Even though September is by far the worst month of the calendar for the U.S. stock market, you shouldn’t bet that this seasonal pattern will continue. That’s because I know of no plausible explanation for why September should be bad for equities. Without the existence of such an explanation, there is a high probability that the pattern is merely a fluke of the data.
Of course, just because I don’t know of such an explanation doesn’t mean none exists. But I have my doubts. Despite asking for many Augusts now for your most plausible theories, I have yet to see any that withstand statistical scrutiny.
In this column I want to focus on the two explanations that have been most often suggested for the “September effect.” The first is a supposed “seasonal behavioral bias” that, as Investopedia puts it, leads “investors to make portfolio changes to cash in at summer’s end.”
The problem with this theory is that, if true, you’d expect that stock markets in the southern hemisphere would experience a pattern that is staggered by six months relative to what we see in the U.S. But I could find no such pattern.
Consider South Africa, where summer lasts from December through March. When ranked according to average return, April is the second-best month of the calendar for the South African stock market (as judged by the FTSE South Africa Index since 1988, per FactSet). March is the third-best month.
Or take Australia, where the summer season is considered to last from December through February. According to my analysis of Australian stock market data from the Organization for Economic Cooperation and Development, reflecting data back to 1960, February and March are no worse than the other months of the calendar, on average.
To be sure, these other countries’ experiences don’t prove that there is no seasonal behavioral bias at the end of the summer. But to argue that one still exists, you have to argue that this bias exists only among U.S. investors. That is a different, and more difficult, argument to make. You should be skeptical.
Mutual fund tax year
The other widely proposed theory to explain the stock market’s poor September performance is that mutual funds’ tax year ends on Oct. 31. According to the theory, fund managers anticipate that deadline by starting to dump their losing positions in September.
This theory is easily tested because beginning in 1986, all mutual funds were required to have the same tax year. Prior to then, mutual funds were free to choose any tax-year end. If the Oct. 31 tax year were the cause of the stock market’s poor September performance, we would expect this seasonal bias to have become more pronounced after 1986.
But that is not the case. Prior to 1986, the Dow Jones Industrial Average’s
return in non-September months on average was 1.8 percentage points better than in September. The comparable spread after 1986 is 1.7 percentage points. So the shift to a universal tax-year-end on Oct. 31 had no apparent effect on the magnitude of the September effect.
Of course, the failure of these two hypotheses doesn’t mean a plausible explanation for the September effect won’t someday be found. But until then, my recommendation is not to bet that the pattern is real. Stock market lore is filled with correlations that are statistically significant but have no real-world significance.
Needless to say, the stock market may still lose ground next month. My point is that, if it does, it won’t be because it’s September.
Mark Hulbert is a regular contributor to MarketWatch. His Hulbert Ratings tracks investment newsletters that pay a flat fee to be audited. He can be reached at email@example.com