What Does the Yield on Subordinated Bank Debt Measure?

We provide evidence that the yield spread on banks' subordinated debt is not a good measure of bank risk. First, we use a model with heterogeneous investors in which subordinated debt is primarily held by investors with su- perior knowledge (i.e., the"informed investor hypothesis"). Subordinated debt, by definition, coexists with non-subordinated, or "senior," debt. The yield spread on subordinated debt thus must not only compensate investors for expected risk (i.e., to satisfy their participation constraint), but also offer an "incentive premium" above a "fair" return to induce informed in- vestors to prefer it to senior debt (i.e., to satisfy an incentive constraint). Second, we test the model using data we collected on the timing and pric- ing of public debt issues made by large U.S. banking organizations in the 1986-1999 period. Findings with respect to issuance decisions lend strong support for the informed investor hypothesis. But rival explanations for the use of subordinated debt, such as differences in investor risk aversion or such as the signaling of earnings prospects by the bank, are rejected. A sample selection model on observed issuance spreads provides evidence for the existence of the postulated subordinated incentive premium. In line with predictions from the model, the influence of sophisticated investors' information on the subordinated yield spread became weaker after the in- troduction of prompt corrective actions and depositor preference regulatory reforms, while the infl uence of public risk perception grew stronger. Fi- nally, our model explains some results from the empirical literature on subordinated debt spreads and from market interviews — such as limited spread sensitivity to bank specific-risk or of the "ballooning" of spreads in bad times. Keywords: market discipline, subordinated debt, bank supervision. JEL-classification: D8, G2, K2