The share of national income going to U.S. workers has been on the decline since 2000 despite long-held economic expectations to the contrary. The reason for this decline has been the source of competing research and fevered debate over the past few years. Purported reasons for the decline range from offshoring and globalization, declining worker bargaining power, and, most famously put forth by Thomas Piketty in his book “Capital in the 21st Century, the idea of capital continuously accruing to the wealthy.

Another theory is that technology may also be key to the decline, not because of the speed of technological change, but rather because of its bias toward one factor: labor. The National Bureau of Economic Research last month published a working paper by economist Robert Z. Lawrence, a professor at the Harvard Kennedy School, who places technological change at the heart of the decline in labor’s share of U.S. income. His findings, however, contradict the idea that “robots” are stealing jobs, a familiar explanation.  His thesis rests on a surprising conclusion—that productivity growth in the U.S. economy has been more tilted toward workers in recent years, even though those workers are not enjoying the fruits of that growth in the form of higher wages.

But before we dive into Lawrence’s paper, let’s step back to remind ourselves of two important factors in the declining share of income going to labor in the United States. The first factor is what economists call the marginal elasticity of substitution between capital and labor. Essentially, this tells us whether a business would use more or less labor if the price of capital declines (or vice versa). An elasticity below one means the two factors are complements, and that a decline in the price of capital would increase demand for labor. But if the elasticity were above one, then more capital would be used, thus lowering demand for labor. Put another way, an elasticity below one means that capital and labor are complements, but if the elasticity is above one then they are substitutes.

And then there is the capital-output ratio in the economy. This is the ratio of the stock of capital previously invested in the economy to the total amount of goods and services produced in the economy in one year. Looking at the movements in the capital-output ratio alongside the marginal elasticity of substitution between capital and labor can tell us about the reasons for the decline in the labor share of income

One way for that share to decline is for the capital-output ratio to rise and the elasticity to be above one. This would be a sign of firms increasingly using capital as the return to capital doesn’t decline. Those with capital would increasingly gain (at the expense of labor) by investing their capital. This interpretation is at the heart of Piketty’s “Capital in the Twenty-First Century.”

The other way would be for the elasticity to be below one and the capital-output ratio to be on the decline. This combination would signal an increasing rate of return to capital, but firms would also be hiring labor as the two are complements. But the return to labor wouldn’t be as high as the return to capital, resulting in income shifting more toward capital. This would be the explanation for a declining labor share of income under a traditional neoclassical model of the economy.

According to Lawrence, the data show an elasticity below one, an increasing capital-output ratio, and a declining share of income going to labor (of course, there is still some debate about this). How does Lawrence explain a declining labor share? Lawrence argues that technological growth has increasingly become biased toward labor. This means that the effective capital-output ratio is actually on the decline and is consistent with a declining labor share.

Lawrence’s analysis contends that technology has increasingly gone toward boosting the productivity of labor rather than capital, resulting in a decline in the effective capital-output ratio. This means the return to capital would go up, as would the demand for labor. Increased productivity of labor is like increasing its supply in that a more productive worker can create more of a good in the same amount of time.

Lawrence also argues that low elasticity of substitution between capital and labor is analogous to the demand for labor being inelastic, meaning that the demand curve for labor is downward sloping and very steep. Movement down that curve—via a shift in the labor-supply curve—results in a decrease in wages that is larger than the increase in employment. So while this isn’t exactly what’s going on, the intuition leads to the same end result: a decline in the share of income going to labor.

Lawrence’s paper is similar to one by economists Ezra Oberfeld at Princeton University and Devesh Raval at the Federal Trade Commission. They also find a declining share of national income going to labor in the presence of an elasticity of substitution explained by technological growth that favors labor. Understanding why U.S. workers are reaping less of the national share of income, then, requires economists and policymakers alike to examine how labor-enhancing technologies are perhaps just as important as globalization and offshoring, the decline in unionization, and the triumph of capital over labor.