Do collaborators learn more when they are co-located? Yes! They are 57% more likely to learn! In this Research Policy paper, @hyejin_youn & I track the careers of 1.6m #scholars between 1975-2018. We find the following... https://t.co/Al62SO9y5r 🧵1/7

When #inventors gain experience, they fall behind the #knowledge frontier of their field & are less likely to produce high-impact #patents. ... but when they #collaborate, this is reversed! Presenting today, with @esposito_ch at @ICSSIConference - @HKUniversity @uclaanderson

Very cool! Kellogg Insight (@KelloggSchool) featured our Research Policy study. People learn less when #collaborating across distance. How does this relate to work-from-home #wfh? Together with @hyejin_youn. @HKUniversity @HKUFB @HKUgeography https://t.co/ZVV4VtGlA1

The inventor gender gap in particpation in China has decreased of recent. This, however, has not prevented #inventor collaboration across the gender divide from rising. And there are huge variations by city. Z. Zhang & @fvanderwouden on #women inventors in #China at #GeoInno2024

So, what do these results tell managers? If your aim is for employees in your team to learn during collaboration, you best co-locate them. What about working-from-home #WFH? Back to the office. Knowledge workers benefit from co-location. Perhaps @elonmusk was right.

What about more rigorous evidence? We match locally to non-locally collaborating scholars on observable co-variates (n = 1,565,663) & run a series of regressions. We find that local collaboration is always significantly positively associated with learning (57% more likely!).

Learning rates are highest in early career, but even scholars in mid or late #career stage have significantly higher learning rates when collaborating locally.

The learning premium from local collaboration increased over time Our understanding of how communication & transportation #technologies have resulted in the "death of distance", doesn't hold for #knowledge spillovers in academia. Distance matters - regardless of your cohort (B)

While learning rates differ across academic fields, we find that local #collaboration is always associated with significantly higher #learning rates than non-local collaboration, except for History.

Yes, in all academic disciplines we find learning rates (share of scholars that learn during collaboration) to drop with geographical distance.

To identify 'learning', we sample scholars with >= 3 pubs, in which the 2nd (t-1) is a collab but the 3rd (t) a solo pub. Learning happens if the scholar introduces a novel topic during the collab & is able to produce this on her own in the next pub. Does distance matter?

Stoked to read this excellent post by @mattsclancy, on how #teams of #knowledge producers are organized spatially. As always, Matt does a superb job linking different literature. I am very happy to see my own work mentioned, and that of my friend & collaborator @esposito_ch.

My excellent co-author @esposito_ch is presenting our new work on the speed of knowledge and the way we organize invention. Sign up if you are interested! We have cool findings and conclusions!

Chris @esposito_ch is about to present our new paper today. If you are at #AOM2022 and interested in #innovation and #collaboration, you want to check it out. We love to hear your feedback, ideas and comments! @AOM_STR @AOM_TIM @AOMConnect

Interested in the organization of invention? My fantastic co-author @esposito_ch is going to present our new paper on how increases in the rate of invention has spurred knowledge obsolescence, driving changes in how we organize invention #AOM2022 @AOMConnect @AOM_TIM @AOM_STR

As knowledge production & disruption increased in the 20th century, the negative returns to experience increased, but so did the benefits for teamwork. Knowledge #obsolescence helps our understanding in shifts in the #organization of #invention, such as the rise of #teamwork! 🥳

... it reduces their probability of high-impact #patenting. #Teamwork helps! Inventors working in teams accumulate #knowledge more quickly and recombine newer knowledge. This increases their probability of producing high-impact patents. And this changed over time! ...

Today at 15:00 I'm presenting a paper with @esposito_ch at #DRUID22. We argue for the importance of knowledge obsolescence in the organization of #invention. Inventors accumulate knowledge over time, but also find their knowledge to become obsolete. This is troublesome, because..

In 20th century, individual inventors had hard time catching up to the advancement of the field. Prediction: their high impact work happens early and team size grows. Both are confirmed with the data. We need to invest more to younger innovators. @esposito_ch @fvanderwouden

@mattyglesias I'm dubious that "something's gone wrong". I'm working on a paper with @fvanderwouden where we find that tech change is moving so fast now, inventors can't keep up with the frontier. Science is a bit different from tech, but the same story should apply.