Three-level qualitative classification of financial risks under varying conditions through first passage times

TL;DR

This paper introduces a three-level qualitative classification of financial risks based on first passage times of a firm's value modeled as a geometric Brownian motion, generalizing classical models to account for time-varying conditions.

Contribution

It generalizes the Black-Cox model by classifying risks into high, medium, and low categories based on asymptotic behavior of default functions, without requiring closed-form solutions.

Findings

Classifies risks using asymptotic behavior of default functions.

Extends risk modeling to arbitrary continuous thresholds.

Provides a framework for analyzing time-varying financial risks.

Abstract

This work focuses on financial risks from a probabilistic point of view. The value of a firm is described as a geometric Brownian motion and default emerges as a first passage time event. On the technical side, the critical threshold that the value process has to cross to trigger the default is assumed to be an arbitrary continuous function, what constitutes a generalization of the classical Black-Cox model. Such a generality favors modeling a wide range of risk scenarios, including those characterized by strongly time-varying conditions; but at the same time limits the possibility of obtaining closed-form formulae. To avoid this limitation, we implement a qualitative classification of risk into three categories: high, medium, and low. They correspond, respectively, to a finite mean first passage time, to an almost surely finite first passage time with infinite mean, and to a positive probability of survival for all times. This allows for an extensive classification of risk based only on the asymptotic behavior of the default function, which generalizes previously known results that assumed this function to be an exponential. However, even within these mathematical conditions, such a classification is not exhaustive, as a consequence of the behavioral freedom that continuous functions enjoy. Overall, our results contribute to the design of credit risk classifications from analytical principles and, at the same time, constitute a call of attention on potential models of risk assessment in situations largely affected by time evolution.