Model Risk Management – The Quest for a Unifying Approach
Modern financial businesses rely on thousands of models to support decision-making from pricing and reserving through risk and capital to management bonuses and shareholder decisions. These models sometimes fail. Forecasts prove to be inaccurate, or decisions supported by models may turn out to be unwise.
What can we do about this? We cannot eliminate the possibility that the future turns out differently to a model prediction. However, we can ensure that assurance we give on models is both truthful and statistically meaningful. We can reverse stress-test models by feeding them awkward simulated data until they break down. We can choose between harsh validation tests that reveal model weaknesses, or we can apply powerless validation methods where a green light is a foregone conclusion. We can foster a culture where people who become aware of model shortcomings are heard rather than silenced.
This 8-part ASTIN Masterclass uses a series of examples to highlight quantitative approaches to model risk management, using examples related to underwriting risk, stochastic reserving and the modelling of asset price changes. Andrew offers tips for actuaries pressured into expressing undeserved confidence in risky models, together with tips better to support decision making in the context of uncertainty.
ASTIN Members can access each episode by clicking the links below. Please login to your website account.
The first two videos are accessible to all. If you are not an ASTIN Member, please consider joining to gain access to this and future Masterclasses!
About the Speaker:
Andrew Smith is an assistant professor in the School of Mathematics and Statistics at University College Dublin and an Honorary Fellow of the Institute of Actuaries. Before he moved to Ireland in 2017, he gained 30 years of insurance experience, specialising in stochastic modelling, including fifteen years as a partner in a major consulting firm.