|Stocks vs. bonds: Is "buy and hold" the best strategy?
The article, "Stocks vs. Bonds--High Rewards Without High Risk" (GSB Chicago Fall 1997) seemed rather wishy-washy. To me, even the concluding statement "The simplest strategy--to buy and hold the market--may also be the best for most investors" conveyed ambivalence.
It is my impression that more and more investment activity has been influenced by the "buy and hold the market" idea. There are understandable reasons for this, including efficient market studies, the fact that the market value of stocks (even back around 1990) was over twice the value of corporate bonds, and historical returns from stocks have been much better than those from bonds.
As noted in the article, "the average real return on the U.S. stock market over the 19261993 period was 6.6 percent. Over the same period, long term government bonds earned a mere 1.7 percent. The difference between the two, 4.9 percent, is called the equity premium."
A similar reading on the historical equity premium, updated through 1997, indicated an annualized real return on the stock market of 7.7 percent; for long term government bonds it was 2.1 percent. The addition of just four years resulted in a 5.6 percent equity premium. It may be tempting to conclude that the case for stocks is even stronger. Before doing so, it would be wise to attempt a determination of probable underlying causes.
Some light may be shed on this matter by examining price behavior through an investment value theory (IVT) lens. In the case of common stocks that involves focusing on the expected rate of future growth in earnings and the rate at which that stream is discounted. A reasonable estimate of the S&P 500 real discount rate at the present time would be around 5 percent (1.5 percent dividend yield plus 3.5 percent expected future growth rate.)
When such momentary forward-looking annualized rates are considered, the current 5 percent rate for stocks and the corresponding real rate of 3.8 percent for L.T. bonds (U.S. government) reflects a difference of just 1.2 percent. Such an equity premium, by itself, should engender a more temperate view of the equity advantage. Using this IVT perspective, one finds highly plausible explanations for the increase in the equity premium from 4.9 percent at the end of 1993 to 5.6 percent at the end of 1997.
A reasonable estimate of the real equity discount rate at the end of 1993 would be 6 to 6.5 percent, i.e., 1 to 1.5 percent greater than it is at the present time. During the intervening period the dividend yield contributed mightily to an S&P 500 price of 970.45 at the end of 1997. Absent that yield decline, the price would have been around 600. That difference, spread over four years, was 13 percent per annum, 56.5 percent of the annualized total return of 23 percent for that period.
There is clear empirical evidence, at least since 1929, that dividend yields have had a strong inverse relationship with the real earnings growth rate (adjusted for business cycle impacts) over trailing periods of 12 or more years. Such trailing experience is presumed to have significant effect on future expectations, similar to the way considerably shorter trailing periods have on inflation expectations. When future growth rate expectations rise, dividend yields decline and vice versa.
To illustrate, from 1984 (an earnings peak year) through 1993, the real earnings growth rate was actually negative. However, informed investors understood that significant restructuring charges and FAS reporting requirements changes made the results look quite a bit worse than they really were. But with an almost 30 percent increase in real earnings between 1993 and 1997, the real growth rate for 1984 to 1997 was about 3.2 percent per year. At least in retrospect, the 1 percent decline in dividend yield did not seem surprising.
If this logical inverse relationship continues over longer time periods (say, five to fifteen years) in the future, a seemingly modest decline in future growth rate expectations could easily cause a 0.5 percent increase in the dividend yield. Other possible influences aside, by itself that much of a yield increase would drop prices about 24 percent. Such an erosion in future growth rate expectations would not necessarily require a recession.
Given this equilibrating function of the dividend yield, it is important to note the increased price volatility inherent to this function with dividend yields at such low levels in recent years. So far this increased volatility has been evident mostly in sharply rising prices. However, if such yields start moving upward, the associated impact on stock prices could result in the perception of greater risk, thereby inducing an upward tilt in the real discount rate. The latter, by itself, might "require" some further (albeit seemingly modest) increase in the dividend yield.
Such considerations could easily justify, at least in some portfolios, something other than unwavering adherence to a "buy and hold" strategy. Furthermore, they might stimulate different approaches to an improved understanding of financial market behavior.
William S. Gray, 50