Our ability to evaluate risk is hedged by all sorts of cognitive biases. It's a miracle that we get anything right.
It’s no miracle at all. Cognitive biases persist because in many cases they are useful. What looks like a bias can in fact be what Gerd Gigerenzer has called a simple heuristic that makes us smart.
Here are a few examples.
1. What are the chances of Manyoo beating S***s on Saturday? We could try to remember Manyoo’s defeats and S***s defeats. The latter come to mind more readily than the former, so we infer than S***s have little chance. The bookies agree. But we’ve just used the availability heuristic.
2. Over-confidence, wishful thinking and the self-serving bias can give us the impetus to undertake risky projects. As a result, people start new businesses, write books and go into the music business in more numbers than rationality dictates.
But the result of this is that society is better off; we get new businesses and products, and more books and music than we would in a rational world. And it’s quite possible that the benefits to the successful outweigh the losses of the unsuccessful.
3. Bandwagon effects can work. There’s lots of evidence that momentum investing works in stock markets. So you can, on average, make money at least if you are early onto the bandwagon.
And when there is wisdom in crowds, the bandwagon is right. If you’re looking for somewhere to eat in a strange town, it might be a good idea to prefer the crowded restaurant to the empty one.
4. The focussing effect - looking at one piece of data and under-rating others - can be a way of identifying what’s important. This new paper shows that some value investors can beat professional fund managers, because they focus on selected data, rather than look fruitlessly at everything.
5. The status quo bias and the “not invented here” bias can be ways of combating our bounded rationality. They help us recognize that the risks of new schemes cannot always be fully measured, and that things that have stood the test of time have done do because they have hidden advantages.
If banks had displayed more of these biases a few years ago, they’d have been more sceptical about new credit derivatives, and we wouldn’t be in the mess we are in.
And in other cases, biases can even out. The gambler’s fallacy (“we’ve had a run of reds, so black must be due”) can cancel out the hot hand fallacy; “red’s on a roll tonight.” Similarly, the base rate fallacy - ignoring prior probabilities - offsets the conservatism bias, our tendency to underweight new evidence.
It is, therefore, no surprise that biased thinking can lead to good results. Which is precisely why biases are so persistent and ubiquitous.
Here are a few examples.
1. What are the chances of Manyoo beating S***s on Saturday? We could try to remember Manyoo’s defeats and S***s defeats. The latter come to mind more readily than the former, so we infer than S***s have little chance. The bookies agree. But we’ve just used the availability heuristic.
2. Over-confidence, wishful thinking and the self-serving bias can give us the impetus to undertake risky projects. As a result, people start new businesses, write books and go into the music business in more numbers than rationality dictates.
But the result of this is that society is better off; we get new businesses and products, and more books and music than we would in a rational world. And it’s quite possible that the benefits to the successful outweigh the losses of the unsuccessful.
3. Bandwagon effects can work. There’s lots of evidence that momentum investing works in stock markets. So you can, on average, make money at least if you are early onto the bandwagon.
And when there is wisdom in crowds, the bandwagon is right. If you’re looking for somewhere to eat in a strange town, it might be a good idea to prefer the crowded restaurant to the empty one.
4. The focussing effect - looking at one piece of data and under-rating others - can be a way of identifying what’s important. This new paper shows that some value investors can beat professional fund managers, because they focus on selected data, rather than look fruitlessly at everything.
5. The status quo bias and the “not invented here” bias can be ways of combating our bounded rationality. They help us recognize that the risks of new schemes cannot always be fully measured, and that things that have stood the test of time have done do because they have hidden advantages.
If banks had displayed more of these biases a few years ago, they’d have been more sceptical about new credit derivatives, and we wouldn’t be in the mess we are in.
And in other cases, biases can even out. The gambler’s fallacy (“we’ve had a run of reds, so black must be due”) can cancel out the hot hand fallacy; “red’s on a roll tonight.” Similarly, the base rate fallacy - ignoring prior probabilities - offsets the conservatism bias, our tendency to underweight new evidence.
It is, therefore, no surprise that biased thinking can lead to good results. Which is precisely why biases are so persistent and ubiquitous.
It's presumably why, as a young person, one is idealistic and "Leftish". As you get older you discover than, actually, being "Rightish" works better - it's what you find out has worked in the past. It's one of the reasons I worry about Obama - if he really does change things, will it necessarrily be for the better?
Posted by: kinglear | January 21, 2009 at 02:32 PM
“red’s on a role tonight” may be a spelling bias rather than a cognitive bias.
Posted by: The Pedant's Apprentice | January 21, 2009 at 02:51 PM
Kinglear, I think you've fallen into the Pipe and Slippers Fallacy. Age is no predictor of politics: there are too many braying young hoorahs and grizzled old communists in the world for the theory of inevitable rightward drift to hold water case by case.
Posted by: Will | January 21, 2009 at 05:12 PM
Over-confidence, wishful thinking and the self-serving bias can give us the impetus to undertake risky projects. As a result, people start new businesses, write books and go into the music business in more numbers than rationality dictates.
But the result of this is that society is better off; we get new businesses and products, and more books and music than we would in a rational world. And it’s quite possible that the benefits to the successful outweigh the losses of the unsuccessful.
If the benefits outweighed the losses, it would be rational to write the book, wouldn't it?
Posted by: ajay | January 21, 2009 at 05:18 PM
Pedant - ta, correction made.
Ajay - yes, I was a little vague there. I think I was hinting partly at the fact that there are huge externalities to innovation (see Nordhaus), so the irrationality benefits society if not the individual. I was also hinting at the fact that the probability of failure is high, even if the rewards for success are big. Whether this makes the venture rational depends upon attitudes to risk.
Posted by: chris dillow | January 21, 2009 at 05:28 PM
Niall Ferguson is exhibit A. His success wasn't in spite of his spectacular wrongness on important topics; it was because he was wrong in the right way at the right time.
Posted by: Alex | January 21, 2009 at 07:03 PM
I think Ferguson is kind of correct on this point.
We do have all sorts of cognitive biases, and most of them evolved in a much simpler world than the one we currently inhabit. They served us very well as a species, and still do to some extent. But they also hurt us a great deal in certain domains (see banking and complex derivatives trading).
Heuristics are useful for yes/no questions, and when the payoffs are not complex. But they kill us when the answers are complex and the payoffs are complex (again see banking and watch the herd follow VaR models until they all fell off the cliff).
So if we're just counting wins v. losses, then yes, heuristics are good (and help to explain the success of the human species), but when things get more complex heuristics are really bad (really really bad). And the impacts of screwing up the global financial system are much more grave than the positives of using heuristics to figure out who is favored in the next footie match.
Posted by: Russell | January 21, 2009 at 11:05 PM
Biases can produce *fast* answers, which if statistically correct can give an evolutionary advantage.
They don't have to produce correct, meaningful or good answers. And it is a cognitive bias to expect that they should have to do so. ;-)
Posted by: dirigible | January 22, 2009 at 12:31 PM