Factor Bias and Heterogeneous Output Quality in Multiproduct Production -- by Yi Lee, Shengyu Li, Mark J. Roberts

This paper develops and estimates a structural model of multiproduct production that allows plant-product quality and factor-augmenting input efficiency. Using plant-level data from Taiwan’s textile industry, we specify a translog multiproduct cost function with input share and output supply equations and estimate the system of equations via GMM. The model recovers plant-specific input efficiencies, output qualities, economies of scale and scope, and substitution elasticities. It does not impose restrictions for non-joint production or input-output separability and does not require imputing product-level input allocations. We document increasing returns to scale, substitutability between labor and materials, and cost complementarities from joint production. Variation in labor shares and product mixes are jointly driven by large and persistent heterogeneity in both input efficiencies and output qualities. A multilateral productivity index shows that variation in input efficiency and output quality is the dominant source of productivity differences across plants.