Attention, please

There is a contradiction at the heart of economics – one that explains why some shares are overpriced and others underpriced, and why investors make systematic errors.

One of the most basic and commonsense ideas in economics is that people economise on the use of scarce resources: this is why we have turned down the thermostat as gas prices have soared.

It’s not just things such as gas and diesel that are scarce, however. So, too, is our attention: we just don’t have enough of it to notice all the facts that can possibly influence share prices. This means that textbook financial economics has a problem. It claims that markets are efficient because all relevant information is in the price. But if our attention is limited, how can this be? We cannot even notice the countless number of facts that matter for shares, let alone price them in. Information is vast, but our ability to attend to it is small. If we economise on scarce resources, markets might not be so efficient after all. 

An embarrassingly obvious fact alerts us to this possibility. It is simply that in order to buy a stock you must first be aware of its existence. But the things that make you aware of it are not necessarily things that make it a good stock. Research by Brad Barber and Terrance Odean, two economists at the University of California, has shown this. They studied users of the Robinhood app, which allows people to trade shares easily. They found that such traders were more likely to buy shares on days when their prices had moved a lot (either up or down) because the app drew attention to these shares. But these stocks did not make money; quite the opposite. Barber and Odean found that the 0.5 per cent of shares that were most heavily bought by the app’s users fell by an average of almost 5 per cent in the month after purchase.

Users of Robinhood tend to be less experienced traders. But even skilled investors lose money because of how they direct their attention, as recent research by the University of Chicago’s Alex Imas and colleagues has found. They studied US fund managers and found a strange discrepancy. Whereas their stock purchases tended to make money, suggesting some skill, their sales were terrible; they would have done better just selling stocks at random. The reason for the difference is that their attention was focused on buying, which they did only after assiduous research. Selling, though, was a different process altogether. It was something done with much less thought, merely as a means of raising money to buy other stocks. The shares they sold tended to be those they held only in small amounts and that had recently either fallen or risen a lot. None of these reasons for selling had predictive ability for future returns.

Such behaviour is only natural. If our mind is on one thing we often don’t notice other things even when they are out of the ordinary. A famous experiment at Harvard University by Daniel Simons and Christopher Chabris proved this. They showed subjects a film of people playing basketball and asked them to count the number of passes. After the film, they asked them if they had seen anything unusual during the film. Almost half said no, even though a woman dressed in a gorilla outfit had walked among the players.

Much the same can happen in the run-up to financial crises. The origin of these is sometimes like the gorilla impersonator – straight in front of us without us realising it. In the 1990s, several Asian countries such as Thailand and Indonesia had been borrowing heavily from abroad for years without financial markets much worrying – until in 1997 they did, causing currencies and stock markets to fall sharply. The fund manager and former Bank of England monetary policy committee member Sushil Wadhwani said in 1999 that current account deficits "appear not to matter until, well, they suddenly do!" His point was vindicated during the euro crisis of 2011-12: Spain, Greece and Portugal had been running large external deficits for years, which didn’t seem to matter – until suddenly they did. Similarly, the US housing market began to turn down in 2006, but financial markets did not worry about this for months – until suddenly that all changed.

Correlation concerns

Warning signs of a crisis, however, are not the only things we don’t pay enough attention to. "People tend to ignore correlations," conclude Erik Eyster at the LSE and Georg Weizsäcker at Humboldt University in Berlin from their experiments. This leads to two types of mistake.

One is the failure to diversify sufficiently among assets with low correlations. For example, holders of risky assets such as emerging market equities often pass up the chance to mitigate that risk by holding bonds. And we don’t appreciate sufficiently that a great virtue of cash is that it protects us when other assets are falling at the same time.

The other error is to overdiversify when assets are highly correlated. For example, investors hold too many shares of companies in their own industry or near where they live. And they own expensive actively managed funds that are correlated with each other, thereby incurring unnecessary charges.

Something else we don’t pay sufficient attention to are background probabilities. There’s a name for this particular error: base rate neglect. We focus so much on the features of specific companies that we fail to see that some types of stock do worse than other types.

One such type are small speculative stocks, which underperform because people pay too much for the slight chance of huge profits. In the past 20 years the FTSE Aim Index has given a total return of just 2.2 per cent. Not only is that way less than the 6.3 per cent total return on the All-Share Index, but it is also less than the rate of inflation.

Another type are newly floated shares. Alan Gregory at the University of Exeter and colleagues have found that newly issued shares underperform comparable ones by an average of over 10 per cent in the two years after flotation – a pattern repeated in the US, as the University of Florida’s Jay Ritter has shown. Recent losses on Aston Martin Lagonda (AML) and Dr Martens (DOCS) fit this pattern.

It’s not just the neglect of background probabilities that lead people to buy into new flotations, though. David Hirshleifer at the University of California at Irvine adds that people aren’t sufficiently attentive to other people’s motivated reasoning. Owners bring their companies onto stock markets not because they generously want investors to share in their wealth, but because they expect to get overpaid for their company – and because they know more about it than outside investors, they often are. Investors, however, don’t pay enough attention to the question: why are the owners selling?

There’s something else we don’t pay enough attention to – the future. We fail to see the power of compounding, which is one reason why we buy actively managed funds: over 10 years an extra half percentage point of annual management charges could cost you £750 for every £10,000 you invest.

