Housing choice in an evolving remote work landscape

Abstract

We estimate a joint model of housing choice along several dimensions to account for changing valuations of housing outcomes due to the COVID-19 pandemic. We consider housing outcomes including housing type, tenure type, the presence of a patio or yard, the number of bedrooms, neighborhood population density, median housing cost, accessibility of amenities, school quality, crime rate, and commute distance. Data used for this analysis were collected in October and November of 2021 from 24 metropolitan areas across the United States. A Generalized Heterogeneous Data Model (GHDM) is used to estimate these housing outcomes as a function of exogenous household sociodemographic characteristics and latent lifestyle propensities. The GHDM also captures jointness caused by unobserved factors, allowing for the estimation of accurate causal effects between outcomes. The results reveal that lifestyle preferences have significant impacts on housing outcomes. Specifically, individuals with a preference for teleworking are more likely to reside in single-family homes in highly populated areas, experience longer commute distances, and exhibit a higher sensitivity to the presence of amenities in their neighborhoods. Additionally, the analysis of tradeoffs between housing outcomes reveals the relative valuations of various housing outcomes. An increased commute distance is found to lead to an increase in single-family homes, reductions in density, and an increased crime rate. Choosing an apartment in a high-density neighborhood is found to lead to reductions in school quality and significant increases in crime rates. Implications of the results for land-use planning, travel demand analysis, and equity considerations are identified and discussed.