Talk by Robert Taylor (DSAS Colloquium)
Room: 248
SPEAKER: Robert Taylor - US Model Validation - TD Bank
DATE: October 10th 2019
ABSTRACT: Financial institutions, such as TD Bank, use models in almost all departments of the bank. Whenever a model is used it poses a risk to the bank for a variety of reasons, such as the uncertainty inherent in all statistical models. At a high level, the purpose of the Model Validation team is to ensure that the model risk taken on by the bank is understood, mitigated, and controlled for. To achieve this, Model Validation tests all models that the bank uses on an ongoing basis to ensure they are statistically conceptually sound, use high quality data, perform as expected, and all limitations are understood. After a brief introduction to Model Validation, my presentation will focus on the models themselves that the bank uses and are validated by my team – the retail expected loss capital models. These models are generally used by the bank to predict credit losses. I will discuss the common model methodologies used, data available for modelling, and statistical testing performed by Model Validation when assessing a model.