We cannot judge fund managers accurately from their track record alone.
A recent paper by Victor Haghani and colleagues and investment managers Elm Partners has shown why. They asked: if you had two coins, one fair and one with a 60 per cent chance of turning up heads, how many tosses would you need to be 95 per cent confident of identifying which one was biased?
The correct answer is 143. This a lot. That’s because there’s a good chance of the biased coin coming up tails a lot, and because 95 per cent confidence is a high bar.
Spotting good funds is much like this. One that has a 60:40 chance of beating the market is a good fund. But even it will underperform often enough to cast doubt upon whether its manager really does have a 60:40 chance of outperforming or is just lucky. We need long periods of data to dispel such doubt.
In fact, spotting good funds might be harder than spotting biased coins. When we toss coins, one toss is independent of the next, so we know how many tests we have. With funds, things are more complicated. Imagine a clueless fund manager buys some good stocks by accident and so outperforms for a few months. If there is momentum in share prices he might then continue to outperform. His outperformance will then look persistent – which is often taken as evidence of skill – even though he’s just lucky.
This means we need to see many years of returns before we can infer much about a fund manager’s ability. Mr Haghani says: “Five to 10 years of track record just doesn’t tell us that much.”
Most people, however, don’t realise this. Mr Haghani and his colleagues asked 700 people that coin toss question, and most of them massively underestimated the true answer: the typical answer was just 40, which is less than one-third of the truth.
People then are likely to see skill where there is in fact only luck. Some separate experiments by the University of Warwick’s Nattavudh Powdthavee and Jordi Brandts at the Autonomous University of Barcelona and colleagues have corroborated this. They found that even intelligent subjects are willing to pay others to predict the toss of (fair) coins, if they see that those people have successfully predicted a few previous tosses.
You might think these experiments don’t tell us much because the subjects were playing for low stakes and thus lacked strong incentives to judge properly. But we have other evidence that people reward luck, even when the stakes are high. It comes from football matches. Romain Gauriot and Lionel Page, two Australian economists, looked at what happened to all shots that hit the woodwork in the five major European leagues: in some cases, the ball rebounded into the goal and in others it bounced out. They found that coaches were significantly more likely to select a player for the following game, and to give him more time on the pitch in that game, if his shot went in. Whether it did or not, though, is a matter of luck. This tells us that even experts with strong incentives reward luck rather than skill.
In light of all this, a finding by a team of researchers in London and Lisbon becomes less surprising. They show that, around the world, investors are more likely to buy funds that have done well in the previous three months – even though three months’ is far too short a period to judge whether a fund manager is skilled or just lucky.
The upshot of all this has been described by Bjorn-Christopher Witte at the University of Bamberg: investors can select for fund managers who get lucky to the neglect of genuinely good ones, especially if the latter deliver only moderate outperformance over time and so do not catch the eye. Market forces do not select against bad fund managers, and might even select in favour of them.
All this poses the question: if we cannot trust past performance, what can we do?
One possibility of course is not to try and to hold tracker funds instead. Another possibility is to lower one’s standards and to back funds you have less confidence in.
A third possibility, though, is to look not just at track records but at strategies. A fund manager with little track record but who follows strategies that have proven success (such as defensive stocks) might well be worth backing. Even this, however, isn’t foolproof. There is a constant danger that investors have wised up to the fact that some shares have done well in the past, and so bid up their prices to levels from which future returns will be poor. This happened, for example, with small stocks in the 1980s: investors learned that these had done well for decades and so piled into small-cap funds only to see a decade of bad underperformance. A similar danger applies today to stocks with what Warren Buffett calls “economic moats”: their great performance in recent years might be reversed if investors have now cottoned onto their merits.
The bottom line here is simple. As economists at City University Business School have concluded: “finding successful funds ex-ante is extremely difficult, if not impossible”. People are, however, unwilling to realise that some things just cannot be known.
