Maybe 1/3 of the US labor force is working remotely now, due to covid-19. Will this persist? What do we know about the productivity of remote work, in normal times? A thread...
Let's start with Bloom, Liang, Roberts, and Ying (2015). This study is remarkable because one of the coauthors is the co-founder of a large Chinese travel booking company. https://nbloom.people.stanford.edu/sites/g/files/sbiybj4746/f/wfh.pdf
This let them run an actual experiment on remote work!
1. Solicit volunteers (249)
2. Half work from home 4 days/week (treatment); half work in the office (control)
3. See what happens over 9 months
Relative to the control group, the home workers:
were 13% more productive
had a 50% lower attrition rate

Based on the results, the company rolled out policy company-wide.
Another study: Sherman (2019) recruits 187 employees of a life-sciences company to work from home (WFH) more often on even-or-odd weeks of a four-week period.
https://pubsonline.informs.org/doi/10.1287/mnsc.2018.3237
This design lets each individual serve as their own control group. Sherman finds self-reported productivity is higher during WFH weeks. The gains are weaker for those who collaborate a lot (but there are no losses associated with WFH), but higher for working mothers.
In both cases, WFH was mixed with colocated work though. What about full-time WFH?

Choudhury, Foroughi, and Larson (2019) provide evidence on a program at the US patent office that allowed examiners to work from anywhere. https://www.hbs.edu/faculty/Pages/item.aspx?num=55312
Patent examiners had access to a policy allowing them to WFH four days a week. As with the Bloom et al. study, CFL find workers using this program are more productive than colocated peers. But this isn't an experiment, so caution is warranted.
However, in 2011 a new policy allowed examiners to work-from-anywhere as much as they wanted. The program was oversubscribed, resulting in quasi-random assignment into the program among interested workers.
Remote workers in the work-from-anywhere program were 4.4% more productive than those in the standard work-from-home program (who were already more productive than colocated workers).
These studies have nice strategies to measure the causal impact of remote work. But call center operators, patent examiners, life-sciences... maybe these professions are unusually well-suited to remote work?

A few other pieces of evidence suggest yes and no.
The tech sector seems well suited to remote work.

A 2019 internal study by google found no difference in the effectiveness, performance ratings, or promotions for teams and individuals whose work required remote collaboration. https://blog.google/inside-google/working-google/working-together-when-were-not-together/
In the same year, Stripe made a large push into remote work, noting: “the technological substrate of collaboration has gotten shockingly good over the last decade” and "our remote employees have outperformed all expectations.” https://stripe.com/blog/remote-hub 
They use a 2011-2016 survey of a representative sample of Portuguese firms. But they don't have info on how many people actually work remotely. Instead they use a measure of remote capability (firm makes apps available remotely x share of workers using computer).
If you just naively compare firms with remote capability to those without, going from a 0->100% remote capable workforce is associated with 15% higher productivity.

But maybe high productivity firms are just more likely to offer remote work.
So they restrict attention to the subset of firms (394 or 6-8% of sample) that change remote capabilities over their observation period and see the impact on those firms (i.e., they include firm-specific fixed effects).
Now they find going from a 0 -> 100% remote capable workforce DECREASES productivity by 7-10%. But this conceals a lot of variation.
It turns out the negative impact of remote capability is concentrated in lower-performing firms: mid-size and large firms do not see reduction in productivity when they switch to remote work; neither do exporters; neither do firms with high-skilled employees.
In fact, firms that do R&D increase their productivity by 9% when they go from 0 to 100% (potentially) remote.
My take-away? For a lot of firms and industries, remote work works!

But we need more studies, especially longer-term and on full-time remote work. I suspect we'll get them soon.
You can follow @mattsclancy.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: