Confirmation Bias, The Investor’s Curse

January 30, 2010

102

Confirmation Bias, The Investor’s Curse

Clinging to the Wreckage of Old Ideas

The problem of confirmation bias – the tendency of people to seek evidence confirming an already held opinion and to avoid looking for that which might upset their carefully constructed mental models has attracted a lot of attention from researchers. It occurs across all domains of human endeavour and triggers all sorts of implausible behaviour, yet investors and institutions remain in its thrall.

We find examples in law courts and doctor’s surgeries, in scientist’s laboratories and the lairs of legislators. So we shouldn’t be surprised to find it coursing through the veins of economists and investors, colouring their every thought and structuring their every idea. Of course a rational market participant, faced with a theory built on a crumbling cornerstone will abandon their ideas and look for some new ones. As you’d expect, therefore, we do no such thing, clinging irrationally to the wreckage of our dreams as they collapse around us.

Disconfirming Science

Science is perhaps the area in which we might expect confirmation bias to be least effective. After all, the processes of science are built around the institutionalisation of disconfirmation. New ideas have to run the gamut of envious colleagues who generally hate nothing more than a smart ass and take great pleasure in proving them wrong. Despite this scientists have generally proven remarkably reluctant to give up discredited theories. Indeed science has often proceeded for decades on the basis of ideas that could easily have been disconfirmed had anyone been sufficiently motivated to do so.

So, for instance, the idea that the universe is stable and unchanging was held way into the twentieth century despite the fact that Newton’s laws of gravitation implied something different. To be precise, they suggested that unless the universe was actually expanding gravity should be causing all the bits of matter in the cosmos to hurtle together towards a central point in a manner calculated to render discussions of globalisation, polar bear populations and national debt levels redundant. Yet when Einstein’s theories also predicted an expanding universe he simply made up a special constant, which he later had to ruefully remove, to stop this unwanted behaviour so ingrained was the assumption of a universe in stasis.

See this excellent paper by Raymond Nickerson for a review of the history and guises of confirmation bias. From ancient number mysticism, through witch hunting to government policymaking, medicine, judicial decisions and science Nickerson traces the ubiquitous course of this insidious curse.

A Paradigm Buster

So even when disconfirming evidence is present beyond all reasonable doubt scientists are wary of abandoning a theory in which they’ve invested a lot of time and effort. In fact usually scientists’ first approach is to modify their existing theory just enough to accommodate the anomalous data. Eventually an accumulation of such anomalies forces some people to acknowledge that there’s probably something a tiny bit awry with the underlying concept, opening the way for a paradigm shift – a radical change in the way that a idea is thought about.

Now given that science is the one area of human endeavour where you might expect confirmation bias to be conquered this doesn’t bode well for less well policed disciplines. To take this idea to its extreme conclusion there’s even a suggestion that researchers into confirmation bias are themselves selectively choosing their theories to fit the evidence.

The standard explanation for these biases is motivational – people are motivated to find facts to fit their own theories, to maintain consistency of their internal concepts. Yet an alternative view is that people make these mistakes because their cognitive capabilities are limited – we’re only capable of thinking about one thing at a time, and we just don’t have the bandwidth to handle alternatives.

While both these factors probably have some involvement it seems to me that a trait like this which tends to bias people in favour of things they’re attached to like a political party, a religion, a sports team or a nation would have an adaptive advantage in generating social cohesion. Anything that made people stick together in a social group is likely to have been beneficial to proto-humans in a dangerous world.

Wason’s Rule Discovery Test

The classic confirmation bias task is the Wason Rule Discovery Test. In this participants are presented with a series of numbers, told that they follow a certain rule and asked to figure out what it is by choosing further sequences of three numbers. The sequence given is 2, 4, 6: what’s the rule?

Pretty much everybody forms some hypothesis and asks questions to confirm it but very few people end up figuring out the correct answer which is any three numbers in ascending order. The reason this is so difficult is that most people don’t ask questions to disconfirm their hypothesis because they’re not trying to break their own rules. This seems to be a fundamentally difficult thing for humans to do, and this difficulty has implications for every area of human activity, not the least of which is investing.

Cocooned CEO’s

Many executives of companies, cocooned in their own little worlds and rarely receiving negative feedback, develop their own intransigent views that are impervious to disconfirming evidence. Stuart Sutherland in Irrationality has a telling anecdote:

"I gave an oral presentation of my results to a party from the distiller’s company, which was headed by the managing director … When I said anything with which he agreed, he would turn to his colleagues and announce …"Dr Sutherland’s a very smart man. He’s absolutely right. When, however, my findings disagreed with his own views, he said, "Rubbish, absolute rubbish." I need never have undertaken the study for all the notice he took of it".

We’ve all met people like this and the more remote they are from the real world and their customers the more likely they are to behave like complete idiots. Managements that completely isolate themselves from their end users should be treated with caution – nothing keeps you grounded like an annoyed customer.

Disconfirming Economics

Economics itself is suffering from a paradigm shift as traditional economics tries to adjust to the onslaught of behavioural psychology. Many of the old theory’s defenders prefer to point out the failings of the new approach – which are many and manifest – yet leave unanswered the very real issues that the errors of economic policies have engendered. To argue, as they do, that behavioural finance isn’t founded on any underlying theory and that it’s often contradictory is fair enough. To suggest that this is grounds for ignoring it isn’t.

Traditional economists are attempting to rescue their discipline in the age old fashion, by tweaking their models to attempt to accommodate the new data. The difficulty with this approach is that if the underlying ideas of behavioural finance are even roughly correct this attempt to save the old models is bound to fail. Behavioural biases such as confirmation bias will inevitably cause people to behave in non-Bayesian (that is, non-rational) and, even worse, unpredictable ways as the plastic brain adapts to the ever-changing world.

The inability of behavioural finance to generate predictions on a grand scale is one of the accusations levelled at it by its detractors. However, the fact that traditional approaches do generate predictions isn’t a saving grace, given that these predictions usually turn out be horribly wrong just when you need them. Better a model that doesn’t give you false confidence in the road ahead than one that continues to confidently send you across the shaky bridge of debt over the chasm of economic doom.

Beating Bias, Being Biased

Individual investors can, at least, attempt to counter confirmation bias. A simple grasp of decision trees is a useful tool which at least forces people to consider some alternative scenarios. Another approach, as we’ve discussed before, is to build feedback mechanisms into your investing techniques.

Of course, the perceptive readers of this journal will note that its pursuit of a theory that economists from the traditional background are plagued by the incommodious effects of a confirmation bias against behavioural finance is itself evidence of confirmation bias. To deny this is futile, guilty as charged. But of course, just as sometimes there really is someone out to get the paranoid so sometimes the irrationally confident really are right. Often it’s better to be lucky than smart.