Team for Speed: Nonparametric Evidence on Heterogeneous Skill-Specific Affinity in Team Production

Abstract

We examine whether team affinity differs across skill dimensions in team production. Using a novel nonparametric framework that accommodates task-level structure, role asymmetry, and latent affinity, we decompose team performance into skill-specific productivity and unobserved match affinity. As an illustrative application, we analyze elite women's bobsleigh data, where performance can be separated into start and riding phases with distinct individual skill inputs. The estimates reveal heterogeneous, task-specific affinities: coordination and complementarity are stronger in the start phase but weaker and more dispersed during riding, underscoring skill-specific heterogeneity in unobserved team affinity.