Since 1965, the All-Share index has given us an average total return between May Day and Halloween of just 1.9 per cent. That's less than the average return on cash, and compares to an average return of 12.8 per cent between Halloween and May Day. There's little sign of this pattern disappearing; in only three of the last 10 years have the six summer months (May to October) seen better returns than the previous winters.
This tendency for the stock market to do relatively badly in the summer is no recent phenomenon. Nor is it confined to the UK. Ben Jacobsen and Cherry Zhang at Massey University in New Zealand have studied the "sell in May" rule for all national stock markets in all history - and they've found that it works almost everywhere most of the time. On average, across the 108 countries they looked at, returns have averaged 6.9 per cent from end-October to end-April, but only 2.4 per cent from end-April to end-October. In 23 of 24 developed stock markets, returns have been higher on average in winter months than in summer ones - the exception being Finland. And they are statistically significantly better in all but seven of these countries.
Why is this? An experiment by Yaron Lahav and Shireen Meer at Emory University in Atlanta gives us a clue. They got subjects to trade an artificial asset under laboratory conditions. Some subjects were shown a brief comedy film before trading and others weren't. They found that the traders who saw the film traded the asset at higher prices than those who didn't. This suggests that asset prices are higher when traders are in a better mood, even if that mood has nothing to do with the asset's fundamental value. Perhaps then the better mood caused by warmer weather and lighter evenings in the spring pushes share prices up to levels which often prove too high.
One otherwise curious fact is consistent with this - that two of the few markets where there is no significant May Day or Halloween effect are in Australia and New Zealand - where the weather worsens in April and May and gets better around Halloween*.
So, what other evidence is there that investors' sentiment might have pushed share prices up too far?
One piece is that mid-caps' prices are high and dividend yields low - which is worrying for the market generally and smaller stocks in particular.
There are (at least) two other measures of investors' moods which have the ability to (partially) predict returns. One is the dividend yield on the All-Share index; since January 1995, the correlation between this and subsequent six-month returns on the index has been 0.35. The other is foreign buying of US equities; since 1995, the correlation between the 12-month running total of such buying has been minus 0.46.
Right now, these point to low but positive returns on the All-Share over the next six months - of 4 per cent in the case of the dividend yield and 1.5 per cent for foreign buying of US shares**.
In one sense, these indicators disconfirm the "sell in May" rule, as they point to positive returns - indeed, slightly better than those available on cash.
In another sense, though, they do corroborate it. They suggest that there's a good chance that the market will fall over the next six months - of around two-fifths if post-1995 relationships continue to hold. And you might think such low returns are insufficient reward for taking on equity risk.
Of course, none of this rules out the possibility that the market will rise nicely. There's so much noise in equity returns that we could get lucky. The question for investors, then, is Harry Callahan's famous one: do I feel lucky?
*You might think this should mean that investors in these markets should buy in May and sell on Halloween. However, this rule is offset by the tendency for these markets (like most others) to be correlated with the US, and so to do well when the US rises, which tends to be the Halloween-May Day period.
**One complication here is that we only have data on foreign buying for February; this forecast assumes that March and April data don't change the current 12-month total buying.
