A Reconsideration of the Risk Sensitivity of U.S. Banking OrganizationSubordinated Debt Spreads: A Sample Selection Approach

[...]In this paper, we analyze banking organization fundingstrategies using a subordinated debt issuance decision modelestimated with data from three deposit insurance regimes:the de facto too-big-to-fail (1985-87) regime, the purchaseand assumption (1988-92) regime, and the post-FDICIA(1993-2002) regime. We argue that banking organizationsubordinated debt issuance decisions can potentially censor thedata available on issuance spreads for subordinated debtinstruments. To test this hypothesis, we estimate a sampleselection model, which incorporates the issuance decisionmodel, and an ordinary least squares (OLS) model, which doesnot correct for sample selection bias, to consider the risksensitivity of observed issuance spreads (over comparablematurityTreasury securities) as well as the effects ofinstrument characteristics, such as issue size and frequency ofcoupon payments, on such spreads. The model is identifiedusing information available to funding managers, such assupervisory ratings, but not to investors. Using this model, wefind that issuance spreads are risk sensitive during all threedeposit insurance regimes when a sample selection model isused. Importantly, we also demonstrate that the sign andsignificance of parameter estimates that do not correct forsample selection bias can be seriously misleading.[...]