Around the world, workers have been losing out. In advanced economies, the share of worker compensation as a portion of national income—known as labor’s share—has been falling: Between 1990 and 2009, the median labor’s share in OECD countries declined from 66 percent to 62 percent.

Labor’s share, the measure that illustrates how much of a nation’s GDP is returned to the workers who produce its goods and services, is sometimes used as one indicator of inequality—as Thomas Piketty highlighted in his book Capital in the Twenty First Century (though there’s some objection to using the number for this purpose). Before the 1980s, this number was mostly stable, then it started a downward trend that has lasted for the past few decades (with some recovery since 2015).

A new working paper from the National Bureau of Economic Research by David Autor, David Dorn, Lawrence Katz, Christina Patterson, and John Van Reenen, economists from MIT, the University of Zurich, and Harvard looks at why labor’s share has declined in recent decades. In a previous study, the researchers found that the average company didn’t see a huge decline in the labor’s share from decade-to-decade. So they instead honed in on various industries, looking at what kinds of companies had suffered the largest decline in labor’s share and why that might be.

The researchers looked at data from the U.S. Economic Census between 1982 and 2012 for nearly 700 industries in six major sectors, including manufacturing, retail, wholesale, services, finance, and utilities and transportation. Looking at how much the four largest firms in each industry accounted for in terms of total sales in the industry, they found an upward trend in concentration in all of the six sectors, meaning that it was increasingly common that just a few firms accounted for the bulk of sales. Since the U.S. Economic Census reports payroll, input, and employment, the researchers were able to observe a negative correlation between concentration and labor’s share—meaning that this trend of so-called superstar firms tends to mean workers taking home a smaller share of the pie. Moreover, the more concentrated an industry had become, the larger the decline in labor’s share.

Why have “superstar firms”—a few select firms that gain larger shares of sales in a particular industry (a feature researchers call “winner take most”)—risen in the past decades? Part of the change might have to do with technological changes in the U.S. economy, and part might be related to monopoly theory: The researchers speculate that these superstar firms might be able to spend a smaller percentage of sales on labor, either because they’re more efficient or because they dominate the market such that their revenues go way up, even if labor costs don’t. They also speculate a few reason for why industry concentration may have occurred in the first place: The internet made prices more transparent, resulting in consumers easily being able to identify the seller with the lowest prices or best quality; the U.S. economy produces goods that have high fixed costs, but low marginal costs—such as software—which benefits large firms; and finally, new technology might be strengthening what economists call “network effects,” or the tendency of people to do the things their peers do, including buying the same goods from the same places.

The authors did not mention the decline in anti-trust enforcement from the federal government as a possible cause. Asked over email about this omission, David Autor, one of the co-authors of the paper and an economics professor at MIT, says while this decline may have played some role, it’s not where the weight of evidence falls, as the greatest rises in concentration are in sectors that saw intensive research and development and patents, and those seeing productivity growth. “This makes us suspect that the fall in the labor share is not driven by fat-cat sectors figuring out how to better collude to raise prices. We cannot rule [out] that this contributes, but it’s not where the evidence seems to point primarily,” wrote Autor.

As for what this means for American workers,  Autor added that falling labor share paired with slow GDP growth is likely to exacerbate inequality, since households tend to “own” labor but not the capital (in other words, stock in companies) that’s becoming more profitable. “If capital ownership were distributed comparably to labor/skills ownership, then the fall in labor’s share would not be dis-equalizing: The rise in capital’s share would have a precisely offsetting effect. But capital is distributed much more unequally than labor,” explained Autor.

To be certain, industry concentration isn’t the sole reason why labor’s share has gone down, but it’s likely part of the story. And if industries in the U.S. become increasingly dominated by just a few firms, the effect on workers might be that their pay becomes an even smaller piece of the pie of overall profit. Recent gains in labor’s share have been small but positive, and economists can only hope that the tightening of the U.S. labor market will continue to improve matters.