Growth and crisis

America's major cities: creating job inequalities

PIcture of San Francisco by Mattia Bericchia on Unsplash

PIcture of San Francisco by Mattia Bericchia on Unsplash

Large American cities are experiencing labor market polarization, characterized by a simultaneous increase in the employment share of both high-skilled and low-skilled jobs. Economists Fabio Cerina, Elisa Dienesch, Alessio Moro, and Michelle Rendal propose that this phenomenon can be attributed to technology shocks that enhance the productivity of highly skilled workers. As these skilled workers increase their participation in the labor market, they also intensify their consumption of personal services, thus generating greater demand for low-skilled jobs.

By Elisa Dienesch

Elisa Dienesch

Auteur scientifique, Aix-Marseille Université, Sciences Po Aix, AMSE

Timothée Vinchon

Timothée Vinchon

Journaliste scientifique

It's 9pm, you leave your office in New York's business district, ask the babysitter to pick up an Amazon package before you arrive, then order a bo-bun to be delivered right on time for your return home. Although fictional, this story describes an increasingly common scenario in major urban metropolises that economists Fabio Cerina, Elisa Dienesch, Alessio Moro and Michelle Rendall call "spatial polarization". In large American cities where technological advances are more rapid than anywhere else,, there is an increasing number highly-skilled workers, but also a considerable increase in jobs requiring no qualifications carried out by low-skilled workers.

Technical progress, a generator of inequality?

The polarization of jobs reflects two phenomena: the decline of the "middle classe" and the rise of extreme inequalities1 . On the labor market, we observe a distribution of skills which represents the distribution of jobs by qualification levels, ranging from the least qualified to the most qualified. The market becomes polarized when there is a simultaneous increase in the proportion of highly-skilled workers (such as managers, engineers, intellectual professions) and the proportion of low-skilled workers (like employees in personal services), at the expenseof medium-skilled jobs (such that technicians, intermediate or administrative professions)2 .

  • 1Acemoglu Daron, Autor David, Skills, Tasks and Technologies : Implications for Employment and Earnings. In Handbook of Labor Economics (Vol. 4, p. 1043 — 1171). Elsevier, 2011.
  • 2Goos M., Manning A., Salomons A., 2014, « Explaining Job Polarization : Routine-Biased Technological Change and Offshoring », American Economic Review, 104(8), 2509‑2526.
Bicycle delivery in the rain in New York City

Picture by Clay Banks on Unsplash

The question arises : what are the sources of this phenomenon? Initially, economic research focused on the decline of medium-skilled jobs, mainly in the industrial sector, often associated with routine tasks. The initial explanation centered around the automation of tasks, where capital replaced routine work3 .

However, more recent studies delve into the polarization of the labor market, aiming to s explain the increase in the number of highly-skilled and low-skilled workers. The main hypothesis formulated suggests that technical progress has played a central role since the 1980s, especially technologies accessible primarly to highly-skilled workers. Highly-skilled workers, who are more productive, and thus receive higher wages, are encouraged to work more, reducing their time spent at home, on domestic tasks. As a result, the increase in hours worked by highly-skilled individuals generate a growing demand for personal services performed by low-skilled workers (childcare, home management, fast food, grocery delivery, dry-cleaning...), generating polarization at the extremes of our skills distribution. These are known as consumption externalities. Researchers have demonstrated that this mechanism is closely linked to the integration of women into the labour market, starting in the 1980s in the United States4 .

  • 3Autor D. H., Dorn D., 2013, « The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market », American Economic Review, 103(5), 1553-1597.
  • 4Cerina F., Moro A., Rendall M., 2021, « The role of gender in employment polarization », International Economic Review, vol. 62(4), pp. 1655–91.

The spatial dimension of labor market polarization

Why should big cities be more suceptible to this phenomenon? Two compelling reasons can be elucidated. Firstly, research indicates that these technology shocks, which favor productivity gains for highly skilled individuals, have been more pronounced in large metropolitan areas5 . Secondly, these consumption externalities inherently exhibit a local nature (as we typically consume services within our residential areas). We often refer to these services as the "non-tradable" sector, to illustrate the highly localized nature of their consumption. Consequently, researchers formulated the hypothesis that labor market polarization manifest more prominently in very large cities, and thus take on a spatial dimension that had not previously been studied in the literature.

