- University of Essex
Wednesday, June 5, 2019
Polo Santa Marta, Via Cantarane 24, Sala Vaona
A large theoretical literature emphasizes the importance of financial networks, but empirical studies remain scarce. We exploit the overlapping bank portfolio structure of syndicated loans to construct a financial network and characterize quantitatively its evolution over time. A spatial autoregressive model provides an ideal methodological framework to estimate a linear interaction function, a frequently used building block of theoretical network models. We provide evidence supporting the existence of spillovers that are economically large and time-varying. During major crises network complexity and uncertainty rise, and loan rates are less affected by peers’ decisions. These findings are consistent with learning externality models where reliance on private signals rises during recessions, and with interaction models that emphasize the role of network spillovers. Counterfactual experiments confirm the quantitative importance of spillovers and evolving network structure on lending rates and quantities.