Agency Issues in Structural Credit Models - Evidences from Bank Industry

Abstract

We empirically examine the agency issues in structural credit models using bank data from 2001 to 2005. We find five independent agency related factors explaining the estimating errors of various structural models from 27% to 70%. They include factors of profit efficiency, cost efficiency, value, debt-equity, and one mitigating debt-equity agency problem. They are scarcely discussed in literature and should be incorporated into credit modeling. Different from previous empirical literature, we also find that, except the Collin-Dufresne and Goldstein (2001), the other three models underestimate default probabilities on average and the one with dynamic interest rate setting, the Longstaff and Schwartz (1995), performs the best.