Approximating default probabilities with soft information

Abstract

We present a new structural credit model that is able to incorporate available soft information, diverse qualitative data and subjective opinions on managerial ability to handle credit events within approximations of default probabilities. We conduct several sensitivity analyses on the model parameters, deploy an empirical exploration of the suggested scheme and simulate realistic lending scenarios. We discover that the proposed model performs exceptionally well throughout the area of elevated type II errors, where loan officers misidentify a nondefault case as a default candidate and wrongly deny loans. Our theory would enable lenders to approve financing in doubtful credit requests and enhance banks' profitability.