The re-entry of banking organizations into securities underwriting raises concerns over the possibility of banks using their lending-generated private information to benefit themselves and the issuing firm at the expense of investors. This paper illustrates the use of syndicate structure by lending banks to credibly commit against such opportunistic behavior and to exploit their proprietary information to lower issuance cost for borrowing firm issuers. We show that lending banks predominantly comanage with a high reputation nonlending underwriter. We present evidence that, relative to investment banks, comanaging issues allows lending banks to lower issuance costs for borrowing firms.

We provide evidence that commercial banks extend their reputation in underwriting syndicated loans and private placements (private debt) to their bond-underwriting activities. In the absence of bond market reputation, private-debt-market reputation enables commercial banks to win underwriting mandates from their loan clients. Furthermore, it allows them to credibly commit to investors against opportunistically using lending information and thereby deliver superior certification benefits in the form of higher issue prices relative to investment-bank underwriters. This pricing benefit is not offset by higher underwriting fees and thus results in lower total issuance costs for borrowers.

A nonparametric frontier approach is used to calculate the overall, technical, pure technical, allocative, and scale efficiencies for a sample of 322 independent banks. The sample was drawn from the Federal Deposit Insurance Corporation tapes on the Reports of Conditions and Reports of Income (Call Reports) for the year 1986. The results indicate a low level of overall efficiency. The main source of inefficiency is technical in nature, rather than allocative. Separate efficiency frontiers are constructed to test the effect of branching. However, the distributions of efficiency measures for branching and nonbranching banks are not found to be different. Coauthors are Richard Grabowski, Carl Pasurka, and Nanda Rangan.