Working Paper: CEPR ID: DP5539
Authors: Susana Iranzo; Fabiano Schivardi; Elisa Tosetti
Abstract: We study the relation between workers' skill dispersion and firm productivity using a unique dataset of Italian manufacturing firms from the early eighties to the late nineties with individual records on all their workers. Our measure of skill is the individual worker's effect obtained as a latent variable from a wage equation. Estimates of a generalized CES production function that depends on the skill composition show that a firm's productivity is positively related to skill dispersion within occupational status groups (production and non-production workers) and negatively related to skill dispersion between these groups. Consistently, the variance decomposition shows that most of the overall skill dispersion is within and not between firms. We find no change over time in the share of each component, in contrast with some evidence from other countries, based on less comprehensive data.
Keywords: Matched Data; Productivity; Segregation; Skills
JEL Codes: D24; J24; L23
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| within-firm skill dispersion (D29) | productivity (O49) |
| non-production workers (J89) | productivity (O49) |
| differences in average skill levels between production and non-production workers (D29) | productivity (O49) |
| skill dispersion (D39) | productivity (O49) |
| production workers and non-production workers (L23) | productivity (O49) |
| complementarity between production and non-production workers (J29) | productivity (O49) |