbers, so I plug them into my planning spreadsheet and determine that, with a 50-50 mix of stocks and bonds, I should be able to generate expected real returns of 4.5 percent annually from my portfolio. If I limit withdrawals to 4.5 percent of my capital per year, my retirement assets should last indefinitely, even taking inflation into account.
But wait a minute! Not all 30-year periods in my table of investment returns look alike. I note that stock real returns from 1951 to 1981 were 5.7 percent, 1 percent lower than the 60-year average. But for bonds, the results from 1951 to 1981 are dramatically different—the real returns are actually negative (-1.2 percent annually). From the end of World War II until about 1980, U.S. bond investors consistently lost a lot of money, taking inflation into account.
If financial conditions in 2012 are more like 1951 than 1981, perhaps I should use the 30 years following 1951 as my historical basis for calculating expected returns. Doing so, the prudent withdrawal rate drops to 2.3 percent.
From here, I’ll focus on expected returns for bonds and will discuss stock returns in a later column.
Three factors determine returns from bond investments: the stated interest rate, the time to maturity and the price the investor pays. Given these factors, it’s possible to calculate precisely the returns over the life of a bond at the time it’s purchased. Looking at some currently available Treasuries, I found that a Treasury bond maturing on November 15, 2021, with a coupon, or stated, interest rate of 8 percent and selling at $152.10, will generate a yield-to-maturity of 2 percent over the coming 10 years, the same as 2 percent Treasury bonds with the same maturity date selling at par. (Bonds are priced at percentage of par, or face value, so $10,000 worth of the 8 percent bond will cost $15,210, while same face value of the 2 percent bond will cost $10,000).
The expected return for a portfolio of bonds works out to be the portfolio’s average yield-to-maturity of the individual bonds. For any level of credit risk and maturity date, all bonds available will bear similar yields-to-maturity.
What drives the current prevailing yield-to-maturity of bonds trading in the market? The fundamental flaw in using historical rates to generate expected returns for bonds is that changes in yields-to-maturity over time aren’t random statistical events but are primarily driven by policies of the Federal Reserve and the U.S. Treasury.
From the end of World War II until the Volcker era at the Fed, the Fed and the Treasury consciously worked to keep interest rates low to minimize the cost of paying off the World War II debt, and later to fund the Vietnam War while rapidly expanding federal social programs. These conditions didn’t change until Paul Volcker took over as Fed chairman with a mandate to wipe out the horrific inflation the occurred from the 1960s to the late 1970s.
Where are we today? The Fed has flip-flopped once again, driving interest rates down in an effort to help the Treasury finance the massive intervention in the financial markets disrupted by the Great Housing Bubble. So yes, 2012 is more like 1951 than 1981 in terms of government interference in bond markets.
If historical returns don’t help, where do we turn? If we could determine the average yield-to-maturity of large bond portfolios, we could do a much better job of estimating expected returns. Fortunately, that information is readily available. The Securities and Exchange Commission requires every bond mutual fund to publish what’s known as the “SEC yield,” which is determined by calculating the yield-to-maturity of the bonds held by the fund. A quick look at the Vanguard website shows that the SEC yield for its very large Total Bond Market Fund is 2.16 percent. The SEC yields for other relevant funds are 2.49 percent for Vanguard’s Long Term Treasury Fund and 3.34 percent for its Intermediate Corporate Bond Fund.
Remember, the SEC yield is stated in nominal yield. We must adjust for expected inflation to arrive at a usable estimate of expected real returns. Subtracting the average 3.7 percent inflation rate of the past 60 years would give an expected real return of -0.36 percent from intermediate term corporate bonds. This is much closer to the 1951 to 1981 returns than the 2.38 percent of the past 30 years. Be forewarned.
