A dull year

In December 2018 I wrote that lead indicators “are now pointing to the index rising just over 10 per cent by next December”. It did.

Of course, one successful forecast might be just luck. But the same lead indicators also correctly predicted that the All-Share index would fall in 2018 and rise slightly in 2017. That means they’ve scored a hat-trick – three successive correct out-of-sample predictions.

My chart shows the general performance of these lead indicators since 1998. There have been some failures. They didn’t see prices falling in 2002-03; underestimated returns in 2006-07; and didn’t anticipate the full extent of the slump in 2008-09, although they did point to a significant fall then.

There are four main lead indicators here:

- The dividend yield. A high yield predicts rising prices.

- Foreign buying of US equities, measured over a 12-month period. The logic here is simple. Investors tend to invest disproportionately in their domestic market. They only invest a lot overseas if they are especially confident. Big buying of US shares by non-Americans is therefore a sign of high optimism. But this means investors are over-exuberant and that shares are heading for a fall. Foreign buying peaked in 2000 and 2007, when stock markets were unsustainably high.

- The ratio of the All-Share index to its 10-month average. This captures the fact that markets are prone to momentum: rises lead to further rises, and falls to more falls.

- The ratio of the money stock in developed economies to the MSCI world index. When investors have lots of cash and few equities, they are likely to rebalance their portfolios away from cash and towards equities, causing prices to rise. And when they have lots of equities and little cash, they are likely to rebalance away from equities, causing prices to fall.

Collectively, these indicators work as a measure of investor sentiment. When sentiment is high, it’s a sign that lots of good news is in the price and bad news not. That means the market is vulnerable to downside surprises. Also, sentiment tends eventually to mean-revert: we rarely stay optimistic or pessimistic for very long. This means that high sentiment often leads to falling sentiment and hence to falling prices. By the same reasoning, low sentiment leads to rising prices.

Such measures, however, only work as lead indicators for longer horizons – 12 months or more. In the short term, pessimistic investors might become even more pessimistic and optimistic ones even more optimistic, so sentiment doesn’t predict returns over shorter horizons.

This raises the question: if returns are so predictable why don’t investors actually predict them and in doing so bid away the predictability? Take for example, December 2018. My lead indicators then predicted good returns. Investors who shared this optimism should have bought then and in doing so forced prices up. Had they done so, the market would not have done as well as it did in 2019 and my lead indicators would have failed.

So, why didn’t this happen? One reason is that some investors are judged on short-term returns. Even if you’re confident of the market rising over 12 months, you’ll not buy if your job and bonus depend on you being right in the next three.

Also, investors rely too much upon their judgment. Professional investors and advisors need to tell their clients stories and to project an air of mastery of complex data. Telling clients that the job of predicting returns can be done by any mediocrity with a simple equation would be professional suicide.

Relying on judgment is, however, dangerous because it leaves us vulnerable to the countless errors to which unaided reason is prone. One in particular is the projection bias. We tend to project our current mood into the future, and fail to anticipate that it will change. When investors are anxious or depressed, therefore (as they were in December 2018) they don’t expect to cheer up. So they don’t anticipate that prices will rise nicely.

The danger of this error, plus the others identified by Nobel laureate Daniel Kahneman and his followers, means that the psychologist Paul Meehl was right back in 1954: simple statistics can beat professional judgment.

So, what do these lead indicators predict for 2020?

Here’s an anti-climax. They point to the All-Share index rising by around 4 per cent, with roughly a one-in-three chance of it falling. While the above-average dividend yield is a bullish signal, the global money-price ratio is still low, which is worrying. Net, our indicators predict a small rise.

Which is just what a naive forecaster would predict.

Which is unfortunate. This means that next year will not be much of a test of my lead indicators: even if they prove correct again, they’ll only do as well as a know-nothing forecaster would. So even a successful forecast would do nothing to vindicate this approach.

But there is a consolation here. Even applying a decent margin of error to this forecast (as of course everybody should to any forecast) implies that 2020 will be a dull year. And there are many worse things than dullness.