Posted by Anthony Curcio on 3/11/16 2:20 PM
Read more about me: Biography
Despite its official end in 2009, the Great Recession still reverberates throughout the global economy. During and after the crisis, many people wondered how the downturn happened. Was it bad luck? Was it a systemic flaw? Was someone, somewhere to blame? And what could we do to prevent another one?
The Great Recession deeply affected the culture of quantitative financial modeling. Those changes are still evolving almost ten years later. Before the crisis, many viewed quantitative models (and the quantitative analysts, or quants, who built them) with a mixture of awe and admiration.
Judgment, experience, and common sense seemed outdated and unreliable when compared to newer and more complex mathematics. The aura of authoritative modelers seemed to be "My model is terribly sophisticated, and frankly, you are not intelligent enough to understand it. If you do not understand it, then you cannot challenge it. Trust me: I am smart."
In many ways, the Great Recession served to shatter this image of invincible modelers. It reminded us all that quantitative models—and their brainy designers—require oversight just like the rest of us. After the trauma of the Great Recession, it was no surprise to hear calls for stronger and more consistent model governance across the financial world.
Starting around 2010, model governance within the federal space gained popularity as an oversight tool. The United States already had audit and internal control standards, but formal model governance was new. In 2011, the Office of the Comptroller of the Currency issued Supervisory Guidance on Model Risk Management:
A guiding principle for managing model risk is "effective challenge" of models, that is, critical analysis by objective, informed parties who can identify model limitations and assumptions and produce appropriate changes.
Auditors (and others who are especially aware of model risk) sense that model governance is increasingly important. As a modeler, I have learned to welcome governance. In fact, I feel at greater risk when governance is not strong and consistent.
Governance is not a conspiracy by auditors and circular writers to make modelers miserable. Rather, good governance greatly increases the accuracy, reliability, and usability of a model. It also protects modelers from being left holding the bag if the model fails.
During my career as a federal credit modeler, I have designed, built, operated, independently validated, and audited quantitative models. These models spanned the Federal space and amounted to loans and guarantees worth almost $2.5 trillion. While serving my clients, I have been on both sides of the proverbial table: I have been the model developer, and I have been the model reviewer (or auditor). These experiences provided a unique perspective on what good model governance means and how it can help the agency that uses the model.
- How to find the most useful resources and guidance for model governance
- Where to learn about best practices in model governance
- How to create strong documentation that will show evidence of model governance to auditors or other oversight authorities