Production Networks and R&D Allocation

This paper investigates how production networks shape firms’ R&D decisions and the resulting aggregate inefficiencies. We develop a dynamic model in which firms form supply chains through a persistent matching process and rely on those links to trade both existing products and newly developed products from R&D. The model features two sources of misallocation: market-power distortions and a network-formation externality. The second novel externality leads firms to fail to internalize that their R&D makes them more attractive to potential partners. We estimate the model using Japanese firm-to-firm transaction and patent data and show that the first-best allocation lies close to the decentralized outcome. Market-power corrections raise R&D incentives for older firms by relieving double marginalization along their long supply chains. Instead, internalizing the network-formation externality tilts R&D toward younger firms that expand their supply chains rapidly. These opposing forces largely offset each other, leaving the planner’s allocation near the decentralized one.