Investors overreact. If you’ve ever believed a share was cheap, you’ve thought that investors have overreacted to bad news. And the dividend yield has for years been a decent predictor of medium-term returns on the All-Share index precisely because, over time, investors react too much to good or bad news, driving prices up or down too far.
Investors underreact. They are slow to sell shares suffering bad news and slow to buy after good news. This is why we see post-earnings announcement drift – shares rise even days after good earnings reports – and why there is momentum in shares and other assets. “The existence of momentum is a well-established empirical fact,” wrote AQR Capital Management’s Clifford Asness in 2014, a fact which has remained true since then.
You might think these claims contradict each other. They don’t. What we have here is an example of a point made by the great political scientist Jon Elster. The social sciences, he said – and investment is applied social science – are a collection of mechanisms. Sometimes such mechanisms generate overreaction and sometimes underreaction. It all depends on the context.
One mechanism generating overreaction is our tendency to overweight new information and underweight older but still important facts. Some recent experiments at the MIT by David Thesmar and colleagues have demonstrated this. They got people to forecast simple statistical patterns and found that the most recent observation “has disproportionate influence on expectations”. People overreact to it.
This is an example of something noted by the Nobel laureate Daniel Kahneman. People, he says, underappreciate the power of regression to the mean; for example, they see a footballer play one or two good games and infer that he’s a great player when in fact he is simply playing unusually well and is about to revert to mediocrity. Investors make the same error. Long-term earnings growth is in fact largely unpredictable: a famous paper by Louis Chan, Josef Lakonishok and Jason Karceski has found “no persistence in long-term earnings growth beyond chance”. Short periods of earnings growth, however, trigger hopes of further growth and hence rising prices – which lead to disappointment. Which is why “growth” stocks have often underperformed except to the extent that they’ve been bolstered by falling bond yields.
Another source of overreaction has been pointed out by Florian Zimmermann and Benjamin Enke. They show that people fail to appreciate that some signals, such as news reports or experts’ opinions, are correlated and so they overestimate the weight of evidence and thus buy more than they otherwise would. Beliefs, they say, “are too sensitive to the ubiquitous ‘telling and re-telling of stories’ and exhibit excessive swings.”
Yet another cause of overreaction is our tendency to follow the crowd. Hans Hvide and Per Ostberg have shown that people’s stock selections are influenced by their co-workers, and Brock Mendel and Andrei Shleifer have shown that investors sometimes “chase noise”; they buy because others are buying in the mistaken belief that those others are correct.
Of course, this isn’t always foolish. Sometimes there is wisdom in crowds and you must be very arrogant to believe you are right and everyone else wrong. At other times, though, this causes asset prices to overreact. It’s a big force behind speculative bubbles.
A further cause of overreaction is that small changes in the evidence can cause big changes in narratives. In the mid-1990s, for example, large current account deficits in several Asian economies were framed as fast-growing economies sucking in capital – until they suddenly became signs of excessive borrowing and a crisis. And in 2006 the cooling off in the US housing market was framed as an innocuous correction – until it suddenly triggered a crisis.
All these are mechanisms generating overreaction, either by individuals or the collective market. Equally, though, there are other mechanisms causing underreaction.
One is our tendency to cleave too strongly to our prior beliefs and not accept that we were wrong, which is why we hold on to falling shares. This is often reinforced by wishful thinking: we want to believe our holdings will come good and the wish is father to the belief.
Such excess attachment to our prior beliefs is more likely to the extent that the belief is part of our identity, which is why many religious and political opinions don’t change in the face of dissonant evidence. If being wrong jeopardises your sense of yourself, you’ll not admit error. If selling at a loss undermines your self image as a good investor, you’ll be less likely to do it and so you’ll underreact to bad news.
By contrast, when our identity isn’t at stake we are more likely to overreact to evidence.
There’s also the fact that we have limited attention. Some shares are simply off our radar so we don’t initially notice their good performance – until they do come to our attention and we start buying. We underreact initially to good news simply because we don’t notice it.
You might think all this is hopelessly inconclusive.
In one sense, that’s as it should be. People are complicated and society is complex, so there are no simple robust general rules.
But this doesn’t mean we know nothing. The long-term success of momentum investing tells us that underreaction is common – sufficiently so that has been profitable, at least in the past.
Equally, the fact that markets have sometimes risen too far or fallen too much, so that returns have been predictable by measures such as the dividend yield, tells us that investors do sometimes overreact, especially in the long-run.
And of course, there is a third possibility – that people neither over- nor underreact but update their beliefs properly as new evidence emerges. The very fact that markets are often efficient – as demonstrated by the fact that active managers underperform over the long-run – reminds us of this. Markets are not always or even often irrational, which makes it all the more important to spot the few occasions when they are.
