Estimating Productivity in the Presence of Spillovers: Firm-level Evidence from the US Production Network.

Abstract

This paper examines the extent to which productivity gains are transmitted across U.S. firms through buyer-supplier relationships. Many empirical studies measure firm-to-firm spillovers using firm-level productivity estimates derived from control function approaches. However, these methods implicitly rule out the interdependence of firms' outcomes and decisions through productivity spillovers. To address this limitation, I develop a framework to jointly estimate network effects and firm-level productivity, while accounting for common productivity shocks across firms and non-random buyer-supplier matching. Using this method, I characterize productivity spillovers over the US production network from 1977 to 2016. My results suggest that having 1% more productive trading partners on average leads to 0.076% higher productivity in the long run. Supplier spillovers, which are driven by both large and small firms, are 4 times greater than buyer effects, which are primarily generated by large firms. Heterogeneity in spillovers within and across sectors also has implications for overall productivity growth: aggregate spillovers tend to be much larger when manufacturers are central in the production network than when retailers and wholesalers are more central.