A combination survival and time series model for predicting time to default.

摘要

Accurate prediction of default rates are an important component of risk management at large financial institutions. The tools and techniques used to predict default is an important area of research. Lately, with the recent economic crisis, research has started to focus on the incorporation of macroeconomic factors into predictive loss models. The natural extension of this research is to forecast the macroeconomic time series forward and use the results to enhance predictions of future defaults. Proportional Hazards regression models have been a focus of research due to their ability to assess not only if a consumer will default, but when. Additionally, PH models provide a framework to incorporate macroeconomic time series forecasts which could improve prediction and provide a method to stress test portfolios under different conditions.