February 2, 2010
The table below shows the distribution of every possible three-year return in the U.S. stock market – measured as of each month-end – between January 1884 and June 2009. This period encompasses 1,506 observations measured over rolling 36-month holding periods.
| U.S Stock Market* 1884 – 2009 Rolling 3-Year Holding Periods 1,506 Observations | |
|---|---|
| Median 3-Year Return (annualized) | 9.50% |
| Average 3-Year Return (annualized) | 10.63% |
| % of Periods with Negative Return | 15.10% |
| Best 3-Year Return (total) | 194.52% |
| Worst 3-Year Return (total) | (80.84%) |
*U.S. stock market returns for the period 1884 through 1926 are derived from the Shiller market index at www.econ.yale.edu/~shiller/data.htm. Returns from 1926 through 1969 represent the “Large Company Stocks” category from Ibbotson Associates. Returns from 1970 onward represent the S&P 500 Index.
Sources: Robert J. Shiller, Standard & Poor’s; Ibbotson Associates; Capital Advisors, Inc.
If the experience of the past 125 years is a relevant guide for the future behavior of U.S. stocks, an investor who is willing to hold the market index for at least three years would rationally have the following three-year expectations at any given month-end starting point:
- An annualized return, including dividends, of around 9.5% on average.
- Around a 15% probability of losing money over 3-years, or slightly greater than 1-in-7 odds.
- A possibility, although very remote, of nearly tripling their money.
- A possibility, although very remote, of losing around 80% of their money.
Nothing about this data should look surprising to an experienced investor. Indeed, something very close to this same distribution of possible market outcomes has been programmed into the financial planning software used by most professional investors for at least the past three decades
But what if this is the wrong distribution?
To address this question, I segmented the 125-year history of the market index into quartiles based on the level of the Shiller P/E Ratio as of each month-end dating back to 1884. I then measured the return distributions associated with the lowest and highest quartiles for the Shiller P/E Ratio to search for differences between the two. The difference was material!
The table below compares the distribution of three-year returns that follow a starting Shiller P/E Ratio of 11.54 or less (11.54 is the threshold for the lowest quartile of all monthly observations since 1884) with those that follow periods that begin with a Shiller P/E of 19.20 or more (the threshold for the highest quartile in the data):
Display article as PDF for printing.
Would you like to send this article to a friend?
Remember, if you have a question or comment, send it to .