We also fail to foresee that changing circumstances will change our preferences. The University of Chicago’s Devin Pope and colleagues have found that houses with swimming pools sell for higher prices in the summer than in the winter and that convertible cars sell better when the weather is better. In both cases, buyers fail to foresee that swimming pools and open-top cars will be useless in the winter.

This is perhaps one reason for the seasonal pattern in share prices: since 1966, the All-Share index has given a total return after inflation of 8 per cent from Halloween to May Day but has lost 0.6 per cent from May Day to Halloween, a pattern repeated in almost all other national stock markets. As the weather improves and the evenings get lighter in the spring, we become more optimistic, but we fail to anticipate that this optimism won’t last into the autumn when the nights draw in. And so prices rise too much. Conversely, in the autumn we become more pessimistic, but fail to see that the pessimism will lift in the spring, and so prices fall too far.

Yes, the future is largely unpredictable even by the best and most attentive minds. But sometimes we fail to foretell it even when it is predictable. That’s because of our lack of attention.

A zero-sum game

So far, we’ve only considered those things to which we pay insufficient attention. But attention is zero-sum: if we pay too little to some things, it is because we pay too much to others.

Some of these things are those that corroborate our pre-existing ideas. We pay more attention to things that make us say 'I knew it' than to things that undermine our ideas. And so we are prone to the confirmation bias, a tendency to become more entrenched in our thinking.

Also, we pay too much attention to information that is already in the price. One reason we do so has been described by Harvard University’s Benjamin Enke and the University of Bonn’s Florian Zimmerman. They show that just as we neglect correlations when diversifying, so we neglect the fact that news reports and opinions are correlated. This can be because experts have similar training and so form similar beliefs, or it can be because we repeat stories that others have told us.

This helps explain perhaps the biggest violation of efficient market theory – momentum, the tendency for past winning stocks to carry on rising and for past fallers to carry on falling. This was first documented by Narasimhan Jegadeesh and Sheridan Titman for US equities back in 1993, but it has since been found in European markets and in non-equity assets such as currencies. “The existence of momentum is a well-established empirical fact,” says Cliff Asness at AQR Capital Management. In the past 15 years, for example, the IC’s no-thought momentum portfolio has risen by an average of 9.3 per cent a year while the FTSE 350 has risen just 1.4 per cent.

The success of momentum investing is what we’d expect to see if our attention were both limited and distorted. A rising share price tends to attract more attention, which brings in new buyers. It also confirms earlier buyers in their belief that they were correct, because we all pay close attention to evidence of our own genius. And this emboldens them to buy more. And because we don’t pay attention to the fact that signals are correlated, we overweight the good news and favourable opinion about rising shares, thus paying too much for them. Through all these mechanisms rising stocks can carry on rising.

All this has an intriguing implication – that perhaps there is a unifying factor underneath the countless errors of judgment that the Nobel laureate Daniel Kahneman, among others, has identified. It is simply that our limited attention causes us to underweight some evidence, thereby overweighting other evidence. “Much of behavioural economics may reflect a form of inattention,” says Harvard University’s Xavier Gabaix.

Of course, there is nothing new about the idea that we have limited attention. Sensible people have always known that the world is more complex than any single person can possibly comprehend. “The 'data’ from which the economic calculus start are never for the whole society 'given’ to a single mind,” wrote Friedrich Hayek in 1945. He thought that the solution to this was to use the price mechanism, because it was a way of communicating a vast amount of information quickly. A rise in the price of tin (to take Hayek’s example) could have many possible causes. Users of tin, however, need not trouble themselves with acquiring knowledge of these. All they need to know is to try to use less tin. In this way, the price mechanism reduces our need to use our attention.

In theory, it could do so efficiently. If some people are focusing attention only on pricing (say) Meggitt shares correctly, while others are focusing on Shell (SHEL) and others on Greggs (GRG) and so on, then it wouldn’t matter that each individual had limited attention. In aggregate, they would price shares correctly.

But the point here is that the mistakes caused by limited attention can sometimes be systematic and so cause systematic price distortions; they don’t always cancel out across thousands of people. When Greggs' share price rose in 2019 (partly because of the attention it got from launching a vegan sausage roll), the company attracted more attention and hence more buying – so that by the summer of that year it was overpriced.

Such inefficiencies don’t contradict Hayek’s argument in favour of free markets. Price adjustments, he wrote “are probably never ‘perfect’ in the sense in which the economist conceives of them in his equilibrium analysis.” For him, they were a useful but imperfect technology. The belief in a perfectly efficient market is not one that the best defenders of a market economy would or should endorse.

But is limited attention really such a bad thing? Not always. Of course, if we’re not paying attention we could miss useful information telling us that a stock is a bargain. Equally, though, we’ll ignore stale information that is already in the price or misleading ideas. Someone who didn’t pay attention last autumn to the hype about cryptocurrencies and non-fungible tokens is better off today than many of those who did.

One of the big dangers for investors is that they become 'noise traders', who act on irrelevant factors rather than genuine signals. If you don’t pay attention, though, you can miss the noise and so not be misled. This is one reason why buy-and-hold investors (at least if they hold tracker funds rather than specific stocks) do relatively well.

Of course, not all of you want to buy and forget. If you must be an active investor, however, at least remember that your attention is limited. What you notice is only a small sample of what there actually is, and very likely it is a biased sample too. Always ask: what am I missing? To what am I not paying attention? The answer could be: a lot.