纽约联储-入职员工的拥挤和粘性研发（英）-2023.11-74页_1mb

Summary

This paper provides a microfoundation for the convex adjustment costs in R&D investment. The authors argue that congestion in worker onboarding—where adding more junior workers decreases the probability of successfully transitioning them into productive senior researchers—generates such costs. They develop a partial equilibrium model showing that firms respond slowly to positive shocks and hoard labor during negative shocks to minimize onboarding congestion, behavior that mimics standard investment adjustment costs models.

Empirical evidence using GitHub data confirms congestion. Software developers (a key R&D workforce) exhibit productivity gains during onboarding, but higher ratios of juniors to seniors reduce the onboarding probability, supporting the model's assumption of a declining ρ function (probability of onboarding). These findings are robust across different specifications and data samples.

A calibrated general equilibrium model with idiosyncratic productivity shocks replicates the hump-shaped and delayed responses of R&D investment to both TFP and monetary policy shocks, matching standard ad hoc adjustment cost models. This demonstrates that congestion in onboarding serves as a plausible microfoundation for sticky R&D investment dynamics.

Key contributions include formalizing labor adjustment costs (congestion) in knowledge production and providing novel, quantifiable evidence from GitHub data, opening avenues for future research into labor market dynamics in high-skill occupations and intangible capital investment.