CECL Prep: Understanding Loan Level Data Requirements

Credit unions are bracing for the impact that the current expected credit loss (CECL) model may have on their institutions’ ALLL and capital levels, and they are adopting transition plans now. One step credit unions will take in this process involves making sure the institution knows what types of data will be needed, what systems will need to be in place, and how much of that data needs to be collected and analyzed. Loan-level information will be required in CECL – a big change for many credit unions, which currently may only archive data at the pool level.

Interagency guidance on CECL issued Dec. 19, 2016 described requirements this way:

Specifically with regard to data, to implement CECL, an institution should collect and maintain relevant data to support its estimates of lifetime expected credit losses in a way that aligns with the method or methods it will use to estimate its allowances for credit losses. As such, the agencies encourage institutions to discuss the availability of historical loss data internally and with their core loan service providers because system changes related to the collection and retention of data may be warranted. Depending on the estimation method or methods selected, institutions may need to capture additional data and retain data longer than they have in the past on loans that have been paid off or charged off to implement CECL.

With this in mind, three areas of credit union data quality on which to focus are:

1.  Adequacy

The CECL guidance is intentionally non-prescriptive, meaning institutions have some flexibility in deciding which methodology or methodologies work best for their portfolio or segments of their portfolio. However, to ensure flexibility in selecting appropriate methodologies, it is important to capture a wide range of data points and retain data longer than credit unions have in the past on loans that have been paid off or charged off. Certain data points might be useful for utilizing one methodology over another. For example, individual loan origination amounts might be useful when applying vintage analysis, while individual loan risk classification would be needed for migration analysis. Some data points, such as individual loan duration, could be used for multiple methodologies. If data is inadequate, identify a timeframe for collecting sufficient data to be able to defend the election of a specific methodology. The Sageworks CECL Prep Guide: Data gives examples of data points for each methodology, along with a checklist for ensuring data adequacy.

2.  Coherency

Storage of data is another aspect of the data needs credit unions will have ahead of CECL. It is possible the credit union stores loan data in different places – mortgages in one system, auto loans in another, etc. Can data be consolidated across those systems, and can the consolidated data then be warehoused in a central data repository in perpetuity? Is there a report writer in place to draw this loan-level data into a single calculation?

It is also possible that issues related to the quality of data being stored can make it more difficult to run calculations for CECL. What data has been tested or historically reconciled to make sure inconsistencies across the systems do not cause problems? For example, are headers consistently applied and understandable across systems? Are data point labels and calculations standardized throughout the institution?

3.  Assurance

What controls are in place to ensure the data is reliable, updated frequently enough to accommodate credit union needs, and secure? Data should be backed up frequently and have redundancy to minimize risk. While many institutions say their data is in good shape for CECL, Sageworks – a financial information company that regularly polls institutions on allowance issues – has found that changes to data collection and storage are needed on a go-forward basis. To have sufficient data to produce lifetime estimates under certain methodologies, more historical data will be needed, so the sooner the data is identified and collected, the more complete it will be.

Mary Ellen Biery is a Research Specialist for Sageworks. She can be reached at 984-242-2578 or maryellen.biery@sageworks.com.