For years now, the exciting field of inequality studies has been puzzled by a seeming inconsistency in the data. Income inequality is clearly going up. But consumption inequality — that is to say, the difference between how much rich people and poor people spend — isn’t.

A worker walks past a Gucci Group NV advertisement in Beijing, China, on Wednesday, March 23, 2011. Beijing's Communist government, faced with a widening gap between rich and poor, moved to make the issue less visible by banning outdoor advertising in the Chinese capital that promotes lavish lifestyles. (Nelson Ching/Bloomberg)

Some said this meant income inequality wasn’t really going up, or if it was going up, it didn’t really matter. “We eat bread, not paychecks,” wrote Will Wilkinson. Others argued that poorer Americans were going into debt to sustain their consumption, or that the fact that richer Americans weren’t spending much more was irrelevant to the fact that they were making much more. But a new paper by Orazio Attanasio, Erik Hurst and Luigi Pistaferri says we’ve got it all wrong: The data we were using is bad, and consumption inequality is going up alongside income inequality.

The problem, says Erik Hurst, an economist at University of Chicago’s Booth School of Business, is that the main survey we’ve been using to track people’s spending habits — the Consumer Expenditure Survey — is breaking down.

Over the last 30 years, the CES has been giving us increasingly implausible readings. For one thing, it’s stopped coming anywhere close to matching our national accounts data, which shows how much we produce. For another, it says that average consumption was 15 percent higher in 1980 than in 2007, which flies in the face of everything we know about income and economic growth since then. And, perhaps worst of all, the CES seems to be delivering particularly flawed readings on the spending of the very rich, which is exactly the group you most want to capture when you’re look at consumption inequality.

So Hurst and his coauthors set about trying to find a measure of consumption that wasn’t so riddled with errors. They looked at areas of the CES that seemed to match our national production statistics better, like food and cars. They used the diary side of the consumption survey rather than the interview data. They set up comparisons between particular subgroups to see how they’d tracked each other over time. They used the Panel Study of Income Dynamics.

Their conclusion? “Across every other measure of consumption we analyzed, consumption inequality increased substantially.” In fact, “not only do these other measures of consumption inequality mirror the overall change in income inequality, the timing of the changes also line up very closely.”

But there has been a moderating force, the researchers say. “Leisure inequality” has gone up, too. Poorer Americans are working fewer hours than richer Americans. About half of this, Hurst says, can be explained by poorer Americans having trouble finding enough work. But the other half can’t be.

“In order to make overall welfare calculations,” the authors write, “one needs to take a stance on how the leisure time is valued. But, as long as leisure has some positive value, the increase in consumption inequality between high and low educated households during the past few decades will overstate the true inequality in well being between these groups.”

That is to say, if you believe that one thing money buys is time, the non-rich are “consuming” more leisure time than the rich, and that should also factor into our thinking on consumption inequality.