The final component of an effective A/L validation is testing the forecasted model outputs against corresponding historical outcomes. We have found this step to be the most beneficial to management and ALCO board members, allowing them to make certain that model inputs and human derived model assumptions are accurate and delivering a true picture of the institution’s IRR. The analysis will include:

• Back testing the model’s ability to accurately forecast asset yields, liability costs, and net interest margin based on historical interest rate movements within an acceptable margin of error. If discrepancies are found, the causes are investigated to discern if a modeling error occurred
• Verify proper model scenarios utilized and model assumption reasonableness based on institution specific needs
• Back testing of deposit offering rate history correlation to key driver rates
• Institution specific non-maturity deposit decay rate analysis
• Loan yield sensitivity assumption review
• Loan pre-payment assumption review
• Loan reinvestment rate assumption review
• Certificate of deposit assumption analysis

As model use increases and is utilized more and more as a strategic planning tool by bank management, the importance of accurate model assumptions is magnified.