As data has become more easily accessible and computers have become powerful enough to run advanced mathematical models, banks have relied more on using models and statistics to assess and manage risk. Models help manage risk, but as the use of models across businesses increase, the misuse of models and model errors can create an even greater level of risk than not using these models in the first place.
Washington continues to actively discourage Wall Street risk-taking which has led to increased regulatory standards. Model Risk Management (MRM) has emerged as a fairly new discipline within Risk Management and is becoming the fastest growing skill set needed for banks as they continue to leverage advanced modeling techniques to measure and manage risk. Even outside of Risk Management, models are used across marketing, human capital management, product management and business operations. MRM frameworks have been adopted by banks to support regulatory agendas and are now becoming more commonly adopted for decision modeling as well.
As a talent consultant in this space, I speak to many leaders on a daily basis about this this subject, from my conversations I have identified below some of the emerging trends we’ll be seeing in 2019 and beyond:
As this area continues to gather more importance in risk management programs, regulators will continue to push financial institutions to invest more within model risk management to focus on a robust development, validation and monitoring capabilities. Regulators expect firms to create MRM frameworks that promote careful model use and develop standarized process for model documentation. The emerging use of Machine Learning models will creates challenges and headaches around the implementation and management of models as well not being able to explain definitively the decisions these models make. Some firms are already hiring and creating AI Model Risk teams to combat this.
As model inventory in banks are increasing, many new models are being designed to create business decisions such as pricing, strategy and liquidity management as well as models being developed for anti-money laundering and fraud detection. The promise and wider application of models have brought into focus the need for an efficient MRM function to develop and validate high-quality models across the whole organization.
Internal Audit Will Become More Important
Internal Audit has become active in managing model risk by actively challenging the modeling teams designs and risk management controls. As a third line of defense and independent to the risk modeling teams. Internal Audit will be used to assess how effective a model risk management framework meets regulatory and business requirements. This has created a staffing need for internal audit teams to hire talent with a background in model development and validation.
Automated Decision and Machine Learning
The challenges of using machine learning models extends to the larger discussion about model risk management. Supervised learning has been around a long time and traditional approaches such as linear regression and logistic regression have been used extensively by banks in risk management applications such as credit risk. With automation becoming more convenient and cost effective across risk and fraud, machine learning and AI models will be used within decision engines and governance programs will need to be properly implemented to prevent exposure for the bank.
Explainable AI (XAI)
This is one for the future. XAI is an AI approach that uses the same technology to solve the problem created by the technology. In XAI, AI is used to read through the code of another AI model and provides an explanation for the outcomes produced by the original AI or machine learning model. XAI can document and translate the learning process of an machine learning or AI model and provide reasoning behind decisions. The technology is used mostly in stealth start ups and RnD programs such as the Defense Advanced Research Projects Agency (DARPA) is spending $70 million to develop this technology. A large retail bank also claims to be investing in XAI for commercial purposes.
The stakes in managing model risk have never been higher. Models are a key part to modern risk management programs and is part of the increased importance of quantitative risk as it’s own discipline. The sooner institutions get started in building value-based MRM programs on an enterprise-wide basis, the sooner they will be able to get ahead of the rising costs and get the most value from their models.
If you would like to discuss this and/or your recruitment plan for 2019 please feel free to reach out to me on email@example.com or (646) 846 3352.