He got subjects to play an urn-guessing game. They were presented with two urns, one with with mostly red balls and another with mostly white, and were asked to guess which of two urns the experimenter was using, based upon seeing one ball drawn. Subjects were split randomly into two groups, one being better paid than the other.
Both the low-paid and high-paid groups got roughly the same amount of guesses right.However, when they were asked: are you a better or worse than average guesser? things changed. Whereas 85% of the high-paid group said they were better than average, only 60% of the low-paid group did. Butler says:
Inequality undermines the confidence of the disadvantaged and boosts the confidence of the advantaged.
This confidence is not just idle talk. In other experiments, Butler found that more confident subjects were more likely to choose to enter winner-take-all tournaments, thus increasing their chances of big payoffs later.
This implies that initially arbitrary inequality can become persistent, because it reduces the desire of the worst-off to compete.(This is, of course, only one of several routes in which this can happen).
It's easy to see that this might have real-world applications.It's consistent with Afro-Caribbean boys doing poorly at school; with working class kids being loath to apply to Oxbridge; and with women not applying for top jobs. It's also consistent with some of the unemployed not wanting to work: why apply for jobs if you'll only be knocked back?
This implies that a popular explanation for inequality might be no explanation at all. It's sometimes claimed that the successful are successful because they worked hard and the unemployed are unemployed because they are lazy. But this research suggests the causality might be reversed. Maybe successful people work hard because prior good luck has led them to think that work pays, whilst others are lazy because bad luck has reduced their confidence. In other words, it could be that the preferences which rightists believe to be the cause of inequality are in fact the effect of it.
We know from research in personal finance that preferences for risk are partly endogenous; people who have experienced recession (pdf) or the death of a child are more risk averse than others. Wouldn't it be remarkable if risk appetite was the only endogenous preference? If ambition and appetite for work are also endogenous, then a moralistic defence of inequality becomes less tenable.