There is a widespread belief that the U.S. labor market has been going through a period of unprecedented change over the past decade or two. On the one hand, David Deming, Christopher Ong, and Lawrence H. Summers questioned this historical claim in their essay “Technological Disruption in the American Labor Market.” That is, they argue that the historical changes in American occupations have been much larger. During various periods of the late 19th and 20th centuries, more recent changes. But they also point to some signs that major changes in the U.S. labor market may be underway, although they have not yet occurred. This essay appears as a chapter in a collection published by the Aspen Economic Strategy Group. Strengthening America’s Economic DynamismEdited by Melissa Kearney and Luke Pardue. You can download individual chapters or the entire book.
For a historical perspective on changes in the U.S. labor market, consider the figure below. Looking back to the late 19th century, about 80% of all jobs in the United States were in agriculture or blue-collar production. But by the 1960s, agricultural jobs accounted for less than 10 percent of all U.S. jobs and were declining rapidly. From 1950 to 2000, the share of blue-collar jobs declines from 40% to 20% of all jobs. On the other hand, professional, clerical, administrative, and other ‘service’ occupations have increased rapidly.
In short, parts of the dynamic U.S. economy have always involved dramatic changes in jobs. Jobs in the United States have evolved dramatically over several decades. Currently, the proportion of clerical and administrative positions has been decreasing since 1980, and retail sales positions are also showing a decreasing trend. However, changes in American occupational categories over the past several decades are not clearly visible in the historical record.
But is the U.S. economy now in a position where the emergence of new artificial intelligence technologies could really lead to dramatic changes in the U.S. job market? maybe! At this stage, it is difficult to know how new technologies will be used and what impact they will have on jobs. History teaches us that when it comes to jobs, new technologies often give and take away. Deming, Ong and Summers point to four shifts currently underway.
Trend 1: Class polarization has been replaced by generic skill upgrades.
[E]Employment growth between 2000 and 2010 was polarized, with large increases at the bottom and top of the wage distribution and declines in the middle. Polarization continued from 2010 to 2016, but to a much lower degree than in previous decades. However, since 2016, the polarization of the labor market has stopped. From 2016 to 2022, both low-skill and middle-skill jobs decreased by about 2 percentage points, while employment in high-skill occupations increased by just over 4 percentage points. Therefore, employment growth since 2016 has been more of a pattern associated with technological improvements rather than a pattern indicative of polarization. …
Trend 2: Employment in low-wage service occupations is stagnant or declining..
The main explanation for employment polarization in the first decade of the 2000s was the rapid growth of service sector jobs. This replaced middle-skilled (often unionized) blue-collar jobs and offered lower wages and fewer employment protections. [T]Growth in service jobs stalled in the early 2010s and remained stagnant for much of the remainder of the decade. Employment in food service and personal care occupations declined sharply in 2020 due to the COVID-19 pandemic, with only a partial recovery by 2024. With the exception of health support occupations, employment in the service sector is now similar to what it was 20 years ago. A few years ago, early 2000s early decade. Service occupations have given back almost all of the rapid job growth experienced over the past decade. …
Trend 3: Rapid Employment Growth in STEM Occupations
Between 2000 and 2012, science, technology, engineering, and math (STEM) jobs declined as a share of U.S. employment, while employment in non-STEM professional occupations grew rapidly (Deming 2017). Deming (2017) documents this surprising fact and argues that the demand for social skills is increasing because complex tasks require social interaction and are not easily automated with current technology. … [W]While social skills-intensive jobs in management, business, education, and healthcare continue to grow, STEM employment is now growing rapidly after declining during the first decade of the 2000s. The share of overall employment in STEM fields increased from 6.5% in 2010 to nearly 10% in 2024. About 60% of this growth is concentrated in computer occupations such as software developers and programmers, but employment has also increased in a wide range of science and engineering fields. The same goes for jobs.
STEM employment growth has accelerated particularly rapidly over the past five years. Additionally, rapid employment growth in business and management positions is concentrated in occupations such as scientific and engineering managers, management analysts, and other business operations professionals. The increase in employment in STEM occupations is also consistent with increased capital investment in AI-related technologies. …
Trend 4: Employment decline in retail sales
Retail sales jobs remained stable at about 7.5% of U.S. employment from 2003 to 2013. From 2013 to 2023, the U.S. economy added more than 19 million jobs. However, during the same period, retail sales fell by 850,000 jobs, causing their share of employment to fall from 7.5% to 5.7%, a 25% decline in just 10 years. The decline in retail sales jobs began before the pandemic but has accelerated over the past year. …
Four trends – an end to polarization, stagnant growth in low-wage service jobs, rapid growth in employment in STEM occupations, and declines in retail sales employment – suggest that the pace of labor market disruption has accelerated in recent years.
The patterns of U.S. economic activity and job growth have been shocked by the pandemic, and it is unclear (at least to me) how these changes will play out. Will working from home, shopping from home, and online healthcare continue to rise, or will they retreat over time? Perhaps new disruptions in the U.S. labor market will not occur across the broad occupational categories used in the historical analysis above, but will instead involve changes in the skills required of workers in current occupational categories.
