Daily Speculations

August 2003

The 1.5 million percent question

In their 2001 masterpiece, Triumph of the Optimists, London Business School professors Elroy Dimson, Paul Marsh and Mike Staunton document the returns in the major global stock markets during the 20th century. Since reading the book, we have been engaging in a spirited debate with the professors over whether the 21st century’s returns will be comparably great. We say yes; they say no. Their latest thinking appears in a forthcoming article on equity premiums for the Journal of Applied Corporate Finance. 

Dimson wrote us: “We cannot unthinkingly extrapolate expected total return from the past total return (something you imply in your email but do not, I suspect, believe). That is the substance of our piece in JACF. Implicitly, however, you are observing that some “low-future-expected-return: adherents derive their views from the historical record of real dividend growth (which is quite a low rate of growth). You are correct to imply that we should not unthinkingly extrapolate expected future dividend growth from past dividend growth.  That would be the same "sin" as Shiller et al criticize. So people who use a forward-looking interpretation of the Gordon model should use forward-looking growth forecasts.  But where would the latter come from?”

We took the liberty of showing the new Dimson/Marsh/Staunton article to Will Goetzmann, an eminent market historian at Yale. His comment: “A major problem with all studies of the equity premium is that  there in no riskless asset, and there is no theory about whether risky government bonds should have a higher or lower expected return than stocks.  In some countries, equity is clearly less risky because government default or hyperinflation is more likely than equity market collapse. So the spread between stocks and bonds is little more than a curiosity.  Jorion addressed this by comparing equity returns with inflation (i.e., a basket of goods) and also against the U.S. dollar, although this was an ex post benchmark.

"Missing from DMS:  St. Petersburg, Shanghai, Teheran and most of South America."