Cities are coming back to life, but offices are still struggling. In a recent interview, Harvard economist Ed Glaeser shared that, on average, office attendance in the U.S. is "down about 19%, relative to pre-pandemic levels." But some cities are struggling more than others. In New York, office attendance is down 32%. In San Francisco, it's down by more than 50%.
Exactly two years ago, Glaeser and I were on a panel at the University of Zurich with other esteemed urbanists and economists. I then warned that the remote work genie would not go back to the bottle (recording here). And that the significant change of office demand post-pandemic will be in new places — closer to people's homes — and not in traditional CBDs.
Since then, it has become clear that office work will never be the same and that office supply will have to spread into new areas to accommodate employees who don't want a long commute. For example, in New York City, companies are shutting down offices in Manhattan and opening hubs in Brooklyn, within walking distance of where 80% of their employees live.
But what people do in practice is not important. To economists, what matters is what they should do in theory. And just like on that panel two years ago, multiple studies get published to explain why remote work can't work and why companies (and whole economies) must rely on intensive in-person interaction.
A recent example is this talk by MIT economist David Atkin, based on a working paper by Atkin, Keith Chen, and Anton Popov (thank you, Soledad, for sending it to me.
Atkin et al. analyzed mobile phone data to "measure face-to-face interactions (or meetings) between workers at different establishments in Silicon Valley." They then used patent citation data from the firms these workers belong to. In simple terms, they wanted to see whether firms, where employees spent more time interacting in-person with colleagues and employees of other nearby firms, were more likely to generate more patents.
The study also tried to gauge "serendipitous" meetings between people from different companies off-work and see whether these lead to "knowledge spillovers" that generate more innovation.
The researchers found that in-person interactions led to a "substantial increase" in the flow of knowledge (as measured by citation activity). They also found that "serendipitous interactions" with people from nearby firms also had a "sizable" effect on the number of citations generated by the companies involved.
There are a few issues with this study. First and foremost, it is based on data from 2007-to 2019, before Covid kicked off a dramatic shift in the way people work. I also have some methodological concerns about controlling for the fact that large and established companies are more likely to spend money on getting their inventions patented — and also pay more, work less intensively, and have more employees than smaller firms. This means that employees from firms that file more patents are more likely to hang out at a cafe and to (afford to) live and work in an expensive walkable area.
But the biggest issue with these types of studies is a failure of imagination. Remote work itself might genuinely be less productive than in-person work. But it has other benefits that make it attractive and, ultimately, more effective on the whole.
These include that hiring remotely increases your odds of hiring top performers. And the fact that hiring remotely increases your odds of matching a specialized employee to a specialized task in a way that unlocks significant productivity.
Remote work also has a more profound impact that is not visible in direct comparison with in-person work. Remote work does not mean the same people are doing the same tasks, but somewhere else. Ultimately, it means that the work itself will be done completely differently.
Let's use the media world as an analogy. A study might find that successful video productions are more likely to be created by teams that live and work next to other teams that work in the same industry. At the same time, a TikTok or YouTube video produced by a random person in a small town (or a few remote collaborators) is much less likely to succeed. However, The second production method makes it much cheaper to produce hundreds of different videos and see which one succeeds. So, even if producing videos remotely is less productive, it also enables a much higher output than ultimately might have higher odds of creating a "winner."
Something similar is happening at work. Remote will usher in a more modular labor market, where a larger number of people could interact to create things in ways that will ultimately be more productive.