As financial institutions progress in the new economic cycle — a cycle defined by gradual interest rate increases, regulatory uncertainty and economic growth — it may be time to revisit the financial models and model processes used to facilitate interest rate risk (IRR) analyses and other risk analyses such as capital stress testing. The accuracy and effectiveness of a model is critical because its outputs may alter the accuracy and effectiveness of related models and impact strategic decisions.
Effective model risk management (MRM) evaluates how risk relates to other models and the role of model risk in an institution’s overall enterprise risk management framework. If done well, an MRM can minimize risk exposure and losses and help an organization operate more efficiently. As a leading provider of independent model validations, MountainView Financial Solutions often sees firsthand what comprises effective model risk management. While numerous variables are incorporated into the development of an MRM platform, MountainView has identified six re-occurring components of highly effective models.
1. Needs Analysis and Strategic Planning. Building an MRM plan requires a significant commitment to analyzing an institution’s unique environment, including but not limited to evaluating model definition; building the model inventory; building a model risk rating process; creating ongoing monitoring processes and limits; identifying model interdependencies; establishing policies and operating procedures, documentation governance, best practices, and roles and responsibilities. Effective planning will capture and address any rogue or misaligned model processes that create inefficiencies.
2. Data Assessment and Management: A well-oiled MRM program will look closely at the data that will be used in its models, recognizing that the quality of data directly impacts model development and implementation success. The use of limited, fragmented, inaccurate, or irrelevant data can significantly impact the precision of a model and model outputs. Moreover, broken data processes, such as data transference between systems, can make it difficult to create effective models and accurately validate models. When the data are poor, the rest of MRM suffers.
3. Model Development and Implementation. An effective MRM program controls the developed model’s implementation. The control framework should ensure the model developed is accurately reflected in the production environment and change control is keenly monitored. The model’s ongoing monitoring tests and thresholds should be established to ensure the model is continually performing effectively in the production environment. Lastly, the controls around the data that feed the production model need to be established to ensure they are consistent with the data used to develop the model; this will ensure that the model output is consistent with upstream or downstream feeder models and dependencies. The hazard associated with downplaying the importance of good model design and implementation range from impaired strategic decisions to heightened regulatory risk and scrutiny.
4. Model Documentation: Comprehensive documentation of a model is a critical component of the MRM framework. Model purpose, limitations, user controls, processes, change control processes, theoretical considerations, alternate methods and quality control procedures will pave the way for a successful governance/regulatory compliance review and will help organizations create a reliable reference point that will outline model specifications.
5. Model Validation. SR 11-7, OCC 2011-12 and other similar industry guidelines require that banks independently validate a model to identify model risks such as problems with a model’s conceptual development, model outputs and assumptions, ongoing model monitoring and verification processes, back testing, governance and documentation. A strong validation will challenge all aspects of your model and specifically report on model risk. Since each institution is unique in its assumptions and business needs, a model validation should be institution-specific and completed by experienced and credentialed model validation experts.
6. Model and Process Optimization: The most effective MRM plans stand out in their ability to help organizations continually review, refine and improve upon their existing model infrastructure. Institutions that find ways to operate more efficiently through workflow automation, for example, will reduce MRM costs across the board. Whether it is the ability to reapply/re-use specific aspects of a model or model process, improve data management for wider use, or identify ways to reduce steps in the development or deployment of a model, institutions that invest in model and process optimization will have a streamlined MRM program that is likely to provide a competitive advantage.