A Markov-Econometric Model for Credit Card Portfolio Loan Loss Forecasting

Zhou Xuan Jenny

Abstract

This research examines credit risk management for credit card portfolios with focus on portfolio level loan loss forecasting using statistical models. The lender should be able to manage and estimate its credit exposure and ensure that enough provisions are made against the risk of losses from the borrower’s default. Existing literatures have approached the retail loan loss forecasting from both portfolio dynamics and statistical analyses. This thesis will develop a Markov-Econometric model which combines the considerations of both portfolio performance and external impacts. Several existing models are first examined by implementing them on a real credit card portfolio and comparing their forecasting results. The proposed model delivers improved forecasting power, as compared to the examined basic models, on the real credit card portfolio employed from a commercial bank.