Evaluating the Risks of Modeling Assumptions Used in Risk Measurement

Over the past few years, risk measurement has become an important, high-profile responsibility for most firms in the financial services industry. With advances in academic theory and in technology, financial risk modeling has grown increasingly sophisticated. Most firms rely on a number of models to analyze their market risks (for example, sensitivity to changes in interest rates, exchange rates, commodity prices, and so on) for asset/liability management. But it is critical to recognize that even the most sophisticated models must make assumptions about key parameters that affect the results of the analysis. This so-called “model risk” reflects the fact that in the real world risk factors are unstable and the historical data upon which many modeling inputs are based can change. This paper discusses model risk, gives specific examples of how model risk can affect fixed-income portfolio valuation, and explains why risk measurement should involve stress testing of key modeling assumptions. If the results of a valuation or asset/liability analysis change dramatically given a small change in a modeling assumption, management may wish to reduce the firm’s exposure to that risk factor, as absolute certainty in financial modeling is an unobtainable goal.