Posts

/by

Observations from ALM Model Validations: For the Real Assumptions, Go to the Source

Do the assumptions outlined in your ALM model’s assumption summary line up with what is actually being utilized to calculate the results? In performing model validations, we have seen numerous instances in which there were discrepancies between reported assumptions and what was used in the actual reporting. This situation creates unnecessary challenges and makes it difficult for credit union staff or examiners to evaluate the reasonableness of the assumptions being utilized. Additionally, it can become much more difficult to understand the results and the changes in results.

For example, in one validation the deposit maturity assumptions actually used in the simulation were roughly 2x as long as the assumptions reported in the assumptions summary report. In another validation, the assumed rate changes (betas) on deposits were different than reported in the summary pages.

Mistakes can happen but it is important to identify issues and correct them as quickly as possible. One of the best ways to do this is to go right to the source. If you do not have reports directly from the model that detail exactly which assumptions are being utilized, you should work with your ALM provider to receive them with each simulation. It is good to check the summary information against the model at least annually and any time there are changes in assumption methodology. Seeing what is actually in the model provides better oversight and helps to avoid the risk of someone unintentionally entering something incorrectly in a summary.

/by

Observations from ALM Model Validations: NEV – Loans Devalue in Rate Shocks – or Do They?

When considering valuation as a measure of interest rate risk, and value volatility as an indicator of changes in interest rate risk, many institutions perform net economic value (NEV) analysis. When working with credit unions, or performing model validations, a concern many have is ensuring the models have the “right” assumptions. What is the “right” discount rate? Should credit risk spreads be incorporated? What effective discount rate or what yield curve should be used to discount cash flows – which method is “more right”?

All of the above may be questions to consider but they are distractions from simple analyses credit union management teams can perform when determining if answers are reasonable. For example, take an auto loan portfolio in which the valuation methodology derives a value of $210M in the base rate environment. This same portfolio devalues to $200M in a +300 bp shock. Said differently, the value volatility in a +300 bp shock is -5.00%. From a quick reasonableness test, this is within a 4-6% devaluation range in a +300 bp shock – very reasonable for an auto loan portfolio.

However, does it change the reasonableness answer if the current book value of the auto loans is $199M? While the devaluation of the loan portfolio is certainly reasonable, the resulting answer implies that the loan portfolio could be sold at a 0.50% gain if rates increased 300 bps instantly. That answer is certainly less reasonable. It is important to remember, in Chapter 13 of NCUA’s Examiner’s Guide, NEV is defined as the fair value of assets less the fair value of liabilities. Would it be reasonable to assume a fair value gain on an auto loan portfolio if rates increased 300 bps?

When measuring NEV volatility, the starting value still matters. High starting values can be driven by low starting discount rates. It is good to evaluate both the effective discount rate and the difference between value and book in the current environment. Some models are unable to calculate an effective discount rate. We have found that sometimes in this situation the effective discount rate does not match what the user intended. If you are in the situation of the model not being able to show the current discount rate, extra attention should be given to the value versus book and how the value compares to book in different environments. Optimistically high starting and shocked values can hide risk and volatility; this connects with the cautions brought out in our blog regarding high starting NEV ratios posted on September 25, 2015.