Why Did Remote Work Persist in Some Cities but Not Others After the Pandemic?

Monte, Porcher, and Rossi-Hansberg (2025) asked whether the COVID-19 pandemic permanently changed city structure by shifting workers toward remote work and whether those changes differed across cities. They analyzed a dynamic urban economic model together with U.S. cell phone mobility data from SafeGraph, housing price data from Zillow, Census and American Community Survey data, and National Longitudinal Survey of Youth data. They found that many large cities stabilized at about 60% of their pre-pandemic commuting levels, while smaller cities largely returned to normal. Cities that shifted to a remote-work equilibrium experienced average long-run welfare losses of about 2.3%.

Why This Article Was Selected for The Policy Scientist

The rapid expansion of remote work has become one of the defining policy and economic questions of the post-pandemic era because it affects labor markets, housing demand, transportation systems, municipal finances, and the long-term competitiveness of metropolitan areas. This article makes a timely contribution by examining why some cities returned to pre-pandemic commuting patterns while others did not. The authors have published extensively on urban economics and spatial equilibrium, and this study extends a well-established literature on agglomeration economies by introducing a dynamic coordination framework. Published in the American Economic Review, one of the field’s most influential journals, the article draws on an exceptionally strong combination of national mobility, housing, and survey data. Although the structural modeling is rigorous, future research would be strengthened by exploiting quasi-experimental causal inference designs to more directly identify the mechanisms underlying persistent remote work adoption. The findings are likely to generalize to many advanced economies with comparable urban structures and labor markets.

Full Citation and Link to Article

Monte, F., Porcher, C., & Rossi-Hansberg, E. (forthcoming). Remote work and city structure. American Economic Review. https://www.aeaweb.org/articles?id=10.1257/aer.20231414

Central Research Question

This study examines whether the dramatic increase in remote work during the COVID-19 pandemic produced permanent changes in the structure of American cities or simply accelerated a temporary adjustment. The authors ask why some metropolitan areas largely returned to pre-pandemic commuting patterns while others stabilized at substantially lower levels of office work. They argue that the answer depends on whether cities possess multiple long-run equilibria in commuting behavior. In their framework, working downtown generates productivity and social benefits because workers interact with one another, but commuting imposes costs. These competing incentives create a coordination problem. Workers are more willing to commute when many others also commute, but remote work becomes increasingly attractive when fewer employees are present in the central business district (CBD). A temporary disruption, such as the pandemic, can therefore permanently shift some cities from one equilibrium to another. The study also evaluates the consequences of these transitions for housing markets, commuting behavior, and economic welfare.

Previous Literature

The article builds upon several major strands of urban economics and labor economics. The first concerns agglomeration economies, which emphasize that cities increase productivity because firms and workers benefit from proximity through knowledge spillovers, labor market matching, and shared infrastructure. Classical urban models explain why employment concentrates in city centers and why housing prices generally decline with distance from downtown. More recent research has examined how advances in communication technology and remote work affect these traditional relationships.

A second literature investigates the productivity of remote work. Previous studies have produced mixed findings, with some reporting productivity gains in certain occupations while others identify declines resulting from weaker collaboration, higher communication costs, and reduced knowledge sharing. Other researchers have documented substantial changes in residential location choices, commercial real estate demand, and urban housing prices following the pandemic.

The present study extends this literature by introducing a dynamic coordination mechanism rather than relying primarily on exogenous improvements in remote-work technology or worker preferences. Instead of assuming that remote work became dramatically more productive after COVID-19, the authors argue that temporary reductions in commuting altered workers’ incentives to return to the office. They demonstrate that this mechanism can generate permanent changes even when the underlying technology changes only gradually. This theoretical contribution connects established models of agglomeration economies with emerging evidence on remote work and urban restructuring.

Data

The analysis combines several large datasets that measure worker behavior, housing markets, wages, and mobility. To examine commuting patterns, the authors use anonymized SafeGraph cellphone mobility data covering millions of devices across hundreds of metropolitan areas before and after the pandemic. These data allow them to identify central business districts and measure changes in trips to downtown locations over time.

Housing market outcomes are measured using Zillow zip code-level housing value data, which provide monthly estimates of changes in residential property values throughout the study period. These data allow the authors to estimate changes in the housing price gradient relative to distance from the CBD.

