Board Oversight of Model Risk Is a Challenging Imperative: A Bank's Board Needs a Sensible, Structured Approach to Oversight of Model Risk, Given the Sheer Volume of Models, Their Wide Scope, and the Complexities Surrounding Their Usage

Article excerpt

[ILLUSTRATION OMITTED]

THE DEPLOYMENT OF mathematical models is growing exponentially, owing to the proliferation of financial instruments and advances in quantitative methods. Every single financial institution in the United States uses models in one form or another.

Scores of new models have been developed and are being used to manage businesses and meet regulatory and reporting requirements. Information provided to the board for review and decisions is increasingly influenced by the hundreds or thousands of models used by the institution. As a result, management and board oversight of model risk and its materiality is gaining in criticalness.

The ultimate goal of model risk management is to reduce model risk to a level consistent with the institution's overall risk appetite. Although the model risk management process needs to have a mechanism for escalating and reporting independently to the board of directors, board oversight over model risk is not widely understood or appreciated.

As with all other risks, the board and its designated committees should follow management's lead in identifying, managing, and mitigating model risks. Understanding how the board's oversight responsibilities are addressed requires definition, particularly in the context of regulators' expectations that the board provide effective, ongoing, and credible challenge to the CEO and management team.

It is the board's responsibility to gain sufficient insight to ensure it's in a position to provide thoughtful, meaningful counsel to management and to exercise skepticism regarding model risk. The board can do this by challenging the assumptions and critically assessing the results, which can be difficult given that models often are voluminous, complex, and subject to radically different outputs.

To help the board fulfill its mandate, management needs to provide the board with enough information, and in the appropriate form, to enable effective challenge without the board members needing to be modelers themselves.

Model Applications

The use of models has increased because they have helped financial institutions reduce the overall uncertainty they face. That is what a model is designed to do--help management make better decisions in the face of complex relationships and uncertainty.

The use of financial models has expanded into an increasing number of areas, including but certainly not limited to:

* Strategic and business planning and forecasting.

* Product development and pricing.

* Customer advice and asset management.

* Valuation of assets and liabilities.

* Management of market risk, operational risk, and credit risk (for example, loan pricing, scoring methodology, loan and lease loss reserves/provision, and risk limit scorecards).

* Trading and portfolio management (for example, value-at-risk limit measurement and hedging, mortgage and asset-backed securities pricing, derivative pricing, and credit value adjustment).

* Regulatory capital (Basel's risk-weighted assets) and the Internal Capital Adequacy Assessment Process, or ICAAP (stress testing, scenario analysis, and economic capital).

* Investment decision support and monitoring.

* Asset and liability management (including net income simulation, economic value of equity and net present value, core deposit behaviors, and loan prepayments).

* Funds transfer pricing.

* Other applications such as anti-money-laundering, customer service, and mortgages and other loans.

An increasing proportion of the recent growth in model applications is driven by regulatory requirements, such as the Comprehensive Capital Adequacy Review (CCAR) and the Basel qualification prescriptions.

Model Risk

Every model provides some level of intelligence.

In 1987 the statistician George E.P. …