BrettArends

My, what short memories investors have.

As the Dow Jones Industrial Average races toward 17,000, on its way, no doubt, to 36,000, it seems that everyone and their broker has either forgotten, or never knew, about one of the simplest and most devastating pieces of stock-market analysis ever written.

It was written just 12 years ago, by Rob Arnott, then at First Quadrant and now at Research Affiliates, and the late, great investment legend Peter Bernstein.

It bore the innocent-sounding title “What Risk Premium Is ‘Normal’?” But behind the title lay some analysis that was far from innocent — and which absolutely all investors, including those just putting money into their 401(k) every month, need to know.

That analysis, in a nutshell: When money managers tell you that stocks “on average” produce returns of 8%, 10% or even 12% a year, they are engaging in a sleight of hand.

And if they are telling you that you can expect something similar in the future, the most charitable explanation is that they either flunked history at school, or math, or logic — or all three.

“We are in an industry that thrives on the expedient of forecasting the future by extrapolating the past,” Bernstein and Arnott wrote. “As a consequence, investors have grown accustomed to the idea that stocks ‘normally’ produce 8% real returns and a 5% risk premium over bonds, compounded annually over many decades.”

U.S. expats give up citizenship to flee taxes

On Lunch Break, WSJs Tanya Rivero explores how American expatriots are giving up their citizenship to escape taxes, how to handle workplace criticism, how children can eat healthy this summer and more. Photo: Getty

Such expectations, they wrote, were based on U.S. stock-return data tracked by a number of sources since the 1920s.

There is nothing wrong with those numbers as far as they go — but using them to extrapolate the future is, as Bernstein and Arnott pointed out, deeply, deeply flawed.

Wall Street’s optimistic arguments about stocks today are based on double-counting and circular reasoning.

First, noted Bernstein and Arnott, a lot of those big returns realized since 1926 were based simply upon the dramatic upward re-rating of stocks which took place during the 1980s and 1990s.

As an illustration: Back in the mid-1920s, according to data tracked by Yale professor Robert Shiller, investors typically paid about $13 for a basket of stocks generating $1 a year in net earnings (Shiller used average earnings for the previous 10 years).

Today, stocks generating a comparable $1 in earnings would cost you a lot more — about $26.

In other words, over that period of time, investors got a huge one-off gain of nearly 100% just because shares got more popular.