There is a rare neurological condition known as the Capgras delusion, in which people believe that their relatives have been replaced by identical imposters. A doctor once asked a patient with this delusion: ‘isn’t it unlikely that your father would have been replaced by an imposter?’ to which the patient replied: ‘Yes. I’d never have believed it unless I’d seen it myself.’
This story shows that prejudices can be good things. A strong prejudice that people don’t get replaced by imposters would have saved the patient from the misleading evidence of his own eyes.
The point broadens. A lot of what we see as investors is potentially misleading: a good day for equities is not necessarily the start of a bull market; some good news about a company needn’t be sufficient basis for a good investment. And so on. A lot of financial news is in fact just noise – mere verbiage that conveys no signal about future returns and therefore is no guide to how investors should act. The investor who doesn’t have any prejudices doesn’t have any way of distinguishing between noise and signal, and will therefore often trade on the former. In doing so, he not only risks losing money for himself, but also distorts prices for other people.
Investors, therefore, need prejudices – or, in fancier language, Bayesian priors. . These tell us what news is noise and so should be ignored. But, of course, we need good prejudices – ones that have a decent chance of being right. What follows are a few prejudices I have.
There are very few ways of systematically beating the market
Of course, hundreds of stocks beat the market. But there are almost 2,000 shares listed on the main market or Aim, so many will do so simply because of random variation. That’s noise. The question is: is it possible to spot the winners in advance? And the answer is: yes, to a very limited degree. We have good evidence that momentum stocks, value and defensives do better than they should.
This evidence consists of two big things. . One is that it comes from multiple data sets. For example, momentum investing works not just in equities around the world, but also in commodities, currencies and even sports betting. Defensive investing also works in assets other than equities.
Secondly, there’s a good explanation for why such stocks should beat the market. What we need is a good reason why intelligent investors leave money on the table – why they haven’t bought underpriced shares and thus bid up their prices. In the case of momentum and defensives, we have such a reason. It is because such strategies expose fund managers to benchmark risk: they could underperform a rising market and so cost the manager his job.
Most other alleged anomalies don’t have such a strong basis in evidence or theory. Of course, researchers have found countless possible ones. But this is simply because there are patterns in random data. As Duke University’s Campbell Harvey has said, most such findings are “likely false”. Investors cannot rely upon them persisting.
Growth stocks, Aim and new flotations are often overpriced
If some stocks systematically beat the market, others systematically underperform it. The most obvious of these are small growth stocks, and newly floated ones. The Aim index has fallen 27 per cent since its inception in 1995, during which time the All-Share index has almost doubled.
A big reason for this is that investors are overconfident. They overestimate their ability to spot future growth and underestimate the fact that business owners know more than they do and use this knowledge to float shares at the best possible time.
Again, these anomalies persist because the smart money cannot easily eliminate them. The fund manager who believes an Aim stock to be overpriced can’t do much about it. Even if he is allowed to short-sell a stock, he might not be able to simply because the share is too illiquid or volatile.
Stock markets are mostly micro efficient, macro inefficient
“The average UK equity mutual fund manager is unable to deliver outperformance net of fees,” concluded David Blake of Cass Business School and colleagues in one recent study. “Active fund managers as a group have underperformed their benchmarks,” says Vanguard's Peter Westaway.
These findings tell us that most fund managers are unable to exploit the few methods we know of beating the market. Investors should therefore steer clear of high-charging funds, as their fees often aren’t justified by better performance.
This might be partly because fund managers are wedded to the use of ‘judgment’ rather than simple algorithms. It’s also partly because their few good ideas are diluted by the need to own dozens of stocks to reduce liquidity risk. But it’s also because of something else; a few anomalies aside, the stock market is to a large extent informationally efficient. If you try to use judgment to beat it – rather than rely upon the tried and trusted defensive, value and momentum factors – you might well fail.
However, this fact is quite consistent with the possibility that the market as a whole will be occasionally mispriced. There’s a simple reason for this. If you think that, say, BP is overpriced relative to Royal Dutch Shell, you can short BP and go long of Royal Dutch Shell. If enough investors do this, the mispricing will quickly disappear. But what if you think the whole market is overpriced? It’s much harder to go short of it without taking on the huge risk that the market will become even more overpriced. “There is little that rational investors can do optimally to exploit, and hence, eliminate excessive volatility,” concluded Bernard Dumas of Insead. This means that mispricing of the aggregate market is more likely than mispricing of individual stocks. The late Nobel laureate Paul Samuelson once claimed that stock markets were “micro efficient but macro inefficient”. He wasn’t wholly right – the good performance of defensives and momentum suggests the market isn’t entirely micro efficient – but he was nearly so.
Gibrat’s law is mostly true
This law says that growth is independent of size – that big companies are as likely to grow well as small companies.
The strongest reason for believing this is simple history. We’ve had more than two centuries of capitalism. If there were a tendency for small companies to grow faster than large ones, we should by now have a flat corporate landscape, with all companies the same size. If, on the other hand, large companies grew faster than small ones, the economy would be dominated by a few giant monopolies. Neither is the case.
One reason for this is that there are massive barriers in the way of a company becoming a giant: diseconomies of scale. Of the 1,976 stocks quoted on the main market or Aim, only 267 (13.5 per cent) have a market cap of over £1bn, and only 46 (2.3 per cent) have a market cap of over £10bn.
This tells us two things. One is that we must be sceptical of valuations that imply that a small company will grow a lot: very few actually do so. The other is that we should not expect small-caps to systematically beat larger ones over the long run. Yes, they can have a few years of outperformance – helped, recently, by the fall in mega-cap miners. But over the long run, small-caps and the FTSE 100 should rise in line with each other – which is what has happened in the past 25 years.
