The evolution of model and algorithmic risk

With increasing volumes of data, and recent advances in technology and computational power, including the introduction of Artificial Intelligence (AI) and Machine Learning (ML), models are at the heart of every financial institution’s operations – the backbone of every function and business line, from product design, to treasury and trading, risk management, compliance, and internal audit. But as financial institutions increasingly rely on the output of models for their decision-making, the focus on model risk – or the risk of errors in the development, implementation, or use of models – has also continued to gain momentum.

There are several reasons for this. Firstly, the evolving technological capability of these algorithms have resulted in a widespread democratisation of model development, enabling individual users to develop and deploy their own algorithmic models without relying on internal IT or traditional model development functions. While this increases the speed of innovation, it also increases the level of risk that organisations are exposed to, as these new generations of models are not subjected to the same robust testing systems and governance structures as traditional ones.

Secondly, there has been increasing stakeholder expectations related to the documentation, accountability, controls, and risk management of models. Regulators, in particular, have been intensifying their scrutiny on model risks, with a particular focus on models that include elements of AI systems and ML algorithms.

In this paper, we present a robust model risk management framework designed to help financial institutions assess and monitor their model risks. We examine the five key pillars that such an organisation-wide framework would require, and propose the use of a central model inventory to monitor models throughout their entire life cycle.

Finally, we take a look at the five stages of developing a robust model risk management framework, designed to help financial institutions implement an effective risk assessment and quantification mechanism for their models.