Created 10/17/1995
Go to
Brad De Long's Home Page

What Will the Stock Market Do?

Brad De Long
U.C. Berkeley

Economists are always being asked this question--even though many of our theories have been constructed to demonstrate that we do not (and why we do not) have very good answers. Stock prices are (or are very nearly) a random walk: they are as likely to go up as down tomorrow, and whether they will go up or down tomorrow has nothing to do (or very little to do) with what has happened about the past and everything (or almost everything) to do with what we will learn in the future.

To say "I can't tell you" is, however, unsatisfying to questioners. So here are two flip answers and one serious answer to the question "What will the stock market do?"

Quick Answers

One answer: I could accurately forecast where the stock market is going next week, next month, or next year, I would not be here: I would be calling in from my yacht.

Another answer is to note that those who have yachts do not have special and accurate insight into the stock market. The best answer was delivered by the turn-of-the-century financier J.P. Morgan. Asked "What will the stock market do?" he replied: "It will fluctuate."

A Reasoned Answer

A serious answer--in my view, the best serious answer--is provided by a line of argument developed by Yale professor Robert J. Shiller (at

Statistical Patterns:

The figure below captures the argument in a nutshell:

At the start of this year, the ratio of the S&P composite to its ten-year lagged moving average of earnings was 24.8--higher than in all except for one year between 1880 and 1986.


If the statistical patterns that held in the past continue to hold over the next ten years, then our best guess is that average inflation-adjusted returns on a portfolio that tracks the S&P composite index will be one percent per year: with inflation running at 2 percent per year and with dividend yields at about three percent per year, the best guess is that the (nominal) value of the S&P composite in a decade will be about where it was at the start of the year.

Important note: It could be a lot higher , and it could be a lot lower in ten years: it is not unusual for stock index values in ten years to be fifty percent higher or fifty percent lower than predicted by this statistical model). At most one-fifth of the variation in capital gains over the next ten years can be forecast from this statistical model.

The bottom line? Expect no nominal capital gains. In fact, expect inflation to erode the real value of a portfolio made up of the stocks in the S&P composite index over the next ten years.

Thus Treasury bonds look a lot better than stocks over the next ten years. Overseas equities (either in developed or in developing countries) look a lot, lot better than stocks over the next ten years--if the statistical patterns established in the U.S. over the past hundred years continue to hold.

Possible Caveats:

Is there any reason to think that past statistical patterns will not hold? Two important factors could disturb them:

To date there is no evidence that either of these possible disturbing factors is at work.


As of today, investments in the U.S. stock market today look like a bad bet.


The real (inflation-adjusted) value of the S&P composite index since 1870, five times the earnings paid by the stocks in the index, and ten times the dividends paid on the index. All series are plotted on log scale--so that equal vertical distances reflect equal proportional changes. January 1996=100. All data courtesy of Robert Shiller (Market Volatility, or


Created 10/17/1995
Go to
Brad De Long's Home Page

Associate Professor of Economics Brad De Long, 601 Evans
University of California at Berkeley; Berkeley, CA 94720-3880
(510) 643-4027 phone (510) 642-6615 fax