Even when automation does not reduce the total number of jobs, it can change who gets paid what. Since about 1980, pay in many rich countries has grown far faster for some workers than for others. Technology is one major explanation, though not the only one. This level looks at the main theories and the arguments over them.

The race between education and technology

One long-standing view is that technology raises the demand for skilled workers faster than it raises the demand for less skilled workers. Economists call this skill-biased technical change. Skill-biased technical change means new technology raises the productivity and pay of skilled workers more than of others. Computers, for example, made educated professionals far more productive, while doing little for, or replacing, many routine workers.

Whether that widens inequality depends on how fast the supply of skilled workers grows. In their 2008 book, Claudia Goldin and Lawrence Katz described a race between education and technology. In most of the twentieth century, American schooling expanded fast enough to keep up. Goldin and Katz argued that the college wage premium rose after 1980 mainly because the supply of graduates grew more slowly while demand for their skills kept rising. On this view, the best remedy for inequality is more and better education.

Labor's share of the pie

A second lens looks at how national income is split between work and ownership. The labor share is the portion of national income paid to workers as wages and benefits, with the rest going to profits, interest and rent. In 2014 Loukas Karabarbounis and Brent Neiman reported that the labor share had fallen in most large economies since the early 1980s, and linked it partly to cheaper computers and equipment, which encouraged firms to substitute capital for labor.

A related claim is that pay has fallen behind productivity. Groups such as the Economic Policy Institute show that output per hour in the United States rose much faster than the pay of typical workers after the 1970s. Critics accept that there is a gap but say it is smaller than it looks. Critics of the productivity-pay gap note that it shrinks when benefits are counted and the same price index is used for both, since health insurance has become a larger part of pay and different price measures rise at different speeds. Even on these measures, much of the gap remains for typical workers.

Superstar firms and employer power

Newer research points to the structure of markets. A 2020 study by David Autor and coauthors found that in many industries, sales have shifted toward a few dominant firms with high profits and few workers relative to their revenue. In the superstar firm theory, labor's share falls as sales shift to dominant firms that use relatively few workers, even if each firm's own labor share does not change much.

Employers may also have the power to hold wages down. In 1933 the economist Joan Robinson named this situation monopsony, from Greek words meaning "one buyer." Monopsony is when employers have enough market power to pay workers less than they produce, because workers have few other places to go. A town with one hospital or one big warehouse is the clearest case, but research suggests milder employer power is common, since changing jobs is costly.

Where it hits

Technology's effects are often local. Daron Acemoglu and Pascual Restrepo studied industrial robots in US labor markets from 1990 to 2007. Acemoglu and Restrepo found that each extra robot per thousand workers lowered local employment and wages, by about 0.2 percentage points of the employment rate and about 0.4 percent of wages. Studies of Germany found smaller job losses, partly because German unions and works councils helped move workers to new roles within firms. Trade shocks, such as rising imports from China after 2000, hit some of the same places.

Worked example: two towns

Imagine two towns, each with a car parts plant that installs welding robots. In the first, the plant lays off a third of its workers. They are mostly older men without college degrees, and the nearest growing jobs are in a city 80 miles away. House prices fall, so selling up to move is hard. In the second, a works council negotiates retraining, and the laid-off workers move to maintenance and quality-control roles. A nearby community college runs a technician program. Apply the theories. Skill-biased change predicts higher pay for those who can program and maintain the robots. The labor share lens asks whether the savings go to owners as profit or are shared. The monopsony lens notes that in the first town, the remaining workers have even fewer employers to choose from, which weakens their bargaining power. The same machine can produce very different outcomes depending on education, institutions and geography.