For small and large institutions alike, financial models are growing more complex. Business leaders at financial institutions are more frequently relying on financial models to make informed and strategic business decisions, ranging from pricing and launching new products, to managing capital and setting reasonable internal risk thresholds.
Concurrently, financial models have been a focus of regulatory scrutiny. Regulators have less tolerance for black box models and express greater interest in getting “under the hood” to ensure that the institution’s models are accurate, reliable and quantify sensitivities and stresses to various economic scenarios.
This added scrutiny has not deterred institutions from going deeper with their analyses. Sure, some financial risk managers still argue that regulatory requirements hamper innovation and strain resources, but in recent years financial institutions have come to understand the importance of statistical modeling and quantitative analytics. The industry is moving beyond “regulatory check-box” mentality and is recognizing the added value of more sophisticated modeling.
“We have seen first-hand the profound impact that a strong financial model can have on a financial institution’s risk planning and go-to-market strategy,” said Jeff Prelle, Managing Director, Head of Risk Modeling at MountainView Financial Solutions, a Situs company. “But financial institutions should be wary of overconfidence. The 2008 financial crisis was brought on by faulty models concealed under layers of confidence.”
Indeed, a model might have all the markings of quality and accuracy but may still fail in the end, and often for preventable reasons. That’s why MountainView Financial Solutions advocates that the shift to more sophisticated financial modeling should coincide with an increased commitment to more thorough model risk management and model governance. Failure to strengthen model governance could counter any benefits garnered from a better mathematical model.
While regulators are increasing the pressure on institutions to adopt an enterprise-wide model risk framework, inclusive of model governance, it is also in the best interest of the institution to establish best practices, processes and controls around financial modeling. It would be unfortunate to build an exceptional model only to find its accuracy diminished by a flaw in the model process.
“Process flaws are not always the most obvious reason for model failure,” said Prelle. “But we always reinforce with our clients that processes for one model are often linked with another model process, so the extra diligence is well worth the effort.”
Prelle explained that the breadth and depth of its model governance framework should mirror the scope and complexity of the institution’s models. The more complex the model, the more robust the process, controls, policies, accountability and documentation. Prelle further states that model governance can be both robust and efficient.
As the complexity of financial models grows, here are just a few key hot spots to watch in your approach to model governance:
Improve Model Inventory and Tracking
Tracking and cataloging all financial models from a single location enables the institution to prioritize models and better understand model interdependencies. Is the model inventory complete? What is each model’s objective? Are the assumptions or outcomes of one model used in another model? What model limitations exist and how do the limitations affect the overall model framework?
Share Knowledge Across the Organization
The model governance or validation process may change hands within a financial institution many times over or there may be various model teams staffed at the business-unit level. Institution-wide knowledge sharing and detailed documentation for both development and validation will create more transparency across the organization. A more standardized approach to model governance will minimize re-work and ensure model validations and processes are consistent and predictable so everyone is operating from the same playbook.
Find Efficiencies In Your Models and Model Processes
You can create efficiencies in your model processes by identifying duplication in your model inventories and working with developers in the model initiation phase. Business units may not be aware of models or solutions in other areas that may be fit for purpose for the area. This will prevent duplicative models in the organization, reducing validation costs and increasing process efficiencies.
Identify and Track Model Overrides
On occasion, a financial institution may use an override to make a model functional or to meet specific modeling needs. However, too many overrides may be a red flag that the model is not working as intended, is not stable or performing optimally. Institutions should be careful to track the reason, performance of overrides (where possible, test against actuals), and frequency of overrides and should be able to defend those decisions to business leaders, boards and regulators.
Monitor Model Performance Over Time
Model performance over time provides model risk managers with an indicator that there is a flaw in the model, or whether the model is beginning to degrade. As part of model governance, ongoing model monitoring will provide visibility into whether a model consistently behaves in a certain way over time. A model that under- or over-predicts consistently over time, even though it does not fail key performance indicator thresholds, would suggests a flaw in the model as well.
Model governance reaches across all phases of the model life cycle from design and development through implementation, validation and ongoing model monitoring. MountainView provides solutions across all phases of the life cycle and can assist you in developing model governance programs or validating the effectiveness of your existing governance. Reach out today by emailing Andrew Phillips, Andrew.email@example.com.