Labor Studies.

• Author: Autor, David
• Position: Program Report

The Labor Studies Program is one of the largest and most active in the NBER. Its nearly 190 members produce more than 300 working papers in an average year. The breadth and depth of questions addressed by Labor Studies members is immense. Research touches on macroeconomic topics such as unemployment and productivity; institutional factors such as minimum wage regulations, labor unions, and globalization; and technological developments including robotics, artificial intelligence, and algorithmic decision-making. It also includes core human capital subjects such as educational investment, the demand and supply of skills, and wage determination; industrial organization topics such as imperfect competition, rent sharing, and firm-specific wage policies; and social insurance and welfare programs such as unemployment insurance, universal basic income, and in-kind benefit programs such as SNAP, Medicaid, and housing assistance. Program affiliates also study urgent social questions, including race and gender disparities in market opportunities, neighborhood quality, treatment by the criminal justice system, and many other subject domains.

Reflecting their intellectual diversity, two-thirds of Labor Studies Program members are affiliated with two or more NBER programs or major projects. Though the pandemic has curtailed some program activities, it has simultaneously opened new horizons. The online meeting environment has allowed many nonaffiliated scholars to participate in program meetings. Meanwhile, researchers who prefer to audit rather than participate in program sessions can watch meetings streamed live on NBER's YouTube channel. In the post-pandemic world, the program will strive to keep these professional and intellectual doors open.

This brief report summarizes a small subset of topics where research by Labor Studies affiliates is burgeoning, including the role of firms in wage determination; the minimum wage; the consequences of advancing technologies for employment and productivity; race and ethnicity in the labor market; and the extent and consequences of racial and ethnic discrimination and segregation. This summary does not do justice to the vast body of recent scholarship by program affiliates, though our hope is that it reveals some important research undercurrents.

Automation, Employment, and Productivity

The role of automation in shaping labor demand, skill requirements, and wage levels has been of intense economic interest for centuries. Even so, this topic has gained further prominence as rapid advances in ubiquitous computing, artificial intelligence, and robotics have imbued machines with the ability to accomplish tasks that require learning, judgment, and dexterity. Labor Studies scholars have taken numerous angles of attack to assess what this has meant for labor markets and to forecast what may lie ahead.

One influential paper in this domain by Daron Acemoglu and Pascual Restrepo explores how the expansion of industrial robotics has affected employment and wages in local labor markets--so-called commuting zones. (1) Harnessing data on industrial robot penetration in other industrialized countries to measure the technological frontier, the researchers calculate predicted robot adoption in the United States within local labor markets based on initial industry structures in those locations. A key finding is that local labor markets with greater exposure to robot adoption saw differential falls in employment-to-population rates (and wages, not pictured) in the 1990s and early 2000s. An independent empirical contribution by George Borjas and Richard Freeman reaches a similar conclusion. (2)

Brad Hershbein and Lisa B. Kahn explore how recessions may accelerate the process of technological change by studying the evolution of skill requirements posted in job vacancies, using a vast database of vacancy postings scraped from the web by Burning Glass Technologies. (3) They show that skill requirements in job vacancy postings differentially increased in metropolitan statistical areas that were hit hardest by the Great Recession, and these increases persisted through at least the end of 2015, long after the recession was over. They interpret this evidence as consistent with adjustment cost models in which adverse shocks accelerate the process of adaptation to new business processes, in this case, so-called routine-task-replacing technologies and the more-skilled workers who complement them. Consonant with these findings, Alex W. Chernoff and Casey Warman argue that the current COVID-19 pandemic may speed the process of automation. They further present evidence that in a large set of countries, the occupations held disproportionately by women are at greater risk of displacement by automation, implying that the post-pandemic labor market may offer fewer of the positions frequently held by women. (4)

Illuminating another facet of the interplay among technological change, demand shifts, and labor market adjustment, Elizabeth U. Cascio and Ayushi Narayan study the impact of the introduction of hydraulic fracturing (fracking) for oil extraction, a technology introduced during the 2000s, on educational investments. (5) Because fracking offers high-paying blue-collar jobs to workers without secondary credentials, it potentially raises the opportunity cost of schooling. As theory would predict--and as many parents would lament--high school dropout rates rose among male teenagers living near shale oil deposits.

What are the long-run implications of advancing automation for skill demands ? A theoretical paper by Seth G. Benzell, Laurence J. Kotlikoff, Guillermo LaGarda, and Jeffrey D. Sachs considers how, in an overlapping generation setting, automation can ultimately lead to worker immiseration by reducing capital formation as long-lived, barely depreciating software capital effectively makes high-skill workers redundant. (6) In related work, Anton Korinek and Joseph E. Stiglitz consider the challenges that artificial intelligence may ultimately pose for income distribution and unemployment. (7) David E. Bloom, Mathew McKenna, and Klaus Prettner place this issue in global perspective by observing that the global labor market will need to absorb roughly three-quarters of a billion new workers between 2010 and 2030. (8) With 91 percent of that growth occurring in low- and lower-middle-income countries, they raise the concern that technological advances may create headwinds because the labor-intensive jobs currently prevalent in developing countries may be increasingly subject to automation.

While most of the papers above focus on the economic implications of machines substituting for labor, work by David Deming presents evidence that as automation proceeds, the demand for human capabilities is rising on another margin: social and managerial skills. (9) Deming argues that as information technology has replaced workers in routine codifiable tasks, it has magnified the value of social skills that allow workers to specialize and collaborate more efficiently. In a related vein, Gaetano Basso, Giovanni Peri, and Ahmed Rahman provide evidence that low-education US immigrants have helped blunt the impact of automation on native US workers. (10)

In work that appears prescient in light of the current pandemic, Nicholas Bloom, James Liang, John Roberts, and Zhichun Jenny Ying examine another labor market manifestation of advancing information technology: remote work. (11) Partnering with a large Chinese travel agency, the researchers conduct a large field experiment in which travel agents were randomly offered the option to work from home. Among those offered the work-from-home option, both productivity and worker satisfaction rose. Ironically, promotion rates conditional on performance fell among those working from home, suggesting that not being in the office may also have hidden private costs.

This growing body of theory and evidence on labor market consequences of automation highlights an enduring...