A National Bureau of Economic Research working paper by Jimmy Charité, Raymond Fisman and Ilyana Kuziemko tackled this with an online experiment in which a random sampling of Americans were asked what tax rate they thought appropriate for someone whose annual income had suddenly increased by $250,000 for reasons involving luck. The researchers asked the question twice. In one version, the income gain occurred in the current year; in the other, it happened five years ago. Surprisingly, the respondents favored a 1.7 percentage point higher tax rate if the person with the income gain had recently started earning the extra money than if the person had been earning it for five years. That may not sound like much, but it is more than half of the gap the same experiment showed between the tax rate favored by Obama voters and the rate favored by those who said they voted for Mitt Romney in 2012.

In other words, respondents favored less redistribution if they believed that the person had already grown accustomed to a higher income. The psychology seems to be something like this: Rich people who have been rich for a while have gotten used to their money, so it would be unfair to tax them heavily. But people who have just gotten rich have not become accustomed to higher levels of after-tax income, so it wouldn’t be as harmful to raise their taxes in the interest of greater equality.