The devil in the details: Dynamic Prediction of loan portfolio profitability with macroeconomic drivers through multi-state modelling
本文提出一个动态监控贷款账户未来预期利润和现金流的框架，整合个体风险、宏观经济趋势和逾期状态转换，帮助贷款机构满足IFRS9或CECL合规要求。
In typical loan portfolios such as mortgages and credit cards, many accounts often experience different stages of delinquency before eventually recovering, fully repaying their balance, or defaulting. From the lender perspective, these events, coupled with the state of the economy, can affect cash-flow and profitability significantly. This paper presents a novel framework for dynamic monitoring future expected profit margins and cash flows of loan accounts, taking into account (i) individual risk profiles, (ii) macroeconomic trends, and (iii) transitions between different stages of delinquency. We make three contributions. First, we show a method to predict future cash flows and profit margins over the life of a loan where the predicted probabilities of an account jumping between delinquency states are obligor specific, time varying, adjusted to be competing risks and dependent on predictions from a macroeconomic model. This model is much more comprehensive model than those in the literature. Second, we investigate different methods to compute optimised cut-offs to be used with the transition probabilities to predict jumps an account is expected to make between different states of delinquency. Third, we illustrate the method using a large sample of 30-year term mortgages and show the expected profit margins for segments of the portfolio. The method will be particularly useful to lenders, who must comply with IFRS9 or CECL.