The wrong sort of value

We all know that the dividend yield has been a good predictor of returns on the All-Share index. Since 2000, the correlation between the yield and subsequent three-year changes in the index has been 0.54. Which means that almost one-third of the variation in returns this century can be explained by the dividend yield alone.

This fact alone is worrying. At 2.8 per cent now, the yield is well below its post-2000 average of 3.4 per cent. History thus warns us that returns will be below average. In fact, if past relationships continue to hold, there’s a greater than 50:50 chance of the market falling over the next three years.

Of course, this low yield is due mainly to last year’s cuts in dividends on many stocks rather than to any surge in prices: the All-Share index is still lower than it was in late 2019. But this is little comfort. It tells us that the market is already pricing in a big recovery in dividends: even a 20 per cent rise would only take the yield to its long-term average. This means rising dividends won’t push up prices, which poses the question: what will?

What’s true of the aggregate market, however, isn’t true of all parts of it. For some sectors dividend yields have been great predictors of returns while for others they have had no predictive power at all.

My table shows this for various FTSE sectors. At the top is the aerospace and defence sector, where the correlation between yields and subsequent three-year price changes has been a whopping 0.85, implying that the yield alone can explain over 70 per cent of the variation in returns. Other sectors where yields strongly predict returns include food producers and pharmaceuticals. At the other end of the table are banks and oil stocks, where yields tell us nothing about subsequent returns.

In fact, the latter should not be surprising. In an efficient market in which all information was in the price high yields would predict not rising prices but rather low growth in dividends while low yields would predict high dividend growth rather than falling prices. In such an efficient market there would only be a link between yields and subsequent price changes to the extent that a high yield compensated investors for a risk that did not in fact materialise. In a classic paper in 2002 the Nobel laureate Robert Shiller showed that this was in fact true for most US shares and that the ability of the yield to predict returns was true only of the aggregate market and not for individual stocks. The US market, he said, is “micro efficient but macro inefficient”. (This is an example of Simpson’s paradox.)

This isn’t quite true in the UK: those high correlations show that there have been some micro inefficiencies, too. What’s surprising is that there seems to be no pattern in my table. I had expected more speculative sectors to have stronger links between yields and subsequent returns because they are more likely to be driven by sentiment and so more likely to become over- or under-valued. But this isn’t the case. Yields on mining stocks are only weakly correlated with subsequent returns and for IT stocks the ability of yields to predict returns is only a little better than it is for the All-Share index. And food producers and pharmaceuticals are not especially speculative, and yet yields here do predict returns.

Whatever the reason for the pattern, investors have a problem. There are some sectors where yields are above their 20-year average, which suggests these are cheap. But some of these are sectors where yields don’t predict returns or do so only weakly, such as oil and miners.

On the other hand, though, there are many sectors where yields are below average and where yields do predict returns: these include food producers, support services and – alas – the whole market.

What we have, then, is the wrong sort of value. Where there appears to be value in the market it tells us little about future returns, but where there are expensive segments is where that expensiveness does convey a signal about the future.

Of course, it’s possible that past correlations will break down. And it’s even more possible that apparently expensive stocks will carry on rising simply because of momentum effects. Nevertheless, all this amounts to yet another reason to be cautious about equities.