Relationship Lending: Do Banks Learn?

This paper explores the implications of asymmetric information and learning in banking. We hypothesize that banks acquire private information about borrower quality, not just when screening the initial loan, but while monitoring the loan’s performance, and that some of the ensuing surplus is shared with relationship borrowers. We develop a general methodology to test for the presence of learning in financial markets, and apply this using a dataset for 7,707 syndicated loans made between 1987 and 2003. Our tests simply require the construction of a variable from the future that is correlated with firm quality and is unobservable by the bank through the entire sample, so it cannot be used directly to price loans. The loading on this proxy in the bank’s pricing equation increases with relationship time, as we would expect if banks were privately learning about firm fundamentals through their relationships with firms. The results are robust to controlling for public, i.e. market-wide, learning. Our findings are the strongest evidence to date that banks acquire valuable private information about borrowers via lending relationships. We find that private bank learning about firm quality particularly benefits higher-quality borrowers, who receive lower interest rates on subsequent loans