Elise Payzan-LeNestour at the University of New South Wales ran an experiment in which subjects could choose to bet on whether an archer would hit a target, winning $2 if he hit but losing $40 if he missed. Some archers were skilled and almost never missed but others were unskilled and did miss: betting on skilled archers was analogous to buying government bonds, but betting on the unskilled ones was more like buying a junk bond with high tail risk and hence a negative expected value.
She found that two-fifths of subjects bet on the unskilled archer even though doing so had a negative expected payoff. This was not because they failed to distinguish between good and bad archers: Ms Payzan-LeNestour shows that subjects inferred the archers' skill from their performance in the way that wholly rational people would.
Instead, some people continued to bet even after knowing the archer was unskilled. This, she says, is because a series of small wins causes some people to keep betting even though they know they shouldn't - perhaps because increased dopamine levels promote risky behaviour. She calls this the "picking up pennies" bias. Because there are genetic differences in dopamine regulation, some people are more prone to this than others.
It is a cliché to question the external validity of laboratory experiments such as these - to doubt whether they apply to the real world. However, we know that people have lost fortunes in exactly the way Ms Payzan-LeNestour describes by trying to pick up pennies in the face of tail risk: they did so by betting on mortgage derivatives in the mid-2000s and, before then, during the junk bond bubble of the late 1980s. Equally, we know that it can be profitable to bet against this behaviour. Just as momentum investors make money by betting against investors' bias to underreaction, so John Paulson's "greatest trade ever" - buying insurance against mortgage defaults in 2007 - was a bet against the picking up pennies bias.
This bias might be especially pervasive because tail risk - by definition - rarely materialises. You can therefore make money for years by holding assets with a negative expected value. For example, a bond that has a 98 per cent chance of making you £1 per year more than cash and a 2 per cent chance of losing you £50 relative to cash has a four-fifths chance of making money even over 10 years - even though it has a negative expected value.
And herein lies the problem. It's difficult to believe that the picking up pennies bias is confined to the professionals. Retail investors might also be prone to it. Perhaps years of decent income from buy-to-letting has encouraged some into the market despite the risks involved - of rising mortgage rates, voids, bad tenants, cuts in housing benefit, or falling house prices.