Bank Lending Channel in Transmission of Monetary Policy in Japan, 2000–2012: Bayesian Estimation Analysis

Abstract

This paper investigates empirically whether a bank lending channel of monetary policy existed in Japan from 2000 to 2012. We extend Bernanke and Blinder’s model and estimate it with the Bayesian method to deal with the identification problem. In particular, we focus on the differential effects of quantitative easing monetary policy regardless of bank size (City banks vs. Regional banks) and firm size (all enterprises vs. small and medium-sized enterprises). We find that the semi-elasticities of loan supply with respect to bank lending rate are larger than those of loan demand, implying a need for larger decline in bank lending rate to stimulate loan demand following an increase in loan supply. We also find that the semi-elasticities of both loan demand and loan supply are almost the same with respect to bank lending rate regardless of bank and enterprise size. Bayesian impulse response function analyses show an increase in bank lending but a decline in spread following quantitative easing monetary policy shock, which is evidence of the bank lending channel. Variance decomposition analyses show that while a large proportion of forecast error variance in bank loans is explained by monetary policy shock, a large proportion of forecast error variance in spread is explained by loan supply shocks. These results also comprise evidence of bank lending channel. However, we find no evidence that loans of smaller banks and loans to smaller firms are more sensitive to monetary policy.