Did computers and schooling push wages apart?

Many economists think they did, at least for a while. The leading idea is skill-biased technological change: a shift in the production technology that favors skilled over unskilled labor by increasing its relative productivity and, therefore, its relative demand. In plain words, if computers make skilled workers more productive and employers want more of them, their pay rises faster than everyone else's. One colloquium of prominent labor economists estimated that technology was responsible for over 40 percent of the rise in US inequality, with trade, the minimum wage, unions and immigration each responsible for 10 to 15 percent.

The measure to watch is the college premium, the extra pay of workers with four-year degrees. It doubled from 1980 to 2005, as demand for college-educated workers exceeded supply. Goldin and Katz estimated that rising returns to education explained about 60 percent of the rise in wage inequality between 1973 and 2005. It is a race, they argue: for eight decades the number of educated workers grew faster than demand for them, lowering inequality, and since about 1980 the reverse has been true. The schooling numbers show it: the average American born in 1945 got two more years of schooling than their parents, while one born in 1975 got only half a year more.

But the story has holes. The timing of the technology does not match the timing of inequality, which began rising in the early 1970s, well before business used the internet in the late 1990s. Education also fails to explain the gap between the top 1 percent and everyone else. And Japan, Sweden and France did not see significant increases in income inequality from 1979 to 2010, although the US did. Emmanuel Saez wrote that differences across countries rule out technical change and globalization as the sole explanation.