Shares are cheap. The dividend yield on the All-Share index is slightly above its post-1990 average, while the non-financial price-earnings ratio is some 35 per cent below its post-1990 average. However, it matters enormously why shares are cheap.
There are two possibilities here, with very different implications for investors. One possibility is that shares are cheap because investors believe they are unusually risky and so require high expected returns to compensate for the dangers of holding them.
The other possibility is that they are cheap because they offer low future growth in dividends and other cash flows.
The distinction matters. If shares are cheap because investors require high expected returns on them, then there's a case for buying them if you are less risk-averse than the average investor because if the perceived risks don't materialise, prices will rise. But if shares are cheap because future growth will be low, there's no reason to expect prices to rise and so no reason to buy. Remember the Ratner principle: some things are cheap for a very good reason.
So, how can we tell whether shares are cheap because they are risky or because they offer low growth?
It's tricky. The obvious measure of growth expectations - analysts' earnings forecasts - have often been wildly wrong in the past, perhaps partly because corporate growth in inherently unpredictable. For this reason, we shouldn't assume that shares are pricing in analysts' expectations.
Nor can we wholly rely upon the obvious measure of risk - the Vix index. This measures short-term volatility rather than long-term risks, and can change very suddenly. What's more, a low Vix can be consistent with uncertainty being high. For example, if one trader expects something bad to happen and another doesn't, there might be lots of trading but little move in prices and hence little volatility and a low Vix, even though there will be lots of uncertainty.
There is, though, another way of adjudicating between the risk and growth explanations. We can look at correlations between equity valuations and subsequent returns. If valuations are low because shares are seen as riskier, there will be a positive correlation between the dividend yield and subsequent returns. If, however, valuations are low because of low growth expectations, there will be no such correlation.
This method tells us that the correlation between the dividend yield and subsequent monthly return on the All-Share index has usually been positive, implying that a relatively high yield is normally a sign of high risk and high expected returns. But there have been three exceptions to this. One came in the late 90s, when euphoria about the 'new economy' raised growth expectations, with the result that low dividend yields (high valuations) were followed by good returns. The second came as the tech bubble burst and growth expectations were revised down. And the third came during the 2008-09 crisis when growth expectations were lowered and when some of the risks investors fretted about actually materialised.
In the last three years, the correlation has been just under 0.2. This implies that variations in valuations have been mostly due to variations in risk; buying on dips has - on balance - tended to pay off. However, the correlation is not especially high, which suggests that there is an element of lower growth expectations in our low valuations.
If we perform the same exercise for sectors, we find three where the correlation between yields and subsequent returns has been unusually low recently: oil & gas, mining and food retailing. This suggests that these have suffered a growth derating.
By contrast, some other sectors have had correlations between yield and return above their long-term average: food producers, telecoms and industrials. This suggests prices here have been swayed more by changes in risk perceptions than growth expectations.
Although these results seem clear - or as clear as we can expect in noisy data - their interpretation is ambiguous.
If you want a bullish message, you could interpret the positive correlation between yield and subsequent return as a sign that the market is indeed cheap now because it is unusually risky, and there's a better than evens chance of this leading to good returns.
But if you want a bearish message, it lies in the fact that the correlation can sometimes change. The fact that the correlation has been positive recently might tell us that investors are due for a nasty awakening. If economists such as Robert Gordon of Northwestern University and Tyler Cowen of George Mason University are right, we face years of sluggish global growth, a possibility that investors are not (yet) heeding. If they come to do so, then share valuations might fall further without this leading to higher future returns.
MORE FROM CHRIS DILLOW...
Chris blogs at http://stumblingandmumbling.typepad.com