This article is published in collaboration with the Washington Center for Equitable Growth.

Recent research about the decline of the labor share has caused economists to reconsider the continued relevance of the “stylized fact” that the labor share is constant. Labor and capital might have split up income at a constant rate for a while, but that’s just not true now.

But what if the fact were never true? A new paper presented at this year’s Allied Social Science Associations meeting claims that the apparent stability in the share of income going to labor never happened. According to the paper, the decline in the U.S. labor share didn’t start in the 1980s or 2000s—it started right after the Second World War.

Why has the labor share declined? The rise of intellectual property, it would seem.

The paper—by Dongya Koh of the University of Arkansas, Raul Santaeulalia-Llopis of the Washington University in St. Louis, and Yu Zheng of the City University of Hong Kong—takes advantage of newly updated GDP data from the U.S. Bureau of Economic Analysis. While the Bureau is constantly releasing new data on economic growth, it also revises previous data. Sometimes those revisions show an increase in total U.S. economic output, and sometimes the revisions show a change in the composition of that output. It’s the latter kind of revision that’s important in this case.

In 2013, the Bureau of Economic Analysis updated its treatment of a variety of issues, including how it treats research and development spending. The BEA previously treated R&D spending as a business expense but, as the BEA realized, it makes more sense to think of spending that could potentially boost a firm’s output as a capital investment. As the authors of the paper show, counting investments in intellectual property as, well, investment significantly increases the amount of investment showing up in the data. According to their calculations, intellectual property products have increased from 8 percent of U.S. investment in 1947 to 26 percent in 2013.

Accounting for this kind of capital investment means that the decline in the U.S. labor share starts much earlier than previously thought. According to the paper, the decline starts in 1947, which would mean the labor share was declining throughout the period it was famously stated to be constant. But not only does the decline start earlier than previously thought—it’s also much larger. It’s actually twice as large. And the increase in intellectual property products explains the entirety of the decline.

Furthermore, the paper’s authors also looked into the elasticity of substitution between capital and labor, a key parameter for understanding the decline of the labor share. The three economists calculate an elasticity greater than one, meaning capital (including intellectual property products) is a substitute for labor. Several other studies that look at the question at a more macro level have also found an elasticity greater than one, while micro-level studies tend to find elasticities below one.

While the overall trend since 1947 is a decline in the U.S. labor share, the data in the paper does show a leveling-off of the share from 1980 to 2000 and then a decline from 2000 onward. The overall trend from 1947 may be downward, but the trends within that trend are also important. The debate about the labor share is still very much alive, as this new contribution shows, and it continues to be a conversation to listen in on.

Publication does not imply endorsement of views by the World Economic Forum.

To keep up with the Agenda subscribe to our weekly newsletter.

Author: Nick Bunker is a Policy Analyst with the Washington Center for Equitable Growth. Prior to joining the Washington Center for Equitable Growth.

Image: Workers on the assembly line replace the back covers of 32-inch television sets at Element Electronics in Winnsboro, South Carolina. REUTERS/Chris Keane.