Workers at remote-first firms were up to 22% more likely to start their own businesses

Tracking 13.9 million workers, economists found that a one-standard-deviation rise in remote work at an employer raised employees' yearly chance of founding a business by 14% to 22%, and the extra startups were more often ones that hired staff or raised venture capital.

Share this:

Remote work did more than move desks into living rooms. It turned more employees into founders. Workers whose employers shifted most heavily to remote work during the pandemic became significantly more likely to start their own businesses, and those new firms tended to be more ambitious than average.

The study is by Alan Kwan of the University of Hong Kong, Ben Matthies of the University of Notre Dame, Richard R. Townsend of the University of California, San Diego, and Ting Xu of the University of Toronto. It was issued as NBER Working Paper No. 33774 in May 2025.

Measuring remote work from web traffic

Firm-level remote work is hard to measure, so the authors built a new gauge from anonymized internet browsing data covering about one billion content events a day. They classified more than 760 million IP addresses as business, residential, mobile or VPN, and measured each firm’s remote work as the share of its employees’ weekday 9-to-5 traffic coming from non-office addresses.

They linked this to LinkedIn employment histories and followed a fixed group of about 13.9 million workers at 137,862 firms who were in wage jobs in February 2020, tracking them from 2016 through 2022. Keeping the group fixed rules out the possibility that remote firms simply attracted more entrepreneurial people later.

To address the worry that firms choosing remote work differ in other ways, they also used a second measure: how far a firm’s employees lived from the office before the pandemic, since long commutes made remote work more likely. A third check used local business closure orders.

A lasting rise in new founders

A one-standard-deviation increase in remote work adoption at a worker’s employer raised their annual probability of starting a business by 14% to 22% relative to the average. Using commute distance instead gave increases of 11% to 13%. The effects appeared in 2020 and persisted through 2022, with no sign of differences before the pandemic.

The spawned entrepreneurs mostly stayed entrepreneurs rather than drifting back to salaried jobs. And among workers who left their jobs, greater remote exposure made it more likely they went into entrepreneurship rather than another job or unemployment.

The additional startups were not just side hustles. The probability of starting a business that later received venture capital rose by 55%, compared with 20% for one without VC backing.

A rough calculation suggests the effect could account for about 317,000 extra business applications a year, roughly 21.9% of the post-pandemic increase.

Time and a safety net to experiment

The authors tested and rejected several alternative explanations. The results held among workers in the same industry, when pandemic boom industries were excluded, and at firms whose employment grew every year from 2020 to 2022, which argues against people founding companies after losing their jobs.

Instead, they conclude that remote work freed up time and reduced employer monitoring, letting workers test business ideas while keeping a paycheck. Consistent with this, the increase was larger when workers entered riskier industries, where testing an idea before quitting is most valuable.

The authors caution that pandemic conditions, including government support and plentiful venture capital, may have amplified the response, and that hybrid arrangements may produce a smaller effect than fully remote work. They also note that firms may lose knowledge when employees leave to start companies, so they may invest less in remote work than would be best for the wider economy.

Study Details:

  • Title: Entrepreneurial Spawning from Remote Work
  • Authors: Alan Kwan, Ben Matthies, Richard R. Townsend, Ting Xu
  • Journal: NBER Working Paper No. 33774
  • Publication Date: May 2025
  • DOI: 10.3386/w33774