Published in: The Review of Economic Studies, 84(7), July 2020 (pp. 1989-2018)

Abstract. In this article, I directly test the hypothesis that interactions between inventors of different firms drive knowledge spillovers. I construct a network of publicly traded companies in which each link is a function of the relative proportion of two firms’ inventors who have former patent collaborators in both organizations. I use this measure to weigh the impact of R&D performed by each firm on the productivity and innovation outcomes of its network linkages. An empirical concern is that the resulting estimates may reflect unobserved, simultaneous determinants of firm performance, network connections, and external R&D. I address this problem with an innovative IV strategy, motivated by a game-theoretic model of firm interaction. I instrument the R&D of one firm’s connections with that of other firms that are sufficiently distant in network space. With the resulting spillover estimates, I calculate that among firms connected to the network the marginal social return of R&D amounts to approximately 112% of the marginal private return.

Joint work with Santiago Pereda Fernández. In preparation for submission.

Abstract. The estimation of spillover and peer effects presents challenges that are still unsolved. In fact, even if separate algebraic identification of the endogenous and exogenous effects is possible, these might be contaminated by the simultaneous dependence of outcomes, covariates and the network structure upon spatially correlated unobservables. In this paper we characterize the identification conditions for consistently estimating all the parameters of a spatially autoregressive or linear-in-means model in presence of linear forms of endogeneity. We show that identification is possible if the network of social interactions is non-overlapping up to three degrees of separation, and the spatial matrix that characterizes the co-dependence of individual covariates and peers’ unobservables is known to the econometrician. We propose a GMM approach for the estimation of the model’s parameters, and we evaluate its performance through Monte Carlo simulations.

Joint work with Alonso Alfaro-Ureña and Jose Vasquez. New version coming soon.

Abstract. Using administrative data for the universe of firm-to-firm transactions in Costa Rica, we study the role and prevalence of “good suppliers”, defined as those upstream firms that provide better, more valuable inputs to their downstream buyers. We then investigate the frictions that might prevent buyers from matching with good suppliers and thus become more productive. Our analysis proceeds in three phases. First, we adapt standard machine learning techniques to the estimation of production functions with many inputs in order to identify the good suppliers in the economy. Next, we quantify the frictions that may preclude buyers from matching with the good suppliers. We do so by empirically estimating a production network formation model through a conditional likelihood approach specifically suited to this problem. Finally, we perform economy-wide counterfactual simulations of industrial policies aimed at supporting good suppliers. The objective of this paper is to study matching distortions in input markets as a microeconomic origin of misallocation in developing economies and to suggest adequate policy responses.

Preliminary and incomplete draft available on request.