Transitioning Loss Methodologies

Chris Emery | Senior Systems Engineer | MST

There are multiple reasons for considering a different allowance methodology. Often the bank’s regulators suggest a change, though that is usually supported by other causations. One reason regulators might recommend and the bank make a change is to improve the accuracy of its allowance results. In particular, if the bank is relying heavily on qualitative factors in calculating its allowance, a better methodology might involve a more quantitative approach. The bank might also want a methodology that allows it to react more quickly to changing environmental factors.

Another reason for change is converting to a methodology that will help the bank prepare for the new FASB accounting standard. CECL, in whatever form it will be approved, will most certainly require substantial amounts of loan level data, A methodology that will give the bank better, more granular data will help it prepare for the new requirements. A different methodology could help the bank start building some of the underlying structure that CECL will require. As well, the bank will need to be more forward-looking in its analysis, so the data the bank gathers today will remain useful for several years.

In addition to more and more granular information, a new methodology might also provide better directional consistency, a key concern of auditors and regulators in reviewing a bank’s approach to determining its ALLL.

In addition to its methodology the bank might also want to change other aspects of its calculation, including how it structures its loan pools. There needs to be a balance between too few and too many pools, and the bank should identify what makes loans similar as they reorganize their pool structures.

Two types of loss methodologies the bank might consider as its looks forward to how it will
approach its allowance in coming years:

  • Loss Migration: Calculating loss rates based on the migration of losses back through the history of a loan in order to assign the losses to risk-stratified segments allows for more granular analysis of loss rates based on risk characteristics.
  • Probably of default/loss given default (pd/lgd): The method combines the calculation of the probably of loans experiencing default events with the losses ultimately associated with the loans experiencing those defaults.

The most important consideration in making a change in methodology is the type and amount of data the bank has that will be required to make the change.

The bank might also consider doing a shadow loss analysis. Different from a parallel analysis, which has as its end a conversion to another approach, a shadow loss analysis allows the bank to test the accuracy or direction of its current methodology.