Hating uncertainty

Chris Dillow

Chris Dillow

It's a cliché – because it's true – that markets hate uncertainty. What's not so well appreciated, however, is that people hate uncertainty even if it is good for them, as some experiments by Stefan Trautmann of Tilburg University and Richard Zeckhauser of Harvard have shown.

They asked people to bet on whether a red or black poker chip would be drawn from a bag of 10 chips. They had a choice of two bags: one containing five blacks and five reds, and one containing an unknown proportion. 77 per cent of people preferred to bet on the five-five bag.

This is the standard Ellsberg paradox, first pointed out by Daniel Ellsberg in 1961; people prefer risk – known probabilities – to uncertainty (unknown probabilities) even if the payoffs to both are the same.

However, Professors Trautmann and Zeckhauser then asked subjects to choose between two successive bets – either both on the five-five bag or both on the unknown bag. Rational people should prefer the latter, as this has a slightly higher pay-off. This is because if you see (say) a red chip come out of a bag with unknown proportions on the first draw, you learn that there's a chance that the bag contains a majority of reds – because the possibility that it contains 10 blacks has been eliminated while the possibility that it contains 10 reds is live. You can therefore improve your chances of winning by betting on red in the second round.

Despite this clear superiority of the unknown bag, most subjects still preferred to bet on the five-five bag. This was not because a single red chip drawn from a bag of 10 conveys little information about the odds. Even when people were asked to choose between two bets on bags of just four chips – where you learn a lot from one draw – they still preferred to bet on known proportions.

The message here is clear. People hate uncertainty – as distinct from risk – so much that they avoid it even if the uncertainty is actually good for them.

A big reason for this is that people don't, in fact, learn from uncertainty in the way that they should. The professors found that, of those who chose to bet on the unknown bag, two-fifths failed to learn from the first draw, in the sense that if red came out first time, they bet on black the second time. For many people, the gambler's fallacy overpowered rationalBayesianupdating.

It's not just the gambler's fallacy, though, that stops people learning. In a separate experiment Jacob Goeree and Leeat Yariv at the California Institute of Technology found that, with similar choices, almost half of people preferred to bet on an uninformative social signal ("this is what others chose") rather than an informative signal ("red came out last time"). Our taste for conformity, therefore, can also override rationality.

In this context, those people who chose the five-five bag over the unknown one might not have been entirely irrational. If you know you lack the discipline or ability to learn from uncertainty, it's reasonable to avoid it. People who avoid uncertainty are like Ulysses (or Odysseus) tying himself to the mast to avoid being tempted into rocks by the Sirens – they're adopting a self-disciplining strategy.

Whatever the cause of the hatred of uncertainty, it has important implications for investors.

First, people avoid assets they regard as unfamiliar. Boston University's Larry Epstein has shown that even moderate levels of aversion to uncertainty can cause people to avoid assets completely. This might help explain the "home bias" in equity holdings – our preference for owning shares based in our home country – or even region – rather than foreign ones.

Secondly – and this is the counterpart of the first – investors like assets which they feel are not uncertain. This preference, however, can be expensive. Hans Hvide at the University of Aberdeen and Trond Doskeland at the Norwegian School of Economics have found that investors who own lots of shares in the industry they work in actually earn below-average returns. In the real world, then, as well as the laboratory, investors sacrifice returns when they try to avoid uncertainty.

Thirdly, bubbles are possible, because investors sometimes follow uninformative social signals ("everyone’s buying!") rather than private ones (this looks expensive). It's no accident that bubbles tend to be biggest in assets that are especially uncertain, such as railway stocks in the 1840s, emerging market shares in the early 1990s or tech stocks in the late 1990s.

Finally, big swings in volatility can cause shares to fall. This is because when volatility varies a lot, we cannot estimate future probabilities of a share price falling. Uncertainty therefore increases. And because people hate uncertainty, shares will fall to price in an uncertainty premium. This means that, after a bout of changing volatility, shares will often be low enough to offer good returns to the investor willing to take on uncertainty.

For this reason there is a strong correlation (0.47 in weekly data since January 2007) between the six-month standard deviation of the CBOE's VIX index (the volatility of volatility) and returns on MSCI's world index in the subsequent six months. A high volatility of volatility leads to good returns. For example, the VIX was very volatile in the second half of 2008 – it rose, fell and rose again – and this led to big rises in shares in the first half of 2009. Conversely, the stable volatility of 2007-08 and early 2011 led to falling prices, in part because it meant there was little uncertainty premium.

Herein lies the immediate relevance of all this. It's reasonable to suppose that equities are now subject to more than unusual uncertainty – in the sense of unknown probabilities. One measure of this is that the standard deviation of the VIX in the past six months has been twice its post-1990 average. More informally – and I suspect more importantly – it is the case that the fate of the market depends in large part upon whether the euro area's debt crisis will end happily or not. And this is a genuinely uncertain prospect, in the sense that it is silly and fatuous to assign a quantifiable probability to it.

This raises the possibility that, if the crisis is resolved, shares could soar. This won't merely be because the chance of disaster will be then priced out of shares – important as this will be. It will also be because uncertainty gets priced out – and the more markets hate uncertainty, the bigger this effect will be.

In this sense, it is possible – we cannot be more precise! – that shares could get an enormous boost next year.

Related topics

Subscribe today

Full access for just £3.37 a week:

• Tips and recommendations - to beat the market 
• Portfolio clinic & Mr Bearbull - build a well-planned portfolio 
• Expert tools - track and manage investments effortlessly
• Plus free delivery to your home or office

Subscribe Now