November 20, 2023
Shifting to work from home
How did work-from-home arrangements in the United States evolve during and after the pandemic?
The COVID-19 pandemic forced dramatic changes on many US businesses and workers. In particular, at the beginning of the outbreak, Americans spent roughly 14 percent of their working hours at home, but within just three months that number jumped to nearly 40 percent.
While many workers have since returned to the office, economists are still trying to understand whether or not work-from-home rates will return to pre-pandemic levels.
In a paper in the American Economic Journal: Macroeconomics, authors Alexander Bick, Adam Blandin, and Karel Mertens documented in real time the rapid increase in work from home over the pandemic and found that the forces that drove the changes are likely to persist into the future.
The speed with which the pandemic hit the economy was unlike anything seen in a generation. Gross domestic product dropped more than 8 percent over the first quarter of 2020, and the unemployment rate went from under 4 percent to 13 percent over the same period.
“It was clear that things were changing very dramatically, but the usual data releases were way too slow to get a good sense of the magnitude of the shock we were looking at. Waiting a month for the next jobs report felt like a very long time,” Mertens told the AEA in an interview. “That's really the origin of this project, which was first launched by my coauthors and quickly after supported by the Federal Reserve Bank of Dallas.”
The authors’ online survey, called the Real-Time Population Survey (RPS), ran from April 2020 through June 2021, collecting responses once or twice a month. The questions closely followed a prominent, conventional survey known as the Current Population Survey (CPS).
The pandemic triggered a change in employers' attitudes toward allowing workers to work from home; it was basically a kind of forced experiment. Some employees found it beneficial, and therefore employers just decided to keep the policy, as the increased flexibility was basically good for everyone.
Beyond collecting data faster, the RPS included questions to inform the work-from-home status for the survey reference week and just before the COVID pandemic hit the United States. They asked respondents specifically about how many days they commuted to work, instead of relying on vague questions about whether they did work from home, which could mean nothing more than checking a few emails.
Further, the authors took steps to address concerns that respondents to online surveys may be systematically different from the general population—possibly more educated or more technically savvy, for instance—which could bias results.
First, they reweighted their responses so that it matched the average CPS responses where the two surveys asked the same questions. The CPS goes to much greater lengths to make sure its samples are representative, so in theory, this technique should help to unbias the results of the RPS.
Second, in a companion paper they show that RPS lines up closely with the CPS along dimensions that were not part of the reweighting, such as hours worked, the industry composition, and the distribution of weekly earnings. In a similar vein, they confirmed that the RPS work-from-home measures are very similar to estimates based on government surveys —such as the American Community Survey and American Time-Use Survey—prior to the pandemic. However, these were not available in real-time, so the RPS work-from-home measure was validated with Google mobility data during the pandemic.
After reweighting, the RPS allowed for a fine-grained look at how commuting patterns unfolded during the pandemic, which the Google Mobility data did not provide.
Initially, the share of workers who commuted every workday declined from 75 percent in February 2020 to 55 percent in May 2020. That number recovered to 62 percent in October 2020, but plateaued and temporarily fell again during the resurgence of the virus in early 2021. By June 2021, the share of full-time commuters was still 62 percent, well below the pre-pandemic level.
The decline in the share of full-time commuters relative to February 2020 mostly reflects a rise in the share of employees who exclusively worked from home. This share quadrupled from 8 percent in February 2020 to 31 percent in May 2020, and declined to 21 percent by October 2020. By June 2021, the share of full-time remote workers was still almost three times the pre-pandemic level.
In contrast, the share of workers who partially commuted dropped from 18 percent in February 2020 to 13 percent in May 2020, but subsequently returned to pre-pandemic levels.
The authors conjectured that the new work-from-home arrangements could have been driven by two different forces.
On the one hand, many workers could have chosen to work from home primarily out of a concern for their health. If this were the case, one would expect work-from-home rates to return to pre-pandemic levels once the virus subsided or vaccines became widely available. On the other hand, employers and employees could have come to appreciate the benefits of more work-from-home days.
“The pandemic triggered a change in employers' attitudes toward allowing workers to work from home; it was basically a kind of forced experiment,” Mertens said. “Some employees found it beneficial, and therefore employers just decided to keep the policy, as the increased flexibility was basically good for everyone.”
Using a quantitative model matched to the survey results, the authors found that as much as two thirds of the post-pandemic work-from-home increase was driven by forces related to the benefits of new work-from-home policies rather than COVID-19 fears.
The authors’ findings suggest that employees are more likely to continue working from home at elevated levels for the foreseeable future, with potentially significant consequences for the rest of the economy.
“Work from Home before and after the COVID-19 Outbreak” appears in the October 2023 issue of the American Economic Journal: Macroeconomics.