トップページ

Default Correlation Model of Loan Portfolios

Yusho KAGRAOKA
Musashi University

Modeling of default correlations in loan portfolios is important to risk management of consumer finance companies.  Macroeconomic variables and attributes of debtors are promising candidates for explanatory variables in default correlation models, however they cannot explain surge in default rates in the past few years.  We propose a time-varying fixed-effects model of credit score.  The model presumes that an unobservable global factor accounts for overall credit status and each debtor has different factor loading to it.  The model characteristic is analyzed through Monte Carlo simulations, and the model is compared with traditional panel data models with individual- and time-specific effects.  It turns out that the time-varying fixed-effects model is appropriate to explain default correlations in loan portfolios.

→Japanese Ver.

→Back to list

ページのトップへ戻る