NAFCU Journal May June 2021

32 THE NAFCU JOURNAL May–June 2021 without additional risk to your institu- tion? In the short run no, but in the long run, probably. Q: What is your advice on extending current loans? What are the risks associated with these extensions as far as loan losses and income? A: I think institutions should do every- thing they can within reason. However, all extensions and modifications need to fit what the borrower can actually pay, or they will not work in the long run. Regulator guidance requires insti- tutions to be pragmatic in this decision. Q: How will higher risk loan portfolio segments, and larger loans that were modified during the pandemic, need to be supplemented now? A: Large loans need to each be treated individually. Some may recover quickly, and some may take a significant amount of time. Obtaining updated quarterly financial information will be import- ant. Any modifications should require information from the borrower. Higher risk segments, where many smaller loans were modified, will need to be tracked more diligently by separating the pool into modified and non-modified loans. Q: What things should we consider when documenting the composition of our loan portfolio, and segmenting it appropriately? A: Segmenting should be a function of risk and should evolve with significant changes in your risk profile. Segmenting COVID-modified loans from non-modified may be necessary. Using credit quality fac- tors as part of your segment structure can also be very effective. Most institutions use static pools; however, risk migrated pools provide better analysis and more precise allowances. If you want to prove your allowance changes to the auditors and regulators, I recommend using migrated analysis. Q: And once the portfolio is properly segmented, any tips for analyzing the correlation between historical losses and current economic factors, like unemployment rates? A: The best way—and the way we perform this task—is by using statistical regression modeling. We utilize regres- sion modeling in determining segment structures as well as identifying exter- nal factors under the current expected credit loss (CECL) standard. Regression modeling works best because it allows you to determine rates of change between external factors such as housing prices or unemployment and charge offs, prepay- ments and default statistics. However, performing this assessment requires more data historically. Q: What are the next steps for our institution after we obtain the results from our stress testing? A: Once your analysis is complete, the results should also affect the following other parts of your accounting, policies and underwriting: 1. Adjustments to your current allowance or qualitative factors based on the results of the test, and new information gleaned from the analysis. What changes in risk were found? 2. Understanding changes in prepay- ments that may affect your other models such as ALM. 3. Updating your underwriting or loan pricing based on results or changes since last test. Stress testing is instrumental to a financially healthy institution. Hopefully these insights will help your credit union approach your stress testing with more confidence and preparedness. While current socioeconomic factors may be overwhelming and relatively unpredict- able, you can set your credit union up for success in the years to come by relying on your data and knowing how to put it to work for you.