My chart shows what I mean. It shows sensitivities of changes in the All-Share index to the lagged dividend yield, measured over five years of monthly data.
On average, since 1985, these sensitivities have been modest but statistically significant. For example, each percentage point above average dividend yield has been associated with All-Share price changes in the following six months being 4.1 percentage points higher than average. The sensitivity of monthly price changes to the yield has been smaller as you'd expect, because a cheap market can get even cheaper in the near term. But it's also been slightly positive.
There's no surprise here: this just tells us that cheap markets see higher subsequent returns and expensive ones lower returns. What might be a surprise, though, is that this average relationship conceals a big variation over time. And in the past five years returns have been more sensitive to yields than usual. During this time, a 0.1 percentage point above-average yield has been associated with prices rising 1.8 percentage points more than usual in the following six months. This means the sensitivity of prices to yields has been four times as great as the post-1985 average.
To understand why this has been the case, we should look at those cases when the opposite was true - when there was a perverse relationship between yields and returns with low yields leading to high returns and high yields to low returns. There have been two such occasions: around the time of the tech bubble and burst; and the financial crisis of 2007-08.
These were times when investors believed, not always correctly, that there had been a paradigm shift in the market. For example, in the late 1990s investors ignored the message of low yields because they thought we were on the verge of a new era of faster growth. While they believed this, low yields led to prices rising.
The opposite happened in 2007-08. Rising yields then were a sign not that the market had become better value, but that disaster was about to strike. Apparently 'cheap' markets got a lot cheaper in late 2008. That led to a negative correlation between yields and subsequent returns.
These episodes tell us that uncertainty and instability weaken the relationship between valuations and subsequent returns. By the same token, economic stability strengthens the relationship. If we're not entering a new era of faster growth, low yields will indeed lead to low returns. And if disaster doesn't strike, high yields will lead to higher returns. In other words, the more normal the times, the stronger the link between yields and subsequent returns.
All this is consistent with the fact that there's a significant link between the correlations in my chart and market volatility, as measured (arguably imperfectly) by the Vix index. High volatility is associated with a weak relationship between yields and returns. And low volatility sees a stronger relationship. Not only have we seen low volatility recently, we also saw it in the mid-2000s - when talk of a 'Great Moderation' was accompanied by a strong relationship between yields and returns.
As I write, the dividend yield is slightly lower than it has been in recent months. The current strong relationship between yield and returns suggests this is a reason to be wary of equities now.
However, we know that the link between yields and returns is unstable. The economic and market stability that generates strong links doesn't last for ever.
It can break down in one of two ways. The nice possibility is that we see a bubble of the sort we got in the late 1990s, when investors convince themselves that valuations don't matter because future growth will be high. The nastier possibility is that we suffer an economic shock that causes a long-lasting rise in yields and hence fall in prices.
Buying now requires one to believe that the nice possibility is more likely than the nastier one. Personally, I don't think there's much evidence to support such optimism.