There’s little predictability in the short term, but a bit more in the longer term
Take, for example, the ability of the dividend yield to predict subsequent changes in the All-Share index. Since 1985, it has explained only 1.8 per cent of the variation in subsequent monthly changes, but 22.9 per cent of subsequent one-year changes, and 44.7 per cent of three-year changes.
I suspect that a similar pattern would be found in any other potential predictor of returns, such as consumption-wealth ratios (although I’d welcome being proved wrong). This suggests to me that attempts to predict market moves are mostly futile for the short run – they are mere noise – but less so for the longer run.
Things take longer than you think
This fact has another implication: things take longer than you think. If the market or a share looks underpriced or overpriced, it can become even more so before it corrects.
Perhaps the most high-profile victim of this fact was the late Tony Dye, chief investment officer at Phillips and Drew. In the late 1990s, he thought equities generally and tech stocks in particular were overpriced. Sadly for him, they became even more so and he was sacked before eventually being proved right.
Similarly, in the early and mid-2000s several pundits warned that house prices were too high and that there was the danger of a financial crisis. They were eventually proved right, but not before those who had followed their advice lost money by betting on it.
Keynes was right: “Markets can remain irrational longer than you can remain solvent.”
Luckily, there’s a way of managing this risk, by using a simple rule proposed by Mebane Faber of Cambria Investment Management. A policy of buying when the market is above its 10-month moving average (measured at monthly frequencies) and selling when it is below it can give us the best of both worlds. It allows us to ride the upward leg of a bubble while getting us out not at the top, but before most of the bubble has deflated.
Sell in May; Buy on Halloween
Although monthly returns are mostly unpredictable, there is one thing that can help us foresee them – the calendar. Equity returns are much higher in the winter than in the summer: since 1966, the All-Share index has given an average return of 8.9 per cent from Halloween to May Day, but minus 0.8 per cent from May Day to Halloween. This is one of the strongest patterns in stock markets: Ben Jacobsen of Edinburgh University and Cherry Zhang of Nottingham University Business School have shown that it exists in almost all markets for most of history.
What’s more, all of the months from May to October (except for August) have given worse returns than all of the months from November to April.
This rule isn’t foolproof, as the fact that shares have fallen since last Halloween reminds us. But then no stock market strategy is. It is, however, one of the strongest market patterns I know.
Risk is big, ubiquitous and fat-tailed
Since 1900, the standard deviation of real annual equity returns has been 21 percentage points. This implies that there is almost a one-in-four chance of us losing 10 per cent or more in a single year.
What’s more, the chance of really big losses is greater than you think. Put it this way: since the start of 1990 the average rise in the All-Share index has been 0.1 per cent per week with a standard deviation of 2.22. If returns were normally distributed, we’d therefore expect to have seen one or two weeks in which the market fell by more than 6.56 per cent. In fact, we’ve seen nine.
Big falls are more likely than they ‘should’ be. This is because extreme returns are distributed not normally but in the shape of a cubic power law.
However, volatility is not the only risk. There are many others, among them:
■ Liquidity risk. The danger of being unable to sell an asset quickly at a decent price. This afflicts housing, but also many alternative assets such as cars or wine.
■ Correlation risk. Previously uncorrelated assets can fall at the same time, especially in bad times.
■ Cyclical risk. Some assets are more exposed to recession than others.
■ Tail risk. Corporate bonds might offer a nice steady return, but there’s a small chance of a massive loss if the company defaults. The same is true for some options strategies that involve selling insurance against market falls. All these dangers mean you must be wary of anyone offering a high ‘risk-free’ return: I would dismiss anyone doing so as a charlatan. They also mean we must be sceptical about measures of risk. Ask: what sample are they drawn from? Do they adequately capture tail risk?
Optimisation is impossible; satisficing is good enough
There is one thing that’s certain about investing. We will make mistakes. We’ll lose money on some investments and miss out on some big gains whatever strategy we follow. All of my prejudices are wrong sometimes: there have been a few great growth stocks; the market has occasionally done well in the summer; momentum stocks have sometimes underperformed; and so on.
We can’t be right all the time. Instead, the best we can do is to follow a few rules that we know to be right on average over the long run. For me, such rules include always holding some cash to mitigate losses, and avoiding the better-known ways of losing money, such as by trading too much or buying lottery-type stocks.
As the late Herbert Simon said, in a complex world optimisation is not possible. Instead, we should satisfice - make do with strategies that are good enough. For example, I hold tracker funds not because they are optimal as efficient market theory says, but because they are a decent low-cost easy second best.
We know less than we think
I’ve argued that prejudices can be a good thing. This poses the question: why, then, do they have such a bad reputation?
One reason is that some prejudices can be plain wrong. In fact, in a noisy world even those that are right on average will be wrong sometimes.
Another is that we can stick to them too much, and not update our beliefs sufficiently as new evidence emerges. This is one reason why momentum investing works: investors who have a strong prejudice against a stock underreact to good news, causing its share price not to immediately rise sufficiently to embody the news. The upshot is that the share gradually drifts up.
And herein lies the problem – not just in investing but in life generally. What’s wrong are not prejudices in themselves – we need them to filter out the massive noise in news – but rather a tendency to cleave too strongly to them. There’s a difference between prejudice and bigotry.
The historian AJP Taylor used to tell the story of when he applied for a job at an Oxford college. “The president of the College concerned said to me sternly: ‘I hear you have strong political views.’ I said: ‘Oh no, President. Extreme views weakly held.’” Investors should follow his lead.
