Dynamic credit score modeling with short-term and long-term memories: the case of Freddie Mac's database

Abstract

In this paper, we investigate the two mechanisms of memory, short-term memory (STM) and long-term memory (LTM), in the context of credit risk assessment. These components are fundamental to learning but are overlooked in credit risk modeling frameworks. As a consequence, current models are insensitive to changes, such as population drifts or periods of financial distress. We extend the typical development of credit score modeling based in static learning settings to the use of dynamic learning frameworks. Exploring different amounts of memory enables a better adaptation of the model to the current state. This is particularly relevant during shocks, when limited memory is required for a rapid adjustment. At other times, a long memory is favored. An empirical study relying on the Freddie Mac database, with 16.7 million mortgage loans granted in the United States from 1999 to 2013, suggests using a dynamic modeling of STM and LTM components to optimize current rating frameworks.