Rise of the Machines: Application of Machine Learning to Mortgage Prepayment Modeling

Abstract

Key to the valuation of agency residential mortgage-backed securities (MBS) is the modeling of voluntary prepayment and default behaviors of the underlying borrowers in the mortgage pool. The proliferation of both pool and loan level data coupled with access to advanced machine learning algorithms have opened the door to the application of machine learning to mortgage prepayment modeling. The modular prepayment model, one that relies on defined functions to predict mortgage prepayment, has dominated the MBS market nearly since its inception. However, machine learning models are beginning to make inroads and, in some cases, replacing traditional modular prepayment models. The modular and machine learning model differ in the following ways: In the case of modular prepayment models, either added or multiplicative, the modeler defines both the functional form of each feature as well as the “tuning” of the parameters passed to each. Machine learning or “second generation” mortgage prepayment models differ in the sense that the modeler “tunes” the hyperparameters which determine the bias variance tradeoff while the machine determines the functional form of each feature of the model. In this article, we propose a machine learning mortgage prepayment model using a boosted gradient classifier, trained at the loan level and generalized to the pool level. A gradient boosted classifier is a tree-based model using an ensemble of weak learners to create a strong committee for prediction.

Key Findings

• ▪ Machine learning mortgage prepayment models are proving competitive, if not superior, to the traditional modular mortgage prepayment model.

• ▪ Key to training a machine learning prepayment model is managing the bias variance tradeoff through the proper selection of the machine hyperparameters.

• ▪ One of the most powerful techniques to improve performance of a machine learning prepayment model is “boosting”; an ensemble method which improves the predictive accuracy of the model by combining the output of many “weak leaners” into a “strong committee.”