Sell in May

It's a iron law that no British fund manager ever has anything useful to say. Imagine my horror, then, when I read Nicola Horlick in the Speccie (rr) saying:

The old stock market adage of ‘sell in May and go away’ is likely to hold true again this year. It seems unlikely that we will see any significant action in the stock market until the autumn.

I say "horror" because this corroborates my own views. Seasonal investing is the only active investment policy I pursue; I switch my pension fund into cash sometime in May and back into equities in early November.
This is because history shows that the "sell in May" rule is amazingly useful. These guys have estimated that it has worked in the UK for over 300 years.
Since 1965 the All-share has given an average total return of just 1.9 per cent in the six months from May 1 to October 31. That's less than cash. But returns in the six months to April 30 have averaged 14.1 per cent. There's almost no chance that such a big difference could be due to luck, and no sign of this seasonal effect fading away in recent years.
Why is the market so seasonal? There are (at least) two possible explanations.
One, suggested by Mark Kamstra (pdf) and corroborated by Maurice Levi, is that investors are prone to seasonal affective disorder. As the nights draw in in September, they become anxious and depressed, and so sell shares. The upshot is that returns are high from late autumn onwards, because the risk premium is high. Similarly, by May, investors have become optimistic because of the lighter days, and have transferred this optimism to equities, with the result that shares become unustainably high then. 
Another theory is that the economy is more volatile in the winter. Yi Wen (pdf) has estimated that half of what we think of as economic cycles are in fact due to Christmas. Investors should therefore need higher returns in winter than in summer, to compensate for this big seasonal economic risk.
Evidence for both theories lies in the fact that Australia has the opposite seasonal pattern to most northern hemisphere markets.
There is, though, a quirk here. In my day job, I've estimated that the seasonal effect is strong even if we control for the predictive power of dividend yields. This means that seasonal changes in investors' appetite for risk are not fully captured by an obvious measure of the risk premium. I find this puzzling.
Despite this loose end, my behaviour shows that I believe in the "sell in May" rule.
Does this mean I agree with la Horlick? Not at all. She's not telling us any of what I've just said. She just gives some rationalist futurological reasons why the market might fall. But the lesson of the success of the "sell in May" rule is that such rationalism is unnecessary, and perhaps even worse than useless. As Deirdre McCloskey has said, there's a huge difference between rationality and rationalism.

Comments

[Ian]

Surely the "Christmas Theory" doesn't apply to Australia ?