Identifying heterogeneous supply and demand shocks in European credit markets

We propose a new model in which relationship-specific supply and demand shocks are non-parametrically identified in bipartite data under mild assumptions. For example, separate heterogeneous supply shocks are identified for each firm to which a bank lends. We show that a simple estimator is consistent, derive its limiting distribution, and illustrate its performance in simulations. Using these methods, we identify the heterogeneous distributions of supply and demand shocks for thousands of banks and firms in 11 European countries using the Anacredit dataset. Our estimates characterize how both quantity and price elasticities, and thus supply and demand curves, have changed in those 11 markets in recent years. The shock distributions exhibit within-firm/bank heterogeneity that is not well-explained by conventional fixed effects approaches, which only capture between-firm/bank heterogeneity. This unexplained heterogeneity correlates strongly with economically meaningful relationship-level characteristics and macroeconomic policy measures. These results have important implications for policy, identification assumptions in empirical work, and modeling exercises.