¶ The work-from-home technology boon and its consequences

NBER link

¶ BibTeX

@techreport{davis2021work,
  title={The work-from-home technology boon and its consequences},
  author={Davis, Morris A and Ghent, Andra C and Gregory, Jesse M},
  year={2021},
  institution={National Bureau of Economic Research}
}

¶ Abstract

We study the impact of widespread adoption of work-from-home (WFH) technology using an equilibrium model where people choose where to live, how to allocate their time between working at home and at the office, and how much space to use in production. Motivated by cross-sectional evidence on WFH, we model WFH as a complement to work at the office. Simulations of the model and recent real estate price data indicate that the pandemic induced a large change to the relative productivity of WFH, one that will permanently affect incomes, income inequality, and city structure.

¶ Notes and Excerpts

We provide evidence on the frequency of WFH prior to the pandemic suggesting that WFH is not a perfect substitute with work at the office. Specifically, prior to the pandemic, very few workers that spent at least some full days working from home spent all days working from home. We thus model WFH and work at the office as potentially complementary in production.

Our benchmark estimates imply an elasticity of substitution (EOS) in production of full days of WFH and work at the office of 3.6, with a 95% confidence interval of 1.002 to 6.105. Since working from home and at the office are complementary, some commuting to the office will occur once the pandemic ends. This suggests that workers will not move en masse to remote, uncommutable areas with low taxes and a low cost of living but may move farther out in their current metro area, to places with long but feasible commutes and lower housing costs

Finally, our work relates to how cities respond to shocks in the short run and the long run. Ouazad (forthcoming) surveys this literature. Our model predicts that the trend towards suburbanization will continue, which is consistent with Ouazad (forthcoming).1 The evidence suggests that natural disasters tend to have only transitory effects on city structure (Davis and Weinstein, 2002; Ouazad, forthcoming), while factors that influence productive capacity, such as transportation, tend to have permanent ones (Bleakley and Lin, 2012; Brooks and Lutz, 2019). Our model predicts that the COVID-induced shock to the productivity of WFH will have long-lasting effects on city structure.

¶ Data Sources

• Current Population Survey (CPS)
  • American Time Use Survey (ATUS): examine time use within one randomly selected day per respondent
    • Leave and Job Flexibility (LJF) module: completed by a subset of ATUS respondents in 2017 and 2018, asks workers directly how frequently they work full days exclusively from home
• General Social Survey (GSS) asks respondents, “How often do you work at home as part of your job?”. Conducted in 2006, 2010, 2014, and 2018.
• 2019 5-year American Community Survey (ACS)
• the 2017 and 2019 waves of the American Housing Survey (AHS).

We estimate the time costs of commuting, t1 and t2, using data from the ACS on the average one-way commute time by workers commuting into Zone

1. Workers living in Zones 1 and 2 commuted an average of 25.7 and 47.7 minutes each way. We estimate the financial commuting cost parameters, τ1 and τ2, using information from a special survey in the 2017 AHS. Our target financial commuting costs for Zones 1 and 2 are $2,226 and $5,565 per year for households working all days at the office.13

¶ Model

Commuting is represented like a pair of taxes on work income and work time. Budget and time constraints for non-WFH person are

Location index is $n$, HH index is $\iota$. Housing is $h$, time at office is $b$. Financial cost of commuting is $\tau$, time cost is $t$.

We explain in Appendix A how a daily model of whether or not to go to work each day, a fixed number of hours in the workday, and a fixed daily cost to commuting maps to an annual model with a multiplicative cost of commuting if households choose the number of days in a year in which to work at the office.

For a person who WFH, the decisions are more complicated because the wage can depend on whether they WFH or commute into work. These households also have some extra terms in their budget constraint for spending on a large home or home office equipment.