Many mortgage lenders are taking on significantly more risk by extending new types of home loans to borrowers. Then, as the real estate boom slows and interest rates and foreclosure rates rise, some of these lenders are wondering aloud about the effectiveness of their credit risk policies.
Similar questions are raised every time the financial services industry tries a new approach that pushes its risk boundaries. In the early 1990s it was adjustable-rate mortgages. In the late 1990s it was credit cards in subprime consumer markets. Now the industry is using higher-risk mortgage products such as interest-only, no-documentation, piggyback, and negative-amortization loans.
As the developer of FICO scores, we at Fair Isaac understand the anxiety that lenders feel when their lending strategies don't deliver the desired results. But we have observed that whenever a lender has a bad outcome, other lenders have good outcomes using the same brand of risk scores to support nearly identical products. The difference seems to be in how well lenders manage their risk strategy in that new or unfamiliar territory.
Given the issues now surrounding new mortgage products, this appears to be a good time to revisit some best practices for using credit risk scores. Here are four we have learned over the years and frequently share with our partners in the financial services industry:
Use all available information. The most important lesson we recommend to any business is to apply all the relevant information you can to every credit decision. Tools such as credit risk scores can become greater contributors when combined with additional predictive information.
Credit risk scores provide benefits by rank-ordering individuals according to an assessment of their history as reflected in credit reports. However, the scores typically do not consider other important factors, such as the person's income or "capacity" to pay or the value of collateral, as these elements are not included in credit reports.
Generally, considering this kind of additional information will foster more informed credit-granting decisions.
Track each portfolio's performance. The ability of a scoring model to rank-order consumers gives lenders vital flexibility. This flexibility is important because the expected risk represented by any given credit risk score can differ between individual portfolios. These differences are caused by a wide variety of factors specific to each portfolio, including the lender's target marketing strategies, brand recognition, the competitive environment, local economic conditions, customer treatment strategies, and consumer lifestyles.
However, within each portfolio the credit scores' rank-ordering should hold true. So as a best practice, we recommend that lenders determine the performance of their customers and prospects within each portfolio, relative to credit score levels. This provides insights that enable the lender to make well-informed decisions regarding score thresholds to use for consumer decisions at different points in the credit cycle.
Monitor score performance over time. Since market factors and lender goals change, we recommend that lenders continually assess the performance of credit scores for their portfolios. This gives the lender the information it needs to adjust its score use in response to market shifts or to meet changing business objectives.
Use champion/challenger testing. For this final best practice, we recommend using controlled strategy testing. This practice is especially important during times of increased uncertainty or when the lender has limited experience in measuring the risk performance of particular market segments or products.
Champion/challenger testing is a systematic way of comparing the performance of a current approach against the performance of a new one that could deliver better results. It's sometimes explained with a horse race metaphor: Two decisioning strategies are run side by side on a fair track to see which one wins. If the challenger wins, it becomes the new champion and remains champion until outperformed later by a different challenger.
This method can be used to improve anything from a product feature or market segment to a strategies for using scores or making other types of decisions.
To help protect the lender's portfolios, champion/challenger testing should be done on a small scale until the lender understands the borrower performance associated with the changing factors. Such controlled testing works well to improve almost any business process.
Used appropriately, credit risk scores can be a key factor in a lender's success with a new product or in a new or rapidly evolving market. By using best practices when managing their risk, lenders can help make sure they come out on the right side of the next market revolution.
