Join our community of smart investors

Why markets crash

The triggers for stock market crashes can be imperceptible, even after the fact. This does not, however, mean that equity returns are unpredictable.
October 16, 2018

Last week’s stock market falls are being blamed upon worries about the trade war, rising US interest rates, and high US valuations. In themselves, however, such explanations make no sense.

To see why, cast your mind back to May. Back then every dog on the street knew that the trade war jeopardised the global economy, that the Fed would raise interest rates, and that US valuations were high. But the FTSE 100 was over 7700 then. So why should it be so much lower five months later when these key facts haven’t changed?

This question tells us that macroeconomic conditions alone do not suffice to determine share prices. As Robert Shiller pointed out back in 1981, prices are much more volatile than “fundamentals” such as profits and interest rates. Instead, price changes and changes in volatility emerge from trading activity itself. We must therefore abandon any idea of a simple hydraulic link between share prices and 'fundamentals', and adopt three other but compatible perspectives.

One is that of the Keynesian beauty contest. Investing, said Keynes, is a like a newspaper competition in which people have to guess which faces other competitors considered the prettiest; these were popular in the 1930s. The point is to anticipate what others will believe, or even what others will believe others will believe. If, for example, I believe that others will believe that rising interest rates will be no problem, I’ll not sell. But I will if I believe others will think they are a problem, or even if I believe others will believe that others will think them a problem. Prices move because of beliefs about beliefs, or even because of beliefs about beliefs about beliefs.

A second perspective is to regard equities as bets. Think of their prices as the probability-weighted average of different payoffs. Let’s take a simple example of just two possible scenarios – a benign one in which the FTSE 100 is at 8000, and a nasty one in which it is at 6000. If investors attach an 85 per cent probability to the benign scenario and a 15 per cent chance to the nasty one, the index will be at 7700. (I’m ignoring discounts rates for simplicity.) If, however, these probabilities shift to 55-45 per cent the FTSE 100 will fall to 7100.

Simple as it is, this example tells us two things: that small changes in probabilities of different scenarios can generate big price moves; and that prices can fall a lot even if investors continue to believe that a benign outcome is most likely.

Our third perspective is that of emergence. Don’t think of the market as if it were a single person. Ben Graham’s notion of a “Mr Market” is misleading in our context. Think of it instead as an ecosystem containing three species: bulls, bears and trend-followers. If bulls and bears are roughly equal, prices will be stable and trend-followers will have nothing to trade on. If, however, bears become more plentiful then prices will fall and trend-followers will amplify the fall. Why might bears come to outnumber bulls? Our other two perspectives give us clues. Small shifts in the probability one attaches to nasty scenarios, or a shift to worrying that others will worry can tip some from bullishness to bearishness, thus beginning an information cascade that drives prices down.

And here’s the thing. The triggers for such shifts can be small and even unobservable. They need not consist in the publication of high-profile macroeconomic data. And they are generally unpredictable.

This doesn’t, however, mean we can’t foresee where the market is heading.

Sometimes we can predict market moves in a different way. We have some lead indicators of returns such as the dividend yield, foreign buying of US equities, ratios of share prices to the money stock, or even the time of year. These all work in the same way – as indicators of sentiment (or of sentiment about sentiment!) A low yield, high foreign buying or high ratio of share prices to money stock all tell us that sentiment is high, or that investors are not worried that others will worry, or that they are attaching high probabilities to benign scenarios. When this is the case, simple mean-reversion warns us that shares are more likely to fall than to rise. Because mean-reversion is more likely to operate over longer time horizons – a year or more rather than a month – so there is more predictability in longer-term returns than in short-term ones.

These lead indicators do not, however, tell us why prices will fall. As Columbia University’s Jon Elster has pointed out, there is a massive distinction between explanation and prediction. Sometimes we can explain but not predict, and sometimes we can predict but not explain. Investors must not forget this distinction.