To study this phenomenon, they classified U.S. cities based on national censuses from 1960, 1980 and 2008. They identified 218 metropolitan areas, home to 63% of the national population in 1980 and 71% in 2008. These 218 areas were further categorized according to their size in 1980, in order to define groups of cities according to their size, using different statistical thresholds (medians, terciles, quartiles). This method allowed them to observe that large cities are experiencing more pronounced polarization, with the proportions of of high- and low-skilled jobs expanding at a faster rate, especially as the definition of a large city becomes more extreme".

  • 5Baum-Snow N., Freedman M., Pavan, R., 2018, « Why has urban inequality increased? », American Economic Journal: Applied Economics, vol. 10 (4), pp. 1–42

A model to identify and quantify the role of technical progress

The authors have developed a theoretical model aimed at analyzing the spatial dimension of labor market polarization. The model is designed to study the location choices of workers whether in small or large cities. It is based on a utility function that considers several factors, includingwages based on skill levels, the prices of consumed goods (including domestic services) and the housing price.

This model serves as a toolto isolate and quantify the impact of different technological changes on the spatial distribution of workers, thereby shedding light on the spatial polarization of jobs. One significant aspect of the analysis involves neutralizing the impact of technical progress on highly-skilled workers, assuming that it remains the same in both small and large cities. The findings reveal that this technological shock contributes to the greater polarization observed in larger cities compared to smaller ones. 

Additionally, the model enablesthe authors to explore the effects of two policy instruments that could be implemented to improve the living conditions of low-skilled (and therefore low-paid) workers in large cities. The results indicates that transfers can indeed improve the well-being of these workers, but the outcome depends on how the policy is funded. It's important to note that if such policies lead to increased prices for labor market services, it may neutralize potential opportunities for low-skilled workers due to reduced demand from more highly-skilled workers."


In 2013, in his book The New Geography of Jobs, economist and geographer Enrico Moretti revealed an important insight : for every one technological job created, it generates five other less-skilled jobs. Every time a software developer is hired in Silicon Valley, five additional are created concurrently: waiters, home helpers or cab drivers. The phenomenon of employment polarization in major cities stands out due to its magnitude and unique dynamics compared to more peripheral areas.

However, the most precarious jobs are often those most affected by the inequalities created by polarization. In a study carried out by the Urban Institute, a team of geographers and urban planners also noted a spatial mismatch between job supply and demand. In the San Francisco Bay Area, only 29% of job seekers find a job within 10 kilometer radius of their residence, leaving the òajority with longer commutes. In a region where the cost of living is 80% higher than the national average, housing is often unaffordable for workers with lower skills. Despite high demand for these jobs, effective solutions to bridge the gap have been limited thus far, hindering cohabitation. Some cities are actively addressing the issue, by creating more housing near jobs centers or improving transportion networks.

Multi-lane bridge between San Francisco and Oakland

Picture by Danielson on Unsplash

What about tomorrow?

The emergence of new artificial intelligence technologies and a new wave of machines, capable of handling tasks previously reserved for human intelligence, raises significant concerns. It is still too early to predict which professions will be most affected. In 2020, economist Michael Webb estimated that artificial intelligence is likely to have a greater impact on high-skilled jobs, as it goes beyond performing repetitive tasks6 . However, in its annual Employment Outlook report, published in July 2023, the OECD estimates that low- and medium-skilled jobs are most at risk. 

The study by Fabio Cerina, Elisa Dienesch, Alessio Moro and Michelle Rendall provides insight into the consequences for major cities. Low-skilled service jobs are likely to increase evenfurther, as evidenced by the success of companies offering home delivery of food or goods. In 2019, Amazon became the world's most powerful brand, surpassing Google and Facebook. The company employs nearly 1.5 million people worldwide, up 83% since the end of 2019.

  • 6Webb Michael, 2019, « The Impact of Artificial Intelligence on the Labor Market ». SSRN Electronic Journal. 10.2139/ssrn.3482150.

Translated from French by

Translated from french by Cate Evans


Cerina F., Dienesch E., Moro A., Rendall M., 2023, « Spatial Polarisation ». The Economic Journal, 133(649), 30-69.