How's Your Credit Migration Model?

Credit Union Management magazine’s Web-only “CFO Focus” column runs the second Thursday of the month.

The past eight years have been traumatic for most financial institutions. We in the lending business witnessed how dramatically loan portfolios can change over relatively short time frames and how detrimental those changes can be.

Federal and state regulators have made it clear they are going to focus more of their resources on making sure credit unions are doing a better job of managing the risk in their investments and loan portfolios than in the past. A credit migration tool designed around stochastic methodology is an essential part of managing credit risk in loan portfolios. Stochastic methodology magnifies the directionality of credit migration and expands its value with statistical analysis that identifies variables which predict the risk factors of loans. A variety of vendors have developed and offer these models in the marketplace.

Our credit migration model was developed over 20 years, based on extensive experience and research. Here are 10 important findings about loan portfolios in just the past seven years:

• A borrower’s financial situation can change quickly—impacting his ability to pay existing debt or take on new debt.
• Loan portfolio book-values are dynamic and continually changing as a result of borrowers’ finances and shifting credit scores.
• Constantly monitoring credit scores (by individual borrower and by pools) is the most efficacious method to forecast impending delinquencies and charge-offs.
• A credit union’s profitability and future existence is critically impacted by its shifting loan yields and loan delinquency and charge-off expenses.
• 80 percent to 90 percent of total delinquencies and charge-offs are attributable to loans that have experienced a drop of two or more credit grades from the original score.
• Less than 10 percent of total delinquencies and charge-offs are attributable to loans that have experienced unchanged credit scores.
• The sooner a credit union takes action when a borrower shows initial signs of financial distress, the better the chance of mitigating a loan loss.
• Monitoring credit score movement in a loan portfolio is a primary step in accurately forecasting necessary placements to ALLL (allowances for loan and lease losses).
• There are profitable opportunities for those credit unions who punctually market to members whose improving credit scores are early indicators of their elevated ability to service debt and possible desire to purchase large-ticket items long deferred due to budget constraints.
• Problem loans can be spotted sooner than with traditional methods and default risk in loan portfolios can be much better managed by using a stochastically derived credit migration model that has determined and then utilizes statistically validated predictors of impending losses.

Based on the above important findings, we conclude that every financial institution needs to generate reports at least quarterly from a reliable credit migration model. In addition, this model, at a minimum, should be able to provide the following 10 key services:

1. upload and review, at least semi-annually, every borrower’s latest credit score and provide alerts for every borrower showing statistically significant score digression or improvement compared to the borrower’s original score
2. track composite credit score movement for every loan pool broken out by collateral code and credit grade to help determine necessary changes in loan policy/procedures, impacts on profitability, necessary mitigation steps, etc.
3. combine a credit union’s unique market-place data—such as mortgage defaults and unemployment rate—with its delinquency trends to form an “environmental index” for the purpose of forecasting loan default rates and required placements to ALLL.
4. adhere to latest GAAP and other pertinent regulatory guidelines
5. identify individual borrowers in a loan portfolio who exhibit improving credit score trends for the purpose of direct marketing 
6. use a credit union’s unique data base and history to create statistically validated ratios to forecast charge-off rates broken out by credit grades, collateral codes, loan-types, as well as for the total loan portfolio
7. identify the risk inherent in individual loan pools on an ongoing basis
8. provide a testing device and validation process relating to a credit union’s concentration risk policy
9. provide a foundation and validation process for a statistically validated risk-based loan pricing tool
10. has been reviewed by federal and state regulators and determined to meet regulatory requirements

Monitoring the shifting sands of borrowers’ credit scores is key to credit unions’ ability to stay financially viable so they can continue to serve the needs of their members. Using a comprehensive, statistically validated credit migration model is fundamental to that end.

Dennis Child recently retired after 40 years as a credit union CEO, and now serves as marketing director of Thompson Consulting and Training, Boise, Idaho. Childs has been associated with TCT for 25 years. Reach him at 435.770.0178. CEO of TCT, Randy Thompson, Ph.D., designs his training and management tools using empirically based, statistically validated methods and research. Reach TCT at 208.939.8366.

Photo credit: Dollarphotoclub.com/Brian Jackson