The views and opinions expressed in this article are those of the thought leader and not those of CeFPro.
By Liming Brotcke, Head of Model Validation/Sr. Director, Ally
How can technology be used to drive model risk management?
Technology can be leveraged to improve model risk management efficiency. Technology that enables tracking of individual models across each stage of their life cycle will improve model governance efficiency by minimizing manual maintenance of a model inventory. It also promotes stakeholders’ full engagement during model development, implementation and use. A dynamic model information system that builds connections and linkage of data and usages could be a powerful tool for quantifying model dependency and interconnectedness. Advanced technologies can also be leveraged to automate the benchmarking and challenger model development as part of the validation activities.
Why is it important to expand definitions and governance groups?
Model development, implementation and use have been evolving rapidly. While conventional statistical regressions and econometric models continue to dominate the modelling methodology choices, there has been a fast recognition and adoption of alternative approaches that rely on machine learning algorithms and other nonparametric modelling approaches by the financial industry. There is a growing interest in moving model implementation from mainframe operation platforms to Cloud and other online venues. Model use has also expanded from core banking business such as deposit and credit lending to servicing functions including HR and Information Security.
As a result, modifications need to be made in order to accommodate those changes while maintaining the independent and effective challenge of the model risk management function.
What skillsets would be most desired within an expanding model risk management team?
Extensive quantitative knowledge of mathematics, statistics or econometrics, and the ability to identify weakness, evaluate impact and assess materiality will continue to be the most desirable skillsets. Meanwhile, MRM teams welcome the following skills including the aptitude for mastering multiple programming languages including SAS, R, Python, the desire to learn machine learning algorithms, and the ability to understand reliance on computer science of certain artificial intelligence solutions. Those skillsets altogether are critical to ensure sustained success of the MRM function as model development, implementation and use has been changing.
What advice would you give when replacing outdated methods?
Replacing outdated methods is like replacing an existing model. Developers are expected to provide justification of why the methods are deemed outdated and why the proposed replacement methodology is appropriate. For example, developers can start with comparing the theoretical construct of both the “dated” and the “new” methods, followed by how differences in theory and assumptions address business relevance and use differently. Then, developers can perform a side-by-side comparison of actual application of both approaches to gain additional insights on caveat/benefit of each methodology in a parallel evaluation.
What do you see ahead for the future of model risk management?
Model risk management is the process of identifying, evaluating, assessing and controlling risk stemming from model fundamental errors, performance deterioration, and inappropriate use of models. Model risk, if not appropriately identified and mitigated, could bring negative impact on an organization’s earnings, capital, liquidity and other areas. While this process is likely to stay intact, MRM needs to incorporate new elements in order to keep up with the advancement in model development, implementation and use.
You may also be interested in…
Have you made your free account?