"What Rate of Return Can You Reasonably Expect . . . or What Can the Long Run Tell Us about the Short Run?"

This is the title of a masterful piece in the March/April Financial Analyst's Journal by Peter L. Bernstein (unfortunately, no relation). If you enjoy finance, Mr. Bernstein's analysis of the expected return for stocks and bonds is pure mind candy. Five stars. (Find a library that has it, copy it, and read it 3 or 4 times.)

Yes, PLB agrees, the real (inflation adjusted) long term return of common stocks over the past 71 years has been about 7%. However, much of that return has been the result of increasing valuations -- look down the PE column of the stock table and note the dearth of single digit numbers. Mr. Bernstein observes that future stock returns may not benefit from this bonanza. He dissects out the effect of multiple expansion by analyzing 63 episodes averaging 35 years with the same starting and ending PE for the period 1871-1976.

The average real return turns out to be 5.7% for these 63 periods. Further, the standard deviation of real returns for these 63 periods was only 1.1%. In plain English, over the long term real stock returns are highly predictable, and not quite as high as we would expect from looking at the raw 1926-96 data.

Nowadays, every stock broker, tax attorney, CPA, and their dogs too, will tell you that the nominal long term return of common stock is 10%. However, if Mr. Bernstein is correct, and if I can perform simple arithmetic, his 5.7% real return added to the current inflation rate of 2.5% gives only a nominal 8.2%.

These numbers fall out exactly the same way from the dividend discount model. In 1926 the Dow yielded 4.5%, and the US economy has been growing at a real rate of 1% since. Add the two together, and you get . . . 5.5% real return, or 8.5% nominal return. (To complete the picture, over the past 71 years earnings multiples have doubled, adding another 1% to the long term real stock return.)

The next piece of the puzzle is bond returns. Here the waters muddy. Mr. Bernstein applies the same sort of analysis to bonds, this time comparing 63 periods from 1803 to 1978 with the same starting and ending yields. He comes up with a real return of 2.7% and an SD of 2.1%. In other words, in the long run, bond returns are actually a lot more unpredictable than stock returns.

Perhaps. However, I can tell you one thing with absolute certainty about real bond returns over the next 10 years; if you buy an inflation adjusted US Treasury obligation your real return is guaranteed to be within 20-30 basis points of 3.6%. You don't have to take my word for this. It's backed up by the full faith and credit of the US government.

So let's take a trip with Professor Peabody in the WayBack machine to, say, 1954. Stocks yield 4.7% and have a PE of 11.8. Government bonds yield 2.7%. At that time one could have reasonably predicted a long term real stock return of 5.7% (either with PLB's estimate, or adding the 4.7% yield to the 1% real growth rate), and a long term real bond return of zero (the coupon being the same as inflation). Clearly, stocks were a much better deal than bonds in 1954.

Fast forward to the present. The real return on the 10 year inflation adjusted treasury will be 3.6%. The real return on stocks will be about 3% (2% dividend yield plus 1% real long term growth) if you believe the dividend discount model, or 5.7% if you take Mr. Bernstein's research at face value. Either way, a dispassionate analysis suggests that stocks will not beat bonds with the same (or perhaps any) margin over the next decade or two.

Let's assume that Mr. Bernstein is correct, and that the real long term return of stocks is 5.5%. Assuming that the correlation of inflation adjusted treasuries with whatever domestic or global you are using is zero, we use the methodology described in the September 1996 issue to estimate return, and the Markowitz algorithm to estimate risk (SD). Below is the risk/return plot for various mixes of IATs and stocks. In each of the bolow one tick indicates a 10% change in composition:




Now, let's assume that the dividend discount model is approximately correct, and that the long term future returns of equity is the same as IATs -- 3.5%:




Finally, assume regression to the mean , a la the 1970s, with real returns of only 1.5%:

Ben Graham advised investors to hold an equal mixture of stocks and bonds. Much as Mr. Graham deserves our admiation and gratitude for putting the art of investing on a firm quantitative footing, this was lousy advice in 1934 (Security Analysis) and in 1971 (The Intelligent Investor). It may just be that "the dean" was simply ahead of his time.

copyright (c) 1997, William J. Bernstein