Mark Twain was a savant about the stock market, as he was about so much else.
"October. This is one of the peculiarly dangerous months to speculate in stocks in," he wrote in a mock diary entry in the novel "Pudd'nhead Wilson." He added, "The others are July, January, September, April, November, May, March, June, December, August and February."
Twain published this bit of satire in 1894, two years before the birth of the Dow Jones industrial average. He wasn't much of an investor himself, but he nevertheless captured the behavior of the stock market from his time to the present.
Stock speculation has been dangerous in all seasons, of course, much as Twain suggested. But, oddly, he was right about something else: a new statistical study shows that October, by some measures, is the most "peculiarly dangerous" of all months for the stock market. The two months that bracket it are almost as risky, the study also says, and autumn Mondays are particularly perilous days of the week.
The research was done by Salil Mehta, a statistician and econometrician who has firsthand experience with risk measurement, on Wall Street and in Washington, where he was director of analytics in the Treasury Department for the $700 billion TARP program, and director of policy, research and analysis for the Pension Benefit Guaranty Corp. He is now an independent consultant and adjunct professor of statistics at Georgetown University, and writes a very wonky and provocative blog called "Statistical Ideas."
In a recent post, "Our Autumn of Discontent," Mehta said that there were some solid statistical anchors for the old notion that fall is a dangerous time for investors.
The cascading market declines precipitated by the collapse of Lehman Bros. in the awful autumn of 2008 seemed to provide dramatic examples of the season's hazards. What's more, history shows that, on a percentage basis, five of the 10 worst days since the Dow's inception have occurred in October, along with one in November, and one in the first half of December. And six of the 10 worst days were Mondays.
Still, this could all be a fluke. Because any 10-worst list is an inherently small sample, and because that tiny list is drawn from a long period, any apparent correlations drawn from it alone have no statistical significance.
There have also been many academic studies on this subject, illustrating apparent seasonal patterns in stock markets around the world but not fully explaining why they might exist. "Sell in May and go away," by now an adage on Wall Street, was common wisdom in London early in the last century. One study of 300 years of monthly English stock data found persistent underperformance in October. But no one really knows why such patterns have occurred.
Mehta doesn't, either, but he has taken a new approach, using as his database the daily returns of the Dow from its inception in May 1896 through the middle of last month. Record-keeping for the Dow has been imperfect, particularly in its early days, so Mehta eliminated unreliable data. He ended up with a sample size of 29,400 days, large enough to allow him to apply some sophisticated techniques.
He sought to determine whether there actually were seasonal patterns for really bad days, which he defined as the 1 percent of days with the biggest stock declines. The cutoff for inclusion in that dubious group turned out to be a daily drop of 3.2 percent, he said. On average, there have been five declines that big every two years, although we haven't had one since November 2011.
The seasonal and weekly patterns that emerged from his sifting are statistically significant, he says. "The autumn does appear to be the riskiest time of year," with October being the riskiest month.
Of the 294 days that make up the worst 1 percent of market declines, 45 occurred in October, 33 in September and 43 in November. Thirty such days occurred in December. That compares with an average of 17.9 for the other eight months. Mehta also found that Mondays were riskier for the market than other weekdays. Some 85 of these big declines occurred on Mondays, compared with an average of 52 for the other four days of the market week.
There is less than a 1 percent probability that any of these patterns occurred by chance alone, Mehta said. "I've established that the deviation is statistically significant," he said, though he added that "I haven't found the causation."
One possibility is very simple, he said. Historically, the stock market has risen more than it has fallen. And because people tend to think in terms of calendar years, it is often the case that they realize their portfolios have risen by the time the autumn comes around, so by then they are ready to take profits, driving down stock prices. Once prices start dropping, major declines become more likely.
And why have Mondays been a troubled day of the week? It could be because investors have time over the weekend to fret and to decide to unload some stocks.
It is possible that there are other reasons for the trends, or no reasons at all. "I wouldn't go overboard on the implications of this," he said in an interview. "I wouldn't recommend selling stocks every autumn because of this, forever," he said.
This autumn, the market has its own particular vulnerabilities. These include stretched valuations because of the market's sharp climb in recent years, he said, as well as uncertainty about the Federal Reserve's intentions, the conflict in Syria and the precarious global economy.
Institutions, traders and Pension funds might want to take this into account and adjust their asset allocations, he said, while ordinary investors are probably best off with a simple, low-cost, diversified buy-and-hold strategy. Some people won't want to veer from a preset plan at all, he said, and that's a valid approach if they are prepared to ride out market declines. "If you're patient, in the long run it should be fine," he said.
If you are willing to tamper a little, he said, you might consider moving a small amount of money into cash when the market is high so you're ready to be opportunistic after an eventual decline. For most people's portfolios, he would limit this to about 5 percent of the total - perhaps 10 percent for those prepared to take greater risk.
"It's not clear, even with the other factors weighing on the market, where things will go from here," he said, "and you could lose out if the market rises in the next few months."
As Twain said, the market is risky in October. But the problem is, it's risky in the other 11 months as well.
No comments:
Post a Comment