I once heard sound advice ‘Start at the beginning’
In order to prepare for CECL, the first thing we have to do is to understand the concept. Compliance with CECL may not start this year, but getting started with CECL does. Aggregating the data today and building the process will help you design a better way to gain insight into your loan portfolio which is all CECL is trying to accomplish.
Let’s boil down what is expected for CECL into 3 steps:
Data Collection - External Inputs - Forward Model
1. The simple process of collecting and warehousing data creates an environment for you to identify inconsistencies such as missing information, original loan data, refreshed credit scores and property values. Not only do you need this information for reporting and analytics, you gain the benefit of clean data.
2. Incorporating external inputs such as economic data in your modeling provides for greater visibility and accuracy determining probability of default and loss severity. Information from this piece of the CECL puzzle further supports pricing and guideline adjustments.
3. CECL language does not provide the exact model specs, so much as it suggests using a forward-looking methodology with cash flows and the ability to track performance and losses. Risk Models exist and have been used for decades but the time is now to prepare your institution for a flexible solution and smooth transition.
Accolade Advisory and Corporate One are strong partners and great resources for Credit Unions. For more information on the CECL model, check out their webinar CECL Series
Institutions are expected to create and maintain an Expected Loss model with forward-looking cash flow analysis. This is not an easy task and most Credit Unions could use assistance with this part but initiating the process can be as easy as aggregating and auditing your data. Reporting will only be as good as the data warehouse so confirm you have the ability to track changes and trends. A small amount of time dedicated to your loan data now will save you mountains of time.
You will be so much more prepared and have more options exploring Expected Loss models with your data element under control.