Terrible as September usually is for the stock market, you shouldn’t be reducing your equity exposure for the next 30 days just because of the calendar.
My argument may strike you as odd, since the statistical case for September being a bad month is quite strong..
For example, since the Dow Jones Industrial Average was created in 1896, September far and away is the worst month of the calendar, on average. It has experienced an average loss of 1.06%, in contrast to an average gain of 0.75% across the 11 other months.
Furthermore, September’s dismal performance has been remarkably consistent: It was a below-average performer in all but one of the dozen decades since the late 1800s. In more than half those decades, in fact, it was in 11th or 12th place in a ranking of monthly average performance.
Why, given these impressive statistics, do I think it’s too risky to bet that this coming month will be bad just because it’s September? Because, as we all remember from Statistics 101, correlation is not causation. Without a plausible explanation for why September should be so terrible for equities, it’s too risky to bet that the historical pattern will persist.
Each statistician has his favorite illustration of this crucial distinction between correlation and causation. Mine comes from David Leinweber, founding director of the Center for Innovative Financial Technology at the University of California, Berkeley. He once found that the indicator most highly correlated to the S&P 500 was butter production in Bangladesh.
I’m sure we can all agree that it would foolish to base our trading strategies on Bangladeshi butter output. But without a good reason for why September has been bad for stocks, Leinweber says the same conclusion applies to it too.
In past years when I have made this argument, many of you have taken it as a challenge to find a plausible story for September’s terrible record. None that you have suggested so far is able to withstand scrutiny. The three most commonly-advanced hypotheses that I am aware of:
1. “Investors are more likely to sell stocks as they return from their summer vacations.” This hypothesis is hard to square with an academic study that found that investors are more likely to sell their stock holdings before their vacations rather than after.
2. “Many mutual funds have fiscal years that end in the fall, prompting them to sell their losers for window-dressing purposes and/or to reduce their tax burdens.” But more than three times as many mutual funds have fiscal years ending in December than in September, according to data from Lipper, and yet December is one of the best months of the calendar.
3. “Investors are forced to sell large amounts of their stock holdings in order to pay their children’s tuition bills at private secondary schools and colleges.” But September’s average loss in recent decades has been less than it was in earlier decades, even as tuition has skyrocketed.
Be my guest if you still want to keep searching for an explanation for September’s terrible record. I’m not holding my breath. But even if you were to find one, Lawrence Tint says that it won’t do you any good. Tint is the former U.S. CEO of Barclays Global Investors and currently chairman of Quantal International, a risk-management firm.
Tint argues that “if a convincing explanation for the September effect were ever found, savvy investors would immediately begin jumping the gun by selling in August, others in turn would try to beat them, and the historical pattern would quickly disappear. Unless you or I are able to discover something nobody else knows about, by the time we know why a pattern exists it’s too late to profit from it.”
What all this means: Your trading strategy for the next four weeks should not be based on this being September. If the stock market falls over the next month, it will be for other reasons.
Written by Mark Hulbert of MarketWatch