How are banks performing ALLL model validation?
Jan 22, 2015
With increased examiner scrutiny around the allowance for loan and lease losses (ALLL) and the expected release of FASB’s CECL model, banks and credit unions are feeling additional pressure to have their ALLL models validated. The release of guidance on model risk management by the Federal Reserve and the OCC in April 2011 outlined specifics of model validation, including the benefits. According to the guidance, model validation can help ensure models are sound, and it can also identify “potential limitations and assumptions” of the model while also assessing their possible impact.
To help bank and credit union executives better understand supervisory guidance on ALLL model validation, Abrigo and Crowe Horwath partnered for a webinar, which addressed the three main components of ALLL model validation. The session was led by Abrigo’ Ed Bayer, and Crowe Horwath’s Mike Budinger and Ryan Michalik.
During the webinar, more than 500 bankers were polled about their model validation practices. The first question asked attendees when they last completed an ALLL model validation.
While over half indicated a validation was conducted within the last year, almost 25 percent had never performed one. The latter took the presenters by surprise, as Mike Budinger noted he’s seen greater emphasis on the model risk management guidance recently. The supervisory guidance, which took effect in January of 2012, states “Banks should conduct a periodic review – at least annually but more frequently if warranted – of each model to determine whether it is working as intended and if the existing validation activities are sufficient.”
The next poll sought how bankers validated the model – either independently in-house, through an independent third-party or a combination of both.
Performing the validation internally received the highest amount of votes at 43 percent, while 26 percent said they use a third-party company. Another 22 percent utilize a combination of internal and third-party resources to complete their validation.
The supervisory guidance does not specify a preference but does state that validation “should be done by people who are not responsible for development or use and do not have a stake in whether a model is determined to be valid.” While some validation work may be performed most effectively by developers and users of the model, it is necessary that any work be subject to critical review by an independent party.