FDIC Center for Financial Research Working Paper No. 2005-06. Do Bank Failures Affect Real Economic Activity. State-Level Evidence from the Pre-Depression Era

摘要

This paper provides empirical evidence documenting the existence of a credit channel during the pre-Depression era using a newly constructed, state-level quarterly time series from 1900Q1 through 1931Q2 for the 48 contiguous states. It also investigates the source and size of the credit channel, and it examines the dynamic effects of bank failures on business failures. Granger-causality tests find evidence that bank failures cause commercial failures at the aggregate U.S. level and over half of the 48 states. The cross-sectional variation allows us to test two explanations of the credit channel discussed in the literature: (i) a reduction in consumption spending from the slow liquidation of failed-bank deposits, and (ii) a decrease in investment spending from a disruption of credit to bank-dependent firms. Our results support both theories, but the evidence in favor of the first is stronger statistically. Branch banking restrictions, state- sponsored deposit insurance, and differences in the agricultural-manufacturing share of commerce do not affect the empirical importance of an independent credit channel. Using aggregate U.S. level data, our structural model indicates that bank failures account for about 25% of commercial failures, and that bank failures have only minor subsequent effects within the banking sector.