Relationship Lending: Do Banks Learn?

There is a vast empirical and theoretical literature that points to the importance of borrower-lender relationships for firms' access to credit. In this paper, we investigate one particular mechanism through which long-term relationships might improve access to credit. We hypothesize that while lending to a firm, a bank receives signals that allow it to learn and better understand the firm's fundamentals; and that this learning is private; that is, it is information that is not fully reflected in publicly-observable variables. We test this hypothesis using a dataset for 7,618 syndicated loans made between 1987 and 2003. We construct a variable that proxies for firm quality and is unobservable by the bank, so it cannot be priced when the firm enters our sample. We show that the loading on this factor in the pricing equation increases with relationship time, hinting that banks are able to learn about firm quality when they are in an established relationship with the firm. Our finding is robust to controlling for market-wide learning about firm fundamentals. This suggests that a significant portion of bank learning is private and is not shared by all market participants.