Connecting the dots: the network nature of shocks propagation in credit markets

We present a simple model of a credit market in which firms borrow from multiple banks and credit relationships are simultaneous and interdependent. In this environment, financial and real shocks induce credit reallocation across more and less affected lenders and borrowers. We show that the interdependence introduces a bias in the standard estimates of the effect of shocks on credit relationships. Moreover, we show that the use of firm fixed effects does not solve the issue, may magnify the problem and that the same bias contaminates fixed effects estimates. We propose a novel model that nests commonly used ones, uses the same information set, accounts for and quantifies spillover effects among credit relationships. We document its properties with Monte Carlo simulations and apply it to real credit register data. Evidence from the empirical application suggests that estimates not accounting for spillovers are indeed highly biased.