What is the optimal structure of the banking industry? Is it one in which firms have close long-term relationships with a single bank or one in which firms have arms-length relationships with multiple creditors? And what is the effect of competition from public captial markets on these relationships? These questions have come to the fore in the current renaissance in the study of banking. Long-term banking relationships, it is claimed, lower the cost of extenal capital to firms and improve the efficiency of investment (Hoshi, Kashyap and Scharfstein, 1991a). Consequently, these relationships may mitigate the severity of business cycles (Greenwald and Stiglitz (1993) ). This line of reasoning has led some to advocate a financial system that promotes close bank-firm relationships, such as the keiretsu system in Japan or the main-bank system in German, rather than a system with arms-length or market-mediated financing relationships as found in the U.S.