Cracking the Mystery of Labor's Falling Share of GDP
Economists are very worried about the decline in labor’s share of U.S. national income. One reason they’re concerned is because when less of an economy’s wealth flows to workers, it exacerbates inequality and increases the risk of social instability. But another reason is that this trend throws a wrench in economists’ models. For decades, macroeconomic models assumed that labor and capital took home roughly constant portions of output -- labor got just a bit less than two-thirds of the pie, capital slightly more than one-third. Nowadays it’s more like 60-40.
Economists are therefore scrambling to explain the change. There are, by my count, now four main potential explanations for the mysterious slide in labor's share. These are: 1) China, 2) robots, 3) monopolies and 4) landlords.
The China hypothesis basically says that the opening of the Chinese and Indian economies, combined with the invention of globalizing technologies like the internet and containerized shipping, dumped a flood of low-cost labor onto the world market, allowing multinationals to shop around for cheap workers while raising their profits. This view received some support from a 2013 paper by Michael Elsby, Bart Hobijn and Asegul Sahin, who found that the trade and manufacturing sectors had the biggest declines for labor income.
One problem with this theory is that, according to Chinese statistics, labor’s share has been falling there, too. From China’s perspective, opening the country to trade brought in a flood of new capital, so capital’s income share should have been the one to fall. This chips away at the globalization explanation, assuming those Chinese government numbers can be trusted.
The robots hypothesis says that as technology gets cheaper, employers are substituting machines for workers. A 2013 paper by Lukas Karabarbounis and Brent Neiman found that costs of capital goods have been getting cheaper, and concluded that companies are substituting technology for human labor. This fits with other research showing adverse effects on wages from the adoption of new technologies like industrial robots.
But there are problems with this thesis as well. A recent study by David Autor, David Dorn, Lawrence Katz, Christina Patterson and John Van Reenen found that the labor share is falling across the whole economy, but not within companies. In other words, companies themselves aren’t substituting machines for workers, as we might expect them to do if robots were getting really cheap. Instead, the economy is simply shifting resources toward a few large companies that are very capital-intensive, and away from the more numerous, smaller companies that use more human labor. Autor et al. blame increasing monopoly power for labor’s decline.
Then there’s the idea that landowners, not corporate overlords, are taking money away from workers. While analyzing the work of French economist Thomas Piketty, Matt Rognlie found that national income accounts showed an increasing amount flowing to owners of land. More recently, economist Dietrich Vollrath examined a paper by Simcha Barkai about rising profits, and found that profits from owner-occupied housing also rose sharply.
Supporters of the other theses have yet to really grapple with the landlords explanation. The reason is that the people pushing this fourth idea justify it based on national income accounts, while supporters of the other three explanations tend to look at corporate behavior up close. When economists speak in different languages, it’s harder to have a debate.
So that leaves us with as many as four competing explanations, each with some reasonably compelling circumstantial evidence in its favor. What to do? Eventually, economists will probably find new, better ways of putting these theories head to head. But in the meantime, it’s worth asking whether some of these explanations could actually be measuring different parts of the same phenomenon.
A recent blog post by Paul Krugman offers a possible insight. Krugman notes that it’s possible that some companies are more capital-intensive and some are more labor-intensive -- think of factories making televisions with robots while others assemble them by hand. When the productivity of the capital-intensive companies improves -- due to mechanization, or the internet, or globalization -- it shifts production toward those companies, and lowers wages in the process.
Now suppose that those capital-intensive companies are a small handful of superstar multinationals, while the labor-intensive companies are a bunch of small, local competitors. Improvement in robots, information technology and globalization would therefore be shifting resources away from the many and toward the few -- in other words, exactly the same phenomenon that Autor et al. describe. Huge companies are probably more capable of building automated factories, using online supply chains to outsource production to China.
So monopoly power, robots and globalization might all be part of one unified phenomenon -- new technologies that disproportionately help big, capital-intensive multinational companies. Meanwhile, technology that augments human labor-power -- for example, cheap energy -- might have languished in recent decades, due to the failure to replace oil and gas with better power sources. Hence, small companies that use lots of workers might be losing out in the age of information technology.
That theory still doesn’t explain how landlords might fit into the picture. But it provides a possible way to unify at least some of the competing explanations for this disturbing economic trend.
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
To contact the author of this story:
Noah Smith at firstname.lastname@example.org
To contact the editor responsible for this story:
James Greiff at email@example.com