Data for CECL: What will you need?
Dec 20, 2016
Financial institutions are still assessing the impact that FASB’s new current expected credit loss model (CECL) will have on their methodology and the resulting allowance for loan and lease losses (ALLL). While the effect on the allowance may be unclear until closer to the implementation deadlines, it’s already certain that calculating expected losses under CECL will require more data than financial institutions are currently using. Indeed, this data will be helpful as institutions run scenarios parallel to the institution’s current ALLL in order to determine which methodologies produce the most accurate estimate of expected credit losses. Some banks are planning to run parallel calculations as soon as the first quarter of 2017.
A central question, then, is how much more data will financial institutions need to gather? At this point, it’s unknown whether having data for 5 years, 10 years or an entire economic cycle will wind up as the “gold standard” for developing “reasonable and supportable forecasts,” as the standards require. Garver Moore, director of special research for Sageworks Advisory Services, says that regulators will expect institutions to utilize data that is reasonably available, so as the institution develops its data sets, one thing to keep in mind is defending what is or is not reasonably available. Clearly, however, starting now and going back as far as you can to gather loan-level data will provide banks with the most flexibility as they develop and test multiple methodologies ahead of CECL implementation. The advance prep work will allow for a more strategic decision about how the institution plans to calculate the reserve under CECL.
Financial institutions that don’t want to be handcuffed by having the wrong data or by having insufficient data can consider a number of suggested areas of focus that Sageworks has outlined in a guide: “Data for expected credit losses in the ALLL.” The guide, available in Sageworks’ CECL Prep Kit, suggests data elements that are strongly recommended, including:
- An unchanging loan identifier that tracks a loan through time. Ideally loan numbers shouldn’t change, but if they do, the institution will need this identifier to track the loan through time.
- A field identifying the data source (which core system, etc.). This field takes on more importance as institutions acquire or leverage multiple data warehouses that could make tracing the data difficult.
- Customer balance
- Book balance
- Coupon rate
- Interest basis
- Segment identifiers, such as call code, product code, geography, etc.
- Credit quality (risk rating, delinquency/nonaccrual/TDR status, credit scores)
- Renewal date and amount
The guide also includes additional data elements that are recommended, along with the rationale for collecting them. These include:
- Payment frequency
- Floor/Ceiling rate (which can be useful in analytics and portfolio risk management reporting)
- Government guaranty information
A few months before the final standard was released, Sageworks surveyed bank and credit union professionals about the readiness of their data archives to run scenarios under CECL, and only about 1 in 3 of those polled said their archives were sufficient. In other words, financial institutions less than a year ago largely reported gaps they are undoubtedly scrambling to fill now. If an institution finds their data is inadequate, options for getting on the road to sufficient data include:
- Contacting the core provider and inquiring about data archiving
- Working with the institution’s internal IT department to create an in-house solution
- Working with a third-party, trusted vendor.
For more information about data problems and solutions related to CECL, listen to the on-demand webinar, “Data Quality Considerations for CECL Measurement.”