Hear the moniker “the one percent” in the context of income inequality, and you will most likely envision a senior corporate executive. There’s a reason for this immediate association—the rise in our economy’s top incomes are mirrored by a rise in income inequality within individual firms over the past several decades. And, given the evidence of the significant rise in executive pay across all kinds of industrial and services firms, there’s considerable support for this intra-firm inequality being a major contributor to the rise of overall income inequality.

But a recently-published working paper from the National Bureau of Economic Research challenges the idea that rising inequality within firms has contributed at all to rising overall income inequality in the United States. The new paper, by Jae Song of the Social Security Administration, David J. Price of Stanford University, Fatih Guvenen of the University of Minnesota, and Nicholas Bloom of Stanford University is part of a developing research literature that looks at how pay variation across employers affects income inequality overall. These papers try to separate rising income inequality among individuals into two components: inequality in pay between firms, and inequality in pay within them.

This is complicated economic analysis. Take, for example, a working paper released last September by several economists who looked at the effects of workplaces on income inequality. They find that rising income dispersion between employers was a significant driver of wage inequality. They find this inter-firm inequality responsible for about two-thirds of wage income inequality in the United States, with intra-firm inequality making up the rest. A paper using West German data and a similar methodology found a similarly large role for inter-firm inequality in that country between 1985 and 2009.

Yet in the new paper, Song, Price, Guvenen, and Bloom utilize a different methodology to uncover results that contrast with the research above. Song, Price, Guvenen, and Bloom attribute the entirety of rising income inequality to inter-firm inequality, finding no role for intra-firm inequality, such as rising CEO pay across industries.

For each year, they look at a specific spot in the distribution of individual income–no matter where that income is earned, meaning the individual employer is not taken into account–and then calculate how much the average income has increased at that point in the distribution over the period they study. They look at the median, or 50th percentile, and find that the average income grew by 18 percent from 1982 to 2012 after factoring in inflation. To figure out the role of firms in income growth, they then calculate the growth in the average wages of firms that employ workers at each point in the individual wage distribution. For instance, they calculate the average wage of the firms that employ workers earning median income and call that the 50% percentile of firms. They do this in 1980 and 2012 and calculate how much those firm average wages changed in the interim.

If the growth rate of average wages at firms at a point on the individual distribution of wages across the economy grows faster than the growth of individual incomes at the same point, then the authors interpret this as a strong firm effect, meaning rising inequality between firms is driving all of the growth in income inequality. And that’s what they find for the majority of the income distribution.

But let’s step back and think about this measure of the inter-firm effect that Song, Price, Guvenen, and Bloom employ in their research. Their methodology posits that the effect of a firm on an individual’s income is measured by the average wages at the individual firm. But this measure is a function of all the individuals at the firm. The average wage can be skewed quite a bit by what economists call “tail inequality,” or the often extraordinarily high salaries of those employees at the very top of any firm. In economics speak, there’s a problem of endogeneity here, meaning there’s a feedback loop between a variable and another that’s supposed to be independent of it. It’s hard to clearly point to a rise in inter-firm inequality when this measure clearly can be influenced by a rise in intra-firm inequality, as Song and his co-authors do.

This isn’t to say that inter-firm inequality hasn’t been a significant driver of income inequality. What the most recent research shows is that the methodology used in this latest paper by Song and his co-authors and the end results aren’t as clear as previous research in this field. The likelihood of no increase in intra-firm inequality given all the other data we have on executive pay seems quite small. In other words, while it’s likely that inter-firm inequality has played a large role, intra-firm inequality played a role as well. And the analysis in this most recent paper by Song and his co-authors isn’t’ persuasive enough to counter those other recent results. Then again, this is a preliminary working paper. Improvements can always be made.