Investigating asymmetries in the bank lending channel. An analysis using Austrian banks’ balance sheet data

In the present paper we use a balanced bank panel data set to obtain an inference on two dimensions of the asymmetric response of bank lending to interest rate changes. The cross-sectional dimension is captured by group-specific parameters whereby each bank’s group membership is estimated along with the model parameters. Moreover, the asymmetric response over time is modelled with switching parameters that depend on a latent state variable. The presence of two latent indicators calls for Bayesian simulation methods. The results show that three bank groups, characterized by the groups' average asset total, differ in their lending reaction to interest rate changes. Some sensitivity analysis comparing the results for different group specifications and the models' out-of-sample forecasting performance confirms our model specification.