Productivity and Structural Heterogeneity in the Brazilian Manufacturing Sector: Trends and Determinants

Abstract

Structural heterogeneity (SH)—i.e. the existence of marked asymmetries in labour productivity among firms, along with low-productivity firms forming a large share of total employment—plays an important role in development theory. But only recently has the availability of micro data made the rigorous measuring of SH possible. This paper makes compatible different databases on manufacturing production, innovation and micro-social data for Brazil—PIA, RAIS, Secex and PINTEC—for 2000–2008 in order to measure SH and analyse its determinants. First, productivity groups are formed out of the universe of Brazilian manufacturing firms using a k-mean cluster methodology. Second, the variables affecting the productivity group to which each firm belongs are tested using an ordered probit model. The results indicate that increasing returns (captured by the firm's market share, the number of employees in innovative activities, workers' years of schooling and the accumulation of workers' experience), the technological intensity of the industry, learning by exporting and public support to R&D have driven productivity growth and reproduced SH through time, as predicted by development and evolutionary theories.