On the Information Content of Bank Loan‐loss Disclosures: A Theory and Evidence from Japan

We develop a model in which banks use loan‐loss disclosures to signal private information about the credit quality of their loan portfolios. The cross‐sectional predictions generated by the model are shown to help to explain previously documented counterintuitive empirical regularities for US banks. We also take advantage of a recent Japanese regulatory policy shift, which first forbade the reporting of restructured loan balances and then forced full disclosure. This policy shift allows us to address a common difficulty in testing signalling theories, in that we are able to construct a timely proxy for the private information that we allege is being signalled. Consistent with our signalling model, we find that banks taking the largest write‐offs turn out later to be the strongest banks, with the fewest restructured loans.