Labor market information comes from the Decennial Census, the American Community Survey, and the National Longitudinal Survey of Youth. These datasets measure remote work participation, hourly earnings, worker characteristics, occupations, and industries over several decades. The panel structure of the National Longitudinal Survey of Youth allows the authors to account for persistent differences among workers when estimating wage differences associated with remote work.

The study also incorporates employment statistics from the Bureau of Economic Analysis and additional demographic information to estimate structural parameters for hundreds of U.S. metropolitan areas. The combination of administrative, survey, mobility, and housing data provides unusually comprehensive evidence on both worker behavior and urban economic outcomes.

Methods

The study develops and estimates a dynamic structural model of city organization in which workers repeatedly choose between commuting to the office and working remotely. The model incorporates several important mechanisms simultaneously: productivity gains from face-to-face interactions, commuting costs, congestion, housing location decisions, heterogeneous worker preferences, and switching costs associated with changing work arrangements.

Unlike many reduced-form empirical studies, the analysis relies primarily on structural estimation and dynamic equilibrium modeling. The authors estimate the model’s parameters using observed commuting patterns, housing prices, labor market outcomes, and worker mobility across U.S. metropolitan areas. They then simulate how cities evolve following a temporary shock such as the COVID-19 pandemic.

The empirical analysis compares the model’s predictions with observed cellphone mobility and housing market data. It also estimates changes in housing price gradients, remote work wage premiums, and commuting behavior across metropolitan areas of different sizes.

The statistical approach is sophisticated and well suited to evaluating long-run equilibrium behavior. Nevertheless, the study does not employ a randomized controlled trial or a modern causal inference design such as difference-in-differences, synthetic control, regression discontinuity, or instrumental variables to identify the underlying causal mechanisms. Consequently, although the structural framework provides internally consistent explanations for the observed patterns, future research would benefit from complementary quasi-experimental evidence capable of more directly isolating the causal effects of persistent remote work on urban outcomes.

Findings/Size Effects

The empirical results reveal striking differences between large and small metropolitan areas. During the early months of the pandemic, commuting declined sharply in virtually every city regardless of size. However, the subsequent recovery differed substantially across locations.

Large metropolitan areas—including New York and San Francisco—experienced permanent reductions in commuting. By the end of the observation period, commuting had stabilized at roughly 60% of pre-pandemic levels on average, with some cities remaining closer to 40% of their previous downtown traffic. In contrast, many smaller metropolitan areas returned almost completely to their pre-pandemic commuting patterns.

Housing markets displayed comparable divergence. Large cities experienced persistent flattening of housing price gradients, indicating that proximity to downtown became less valuable after remote work expanded. Smaller cities largely returned to their previous housing price gradients as commuting recovered.

The estimated structural model predicts that larger metropolitan areas are considerably more likely to possess multiple stationary equilibria because stronger agglomeration forces make coordination among workers more important. Under these conditions, the pandemic acted as an equilibrium-selection event, permanently shifting many cities toward a high-remote-work equilibrium. Smaller cities, by contrast, generally possess only a single equilibrium and therefore naturally returned to their previous commuting patterns after pandemic restrictions ended.

The welfare analysis suggests that these permanent transitions impose measurable but relatively modest costs. Across cities estimated to have shifted toward remote-work-intensive equilibria, average long-run welfare losses equal approximately 2.3%. Estimated losses reach approximately 3.2% for New York, 2.8% for San Francisco, and roughly 3.7% for Los Angeles and San Jose. These losses reflect reduced productivity from weaker agglomeration effects, although they are partially offset by lower commuting costs and improved flexibility for workers.

Conclusion

The study argues that the pandemic permanently altered the structure of many large American cities not because remote work suddenly became dramatically more productive, but because temporary reductions in commuting changed the incentives governing workers’ collective decisions about office attendance. This coordination mechanism explains why otherwise similar cities experienced fundamentally different long-run outcomes following a common national shock.

By integrating structural economic theory with multiple large-scale datasets, the authors provide a coherent explanation for persistent changes in commuting, housing markets, and urban organization. Their findings suggest that the long-run effects of remote work depend not only on technology and worker preferences but also on the underlying economic structure of individual cities. The results have broad implications for urban economics, labor markets, transportation planning, commercial real estate, and local public finance, while demonstrating that temporary disruptions can permanently reshape complex economic systems when multiple long-run equilibria exist.