Understanding credit behavior (or any business) at the aggregate portfolio level is becoming increasingly important in an environment where competition abounds and management is under constant pressure to stress test baseline (business as usual) forecasts under a variety of conditions. The term stress test refers to applying different assumptions to your business' portfolio-especially in terms of anticipated economic conditions - to see their impact on demand, revenue, or expenses. In the credit world, significant emphasis is placed into predicting charge-offs because of the huge monetary impact it has on the company's financial status. "Charge-offs" refer to the dollars charged to consumer credit cards that are typically not paid within 120 days. An understanding of the financial risk associated with this type of behavior allows the issuer to set up strategies for accepting new applicants, better manage their current portfolio, develop more efficient collection policies, and decide which are the best territories for market expansion.
For example, what would happen to a portfolio's risk level if general economic conditions cause credit conditions to tighten? Drilling down further, the executive might want to know the impact of increased debt burdens over the next year on specific geographic areas. Besides using the corporation's internal data sources like delinquencies and credit scores as leading indicators of risk, it is often useful to have external sources that could capture trends and potential changes in economic conditions. Furthermore, understanding the portfolio at the geographic level can enable management to determine where to grow the portfolio, maximize profits, as well as when and where to implement conservative policies in future. The following discussion highlights how Equifax's Knowledge Engineering Division helps customers understand this behavior by developing a stress test simulation tool for strategic planning purposes.
Here is a step-by- step procedure for stress testing your portfolio:
Step 1: Collect data over time (at least twothree years by quarter) with respect to some level of geography - preferably at the state or MSA level. Data should include a balance of internal variables (variables under control of the firm like delinquencies, average credit limits, and average credit scores) and external factors such as variables reflecting general economic trends. Equifax, for example, has developed a database of geo-economic data aggregated at the MSA level covering every county in the U.S. In the example that follows, we are interested in predicting the charge-off rate for each of 231 MSA's defined as:
Charge-off Rate (%) = (Charge-off Dollars / Outstanding
Dollar Balances) %100
Where Outstanding Dollar Balances refer to the amount of money charged to a credit card within that MSA that has yet to be collected by the issuer.
Step 2: Estimate a pooled cross sectional time series regression model. "Pooling" here refers to combining all cross-sectional (MSA level) data that is available over time into a single regression equation. The basic idea behind using pooling models is that economic health is not only a time related phenomenon, but a relative one. For example, although the U.S. economy is healthy, it is part of a global economy that is currently experiencing economic difficulties. Likewise, within the U.S., some local geographies are outperforming the norm while others are under-performing. By evaluating these relative differences along with the typical changes we see over time, we can quantify reasons for the variation without collecting a long history of data that is often expensive or impossible to obtain. A good approach is the use of an estimation technique called Heteroscedastic Time-Wise Autocorrelation Regression. Although the actual procedure is statistically complex, it produces an equation looking exactly like a textbook linear regression equation. …