Intertemporal Tradeoffs of Remote Work

Featured in print Digest

This figure is a line graph titled, Engineers’ proximity to teammates and online feedback. The y-axis is labeled, average number of comments per program, ranging from 3 to 9, increasing in increments of 1.5. The x-axis is time, ranging from August 2019 to December 2020. There is a vertical dotted line at around march 2020 that is labeled, march 6th: approximate date of office closures due to COVID-19.  The graph features two lines: One-Building and Multi-Building. Before reaching the vertical dotted line, which indicates office closures, the One-Building line begins at a higher level compared to the Multi-Building line. The One-Building line starts at 7.5 and climbs to approximately 8.75 before the office closures. Meanwhile, the Multi-Building line shows some variation but generally remains within the range of 6.5 to 7 comments. To the right of the dotted line, the paths of the two lines converge and both start descending towards the lower right section of the graph. The One-Building line falls from 8.75 to just below 7.5, and then further declines to around 4 comments by December 2020. Similarly, the Multi-Building line also decreases, reaching approximately 4.5 comments by December 2020, The source line reads: Source: Researchers’ calculations using data from a Fortune 500 firm.

The COVID-19 pandemic sharply increased the number of people working remotely, thereby reducing the frequency of in-person interactions between coworkers. Firms are divided about the effects of remote work, with some top firms recalling workers to the office and others embracing remote work.

In The Power of Proximity to Coworkers: Training for Tomorrow or Productivity Today? (NBER Working Paper 31880), Natalia EmanuelEmma Harrington, and Amanda Pallais find that sitting near coworkers increases knowledge sharing at the expense of short-term productivity. The researchers focus on the main campus of a Fortune 500 firm where software engineers work in two buildings located several blocks apart. Before the pandemic, some engineers were on teams where all members sat in the same building, while other teams were split across buildings. Multibuilding teams held their meetings online, operating more like remote teams even before the pandemic. Once the pandemic hit, the offices closed and all teams worked remotely.

Working alongside coworkers increases feedback and knowledge sharing but reduces short-term productivity, leading to tradeoffs over the short and long run.

Sitting together increased online feedback. Pre-pandemic, engineers on the one-building teams received 22 percent more comments on their code than those on the multibuilding teams — a gap that largely vanished after COVID-19 struck and everyone began working from home. Junior engineers were the main beneficiaries of these comments. However, increased mentorship came at a cost, particularly for the senior engineers who did more mentoring. Sitting near colleagues reduced monthly program production by 23 percent, with an even greater reduction for senior engineers.

Suggestively, proximity had divergent impacts on workers’ careers in the short and long run. In the short run, junior engineers on co-located teams were less likely to receive pay raises, as they spent more time soliciting and responding to feedback at the expense of writing programs. However, after the pandemic closed offices, these engineers were more likely to receive pay raises, consistent with them having accumulated more human capital. Workers who had been trained on one-building teams were also about twice as likely to move to higher-paying firms as were engineers on multibuilding teams (an absolute difference of 1.2 percentage points).

Women, who were underrepresented among software engineers, faced bigger tradeoffs from remote work. Pre-pandemic, women on one-building teams received 40 percent more feedback than their multibuilding counterparts, twice the gap among men. The gender difference was largely driven by follow-up questions, which suggests that women are more willing to ask for additional feedback when seated near their mentors. On the other hand, senior women do much more mentoring in person, leading to a larger decline in their own output when working near colleagues.

Remote workers have externalities on interactions between in-person workers. Pre-pandemic, hiring just one team member in another building, instead of in the same building, depressed online feedback between co-located teammates. This suggests that it may be more efficient to sort teams into fully remote or fully in-person groups rather than having teams with both remote and in-person workers.

—Steve Maas