A woman signs up for a resume writing workshop. A new working paper finds that the adoption of new technologies such as Microsoft Office and Java leads to increases in pay inequalities between more- and less-educated workers.

A new working paper published today by the Washington Center for Equitable Growth explores the relationship between the adoption of new technologies in occupations across the U.S. economy and how technology usage within occupations is contributing to growing income inequality. The four authors—economists Enghin Atalay and Phai Phongthiengtham at the University of Wisconsin-Madison, Sebastian Sotelo at the University of Michigan, and Daniel Tannenbaum at the University of Nebraska—examine how the adoption of these new technologies, among them the Microsoft Office suite of software products, the operating language Unix, and the programming language Java, affect the demand for routine and nonroutine occupational tasks, and thus earnings inequality, between 1960 and 2000.

The four authors build upon their earlier working paper published in 2017, which showed that since the 1960s, occupational tasks in the United States have shifted broadly away from routine tasks and toward nonroutine tasks—a shift that occurred within occupations rather than between them. I wrote about these changes in the nursing occupation, for example, throughout the 20th century here. For that earlier working paper, the authors constructed an open source database of occupational characteristics, including skills requirements, technology use, and work activities, from text analyses of job ads published between 1940 and 2000 in The Boston Globe, The New York Times, and The Wall Street Journal. Now, in the new working paper, they built upon this prior work with data on 48 new technologies to document the relationship between the tasks that workers perform and the technologies they use.

The authors use these data to construct a general equilibrium model that shows the introduction of new technologies shifted workers’ tasks within occupations toward nonroutine analytical tasks. This increase in demand for nonroutine analytical tasks such as problem solving, intuition, creativity, and persuasion translates into greater demand for highly educated workers who are relatively better at conducting these types of tasks.

They find that the adoption of new technologies in turn led to an increase in pay inequality between more- and less-educated workers. Their analysis shows that the introduction of a range of new technologies is responsible for 17 percent of the increase of the difference in earnings between college- and high school-educated workers from 1960 and 2000. The one exception is Microsoft Office, which instead increased demand for nonroutine interactive tasks and thus slightly decreased the overall skills premium and earnings inequality.

The four researchers leveraged their rich database on occupational characteristics such as task content, skill requirements, and technology usage during the post-industrial period to ask important questions about occupational sorting and the advent of new technologies on earnings inequality. This same database could be leveraged by others interested in how labor markets have changed over time. Open questions about shifts in educational requirements among and between occupations or the changes in characteristics of internships, apprenticeships, and other temporary training opportunities could still be explored. There are plenty of analyses that researchers could conduct with these data to examine changes in particular occupations over the course of the 20th century.