Aftershock

From our Bureau of Panic Peddling

Aftershock

So what happens when the extravagantly back-tested research of two eminent economists strongly suggests that equity returns are in for a long hard road going forward? Should we stay the course, panic and sell, or stock up on canned food and bottled water?

The short answer is "stay the course." Even Fama and French, whose research I am referring to, don't advocate holing up in an elaborate Montana rabbit warren filled with guns and Spam. The research quite simply is a call for reflection, recognition perhaps, that 20-30% annualized returns are unlikely to last forever.

The paper by Eugene Fama and Kenneth French is to some degree revolutionary, because it questions some very basic and long-held premises about the risk premium that stocks enjoy over risk-free investments, like government bonds (Fama and French use 6-month commercial paper as a proxy for the risk-free rate).

Let's cut to the chase. Here is the most delicious quote of the study: "If we use the average growth rate of real dividends for 1950-1999, 1.61 percent per year, to estimate the expected future growth rate, the expected real stock return is 2.93 percent. The riskfree real interst rate for 1999 is 2.24 percent, so the estimate of the expected equity premium is 0.69 percent. If we replace the 1950-1999 dividend growth rate with the higher average growth rate for 1872-1999, 2.15 percent per year, the expected real stock return rises to 3.56 percent, and the expected equity premium is 1.32 percent."

Background

There are two widely used methods to calculate future equity returns. One uses dividend yields and dividend growth rates to calculate future returns, while the other essentially uses past returns and dividend yield to calculate future returns. The first of these methods is known as the Gordon Model, and is favored by Eugene F. Fama and Kenneth R. French in their recent study examining the equity risk premium.

The formula used in the Gordon Model is as follows for those of you who are interested[1]:

Pt = Dt 1/(r-g) = Dt(1 g)/(r-g)
P is price, D is dividend, r is the discount rate or cost of capital, g is the dividend growth rate, and t symbolizes a given time (t 1 is a year going forward). The formula assumes a constant rate of dividend growth, which is the same as the present growth rate.

The theory behind the Gordon Model is similar to that behind federal monetary policy. It's one of supply and demand. Companies raise dividends to attract capital. Generally at these points in time, stock prices are low and therefore anticipated returns are higher. When dividend payments go down, demand for stock is high, equity is highly valued, and therefore anticipated premiums going forward are lower.

And how does the Gordon Model pan out? According to Fama and French, over time (they use the time period 1872-1999) the Gordon Model is roughly 2.5 times as precise as the annual average realized real return method. Data shows that the standard error of the mean was significantly lower using the Gordon Method. The paper points out that even from 1950-1999, when the Gordon Method underpredicted returns by about 50%, the error of the predicted Gordon real returns was 0.87%, compared to 2.38% for the average realized real returns.[2]

The expected Gordon equity premium from 1872-1999 is 3.64%. The estimate from real returns is 5.73%. Fama and French hold that much of this difference results from a huge gap of a Gordon estimate of 3.40% vs. a realized equity premium of 8.28% from 1950-1999. It wasn't always so. From 1872-1949, the Gordon model predicted returns of 3.79%, and average realized returns came out to 4.10%. From 1959-1999 that small gap became a chasm. Fama and French use many of the pages of their study trying to explain this gap and to understand its implications for future returns.

Simply put, the reason for the gap boils down to a larger spread between the average rate of capital gain and the average dividend growth rate. Fama and French ascribe the difference largely to unexpected capital gains which occurred as a result of a declining discount rate. The fact that dividend rates are so low (1.32% in 1999 with a negative dividend growth rate), while capital gains have been so high for so long does not bode well for equity prices going forward - if one subscribes to the Gordon Method.

Where does that leave us?

Bear in mind that at the end of 1996, the Gordon Model predicted expected annual return going forward to come in at 4.8%. The Wilshire 5000 enjoyed annual returns of roughly 29%, 22%, and 22% from 1997-1999. In 1929, the Gordon Method predicted returns going forward would be 16.5%. The actual annualized return going forward though 1937 was -3%[3].

For an additional dose of perspective, it's worth noting that the dividend rate in 1937 was 7.6%. In 1999, as noted above, it stood at 1.32%, down even further from 1.9% at the end of 1996. This reveals why the Gordon Model predicts such a grim outlook. Fama and French note that in order for the Gordon predicted equity premium to return to the 3.40 average value for 1950-1999, given a 2.24 real interest rate, the growth rate for dividends would have to rise to 4.32 percent a year for the indefinite future.

The Gordon Model predicted risk premiums of 1.27 percent and 1.71 percent for 1980-1989 and 1990-1999. Back-tested from 1872, the data predicts returns that were roughly half of actual returns. So, for me at least, a read of the Fama/French paper does not necessarily say we will see a return to those absurdly low equity risk premiums, but that we could see those levels.

In a nutshell, past returns do not guarantee future performance. Indeed, it's important not only to remember that "risk" is an important part of the risk premium, but that there are no guarantees of any premium over any time period, regardless of how diversified you are.

Footnotes:

[1] P.V. Viswannath, Valuing Stocks, Pace University, Lubin School of Business, New York, NY, Febuary 1999.

[2] Eugene F. Fama and Kenneth R. French, The Equity Premium, University of Chicago Graduate School of Business, Chicago, IL, July 2000.

[3] Edward Renshaw, Valuing the Earnings and Dividends Associated with the S&P 500, State University of New York at Albany, Albany, NY, March 1997.