We study the adoption of remote work within cities and its effect on city structure and welfare. We develop a dynamic model of a city in which workers can decide to work in the central business district (CBD) or partly at home. Working in the CBD allows them to interact with other commuters, which enhances their productivity through a standard production externality, but entails commuting costs. Switching between modes of labor delivery is costly, and workers face idiosyncratic preference shocks for remote work. We characterize the parameter set in which the city exhibits multiple stationary equilibria. Within this set, a coordination mechanism can lead to stationary equilibria in which most workers commute or most of them work partially from home. In these cases, large shocks in the number f commuters, like the recent lockdowns and self-isolation generated by the COVID-19 pandemic, can result in dynamic paths that make cities converge to a stationary equilibrium with large fractions of remote workers. Using cell-phone-based mobility data for the U.S., we document that although most cities experienced similar reductions in CBD trips during the pandemic, trips in the largest cities have stabilized at levels that are only about 60% of pre-pandemic levels. In contrast, smaller cities have, on average, returned to pre-pandemic levels. House price panel data by city show consistent changes in house price CBD-distance gradients. We estimate the model for 274 U.S. cities and show that cities that have stabilized at a large fraction of remote work are much more likely to have parameters that result in multiple stationary equilibria. Our results imply welfare losses in these cities that average 2.7%.

Information is critical for migration decisions. Yet, depending on where individuals reside and who they interact with, they may face different costs of accessing information about employment opportunities. How does this imperfect and heterogeneous information structure affect the spatial allocation of economic activity and welfare? To address this question, I develop a quantitative dynamic model of migration with costly information acquisition and local information sharing. Rationally inattentive agents optimally acquire more information about nearby locations and learn about payoffs in other locations from the migrants around them. I apply this model to internal migration in Brazil and estimate it using migration flows between regions. To illustrate its quantitative implications, I evaluate the counterfactual effects of the roll-out of broadband internet in Brazil. By allowing workers to make better mobility choices, expanding internet access increases average welfare by 1.6%, reduces migration flows by 1.2%, reduces the cross-sectional dispersion in earnings by 4%.

Labor market conditions differ widely across regions within countries. Yet, migration patterns often do not respond to these regional differences. Are workers' limited migration responses due to a lack of information on the potential net gains from regional migration? To answer this question, we analyze the mobility decisions of all formally employed workers in Brazil for the period 2000-2014. We use a model of workers' migration decisions to derive novel moment inequalities that allow us to test for the content of migrants’ information sets. Beyond our specific application, the moment inequalities we derive may be used in the context of discrete choice models in which the agent’s consideration set is arbitrary large and the agent’s expected utility from each choice is unobserved to the researcher. Our results indicate that workers face indeed substantial information frictions in their migration decisions. We find that workers on average do not perceive variations in wages across locations that are below the inter-quartile range of the wage distribution. Workers located in regions with a better access to internet and a higher population density have more precise information overall, and workers in general have more precise information about nearby regions.

As states have diverged in their relative demand for skilled workers, the children born in these states have also diverged in their probability of graduating from college. We ask how much of this spatial divergence in college attainment can be explained by the growing differences in the types of people who live in these locations, or by the increased importance of frictions to geographic mobility, such as migration costs and out-of-state tuition. First, using panel data on individual education choices and college characteristics, we show that children are less likely to go to college when they are located in a labor market, at age 17, with a lower skill premium and lower access to quality colleges. Second, we build a dynamic spatial equilibrium model with college choice, financial frictions, migration, and endogenous heterogeneity across locations in the higher-education system. We calibrate the model and show how a spatially uneven skill-biased technical change shock can reproduce most of the observed divergence in college attainment. We find that migration costs, rather than financial constraints, out-of-state tuition, differences in college quality, or increased sorting, is most important for explaining the growing gap in college attainment.