At least one good indicator is telling us to sell equities now – but we should ignore it.
This indicator is the10-month average rule, proposed by Mebane Faber at Cambria Investment Management. He showed that an investor who had sold the S&P 500 when it fell below its 10-month moving average and bought it when it rose above that average would have made better risk-adjusted returns than on a simple buy-and-hold strategy. The same has been true in recent years for emerging markets and the All-Share index, especially for more bubble-prone sectors such as miners or IT.
With the All-Share index now well below its 10-month average, does this mean we should sell?
Usually, I believe we should trust rules rather than judgment. This case, however, is an exception. We should ignore this rule now. To see why, let’s look at when the rule has worked best.
One such period was during the bursting of the tech bubble. The 10-month rule would have got us out of UK equities in November 2000 and kept us out until 2002, thus saving us a 13 per cent loss. The rule would also have got us out of the market from November 2007 until May 2009, thus saving us a 30 per cent loss.
But these two episodes have something else in common. On both occasions, the dividend yield on the All-Share index was also low, at under 3 per cent. It, too, was telling us to sell.
Today, of course, things are different. The dividend yield is well over 5 per cent. Insofar as history is any guide, this is a buy signal. What’s more, history suggests that the signal from the dividend yield is stronger than that from the 10-month rule.
We can test this with a simple regression equation. This has two independent variables: the dividend yield, and a dummy variable which is one if the index is above its 10-month average and zero if it is below it. If we regress subsequent annual changes in the All-Share index since 1990 upon these two variables, we get the equation: returns = (9.1 x yield) + (5.4 x dummy) – 30.1. This equation, which explains over a fifth of the variation in annual returns since 1990, tells us that both the 10-month rule and the yield predict returns.
But the yield is a stronger predictor. So much so that this equation is (as I write) now predicting a rise in the index of almost 20 per cent over the next 12 months.
Of course, history might not be a good guide. Perhaps today’s high yield isn’t the buy signal it would have been in the past, because it tells us not that the market is cheap, but that the economic outlook is poor and risky.
So let’s forget statistics and think about theory instead. Why does the 10-month rule work? It does so by exploiting investors’ under-reaction. When investors under-react to bad news, shares will be overpriced even if they’ve fallen below their 10-month average. In this situation, the 10-month rule will work well.
And this is what happened in 2000. Investors cleaved to their prior belief that tech stocks had great prospects even though the newsflow was deteriorating. Stocks were therefore overpriced even when they were below their 10-month average. Similarly, in 2007 investors under-reacted to the freezing up of interbank markets in August of that year, and stuck too strongly to their earlier optimism.
On both occasions, price subsequently fell as bad news gradually hit shares. And the 10-month rule protected us by getting us out in advance.
Is today really like these two occasions? Prices have not drifted down: we’ve suffered one of the worst three-month drops on record. And it’s hard to believe that investors are cleaving to 'business as usual' prior beliefs.
Indeed, it’s possible that instead of under-reacting, the market might have over-reacted. Risk-parity traders – those who try to keep constant volatility on their books – have sold as volatility has soared. And value investors who would ordinarily buy and stabilise the market have been sidelined by fears of being swamped by momentum sellers.
When markets overreact, the 10-month rule sends a misleading signal. It might be doing so now.
Yes, there’s a risk to this view. It’s possible that the economic damage done by the virus will last longer than we expect. This is not simply a matter of government guidance – if they lift the lockdowns but people self-isolate as the virus continues to spread, damage will also be done. If this is the case then we will see further stock market falls.
In such an event it will look as though the 10-month rule has worked. But this won’t be because of the merit of the rule – which is that it alerts us to the fact that markets sometimes underreact – but because a risk has materialised.
This risk is a strong justification for us not betting all-in on the market recovering. It’s not, however, justification for following the 10-month rule itself.
