On Infectious Model for Dependent Defaults

TL;DR

This paper introduces a two-sector Markovian infectious model for dependent defaults, extending Greenwood's model to include bidirectional causality, and demonstrates its superior statistical fit and risk evaluation capabilities using real data.

Contribution

It presents a novel two-sector infectious default model with bidirectional causality and introduces new risk measures, CRES and CRVaR, validated with real data.

Findings

The proposed model outperforms previous models based on BIC.

The model effectively captures bidirectional default dependencies.

CRES and CRVaR provide useful risk assessments.

Abstract

In this paper, we propose a two-sector Markovian infectious model, which is an extension of Greenwood's model. The central idea of this model is that the causality of defaults of two sectors is in both direction, which enrich dependence dynamics. The Bayesian Information Criterion is adopted to compare the proposed model with the two-sector model in credit literature using the real data. We find that the newly proposed model is statistically better than the model in past literature. We also introduce two measures: CRES and CRVaR to give risk evaluation of our model.