Tuesday October 10 in room A822 at 13:00
Title: Worker Mobility and the Diffusion of Knowledge (with Jeremy Lise, Guido Menzio, and Gordon Phillips)



"We develop a theory of teams to measure the way knowledge diffuses across workers.
We extend the frictional sorting framework of Shimer and Smith [2000] to allow
for workers to influence each other's knowledge. Workers can search on-the-job and
leave their team to start a new team, carrying some of their knowledge with them. In
contrast to standard sorting models, a firm's type is no longer exogenous; it is coworker
human capital. Using a new methodolgy, we estimate the knowledge diffusion process
and the degree of worker complementarities in production with micro wage data and
job mobility patterns from the LEHD, as well as startup patterns from the Integrated
LBD. Our estimated parameters imply both positive peer effects (lower types learning
from higher types) and negative peer effects (higher types dislearning from lower types).
We find that eliminating positive peer effects would lower output by 7%, eliminating
negative peer effects would increase output by 11%, and eliminating on-the-job-search
would reduce output by 17%. Our estimates also imply strong production complementarities.
Setting production complementaries to zero, but still allowing for learning
and worker flows, would reduce output by 52%. Remarkably, under the estimated
parameters, welfare would only be .02% higher if we reshuffled workers to achieve the
planner's allocation."




Subscribe to seminars using Google Calendar