Bridging the GAAP: Data for CECL
CECL preparation teams materializing in institutions across the country are asking what data they need to comply with the Current Expected Credit Loss (CECL) allowance accounting standard.
The Board of Governors of the Federal Reserve System, the Federal Deposit Insurance Corporation (FDIC), the National Credit Union Administration (NCUA), and the Office of the Comptroller of the Currency (OCC) issued a Joint Statement on June 17, 2016, summarizing key elements of the new allowance accounting standard and providing initial supervisory views with respect to measurement methods, use of vendors, portfolio segmentation, qualitative adjustments, allowance processes – and data needs. In response to the question, “What should institutions do to prepare for the implementation of CECL?” the agencies provided multiple directives, chief among them:
- Review existing allowance and credit risk management practices to identify processes that can be leveraged when applying the new standard.
- Identify currently available data that should be maintained, and consider whether any additional data may need to be collected or maintained, including:
- Origination and maturity dates
- Origination par amount
- Initial and subsequent charge-off amounts and dates
- Recovery amounts and dates by loan
- Cumulative loss amounts for loans with similar risk characteristics
Breadth and Depth of Data is Important for CECL
While most financial institutions seem to be focused on the breadth of their data, they should also focus on depth – not how wide a net they can cast, but how deep they can go.
As institutions attempt to pull together information for life-of-loan calculations, the value of historical data cannot be overstated – if there is sufficient depth. A bank with eight years of loan-level data, for example, could consider multiple life cycles of five-year auto loans for a historical model that can be applied to future expectations. Whereas, an institution with only three years of data and a weighted average portfolio life of ten years may lack sufficient depth to most appropriately estimate expected losses.
Furthermore, while there is no single equation every institution can use to become CECL compliant, examining the who, what, when, where, and why of CECL data will help you determine your data needs and a methodology suitable for the data you have.
WHO: What departments should be involved in CECL implementation?
Think about having a team of people charged with building a bridge across a ravine. Would you want people with all of the same skills? Or different areas of expertise? If you have people who can do construction, but no one to cut the trees to make the planks, you won’t get very far in your project. It’s similar with an implementation team. A team comprised only of people from risk management more than likely will not be as effective as if others are involved. As bright as they might be and as good at assessing risk, they more than likely will need help extracting the data from the institution’s databases. Continuing the analogy, they can be a major asset in building the bridge but need help making the planks. That’s where IT comes in: to unearth the data and build the planks.
Collaboration among departments and combining multiple skills and expertise will make the transition to CECL more efficient and effective. Each department with influence on the reserve or that is capable of providing data for the reserve should be represented on the team. According to MST Advisory Group Senior Advisor Shane Williams, “Financial institutions should consider including members from IT, allowance, compliance, internal audit, loan review, treasury, internal review, model validation, and profitability departments.”
WHAT: What types of data do you need?
If you’re building a bridge, what kind of wood do you need to support the weight of people crossing it? What are you going to use to suspend the bridge that will hold the weight? Are there enough trees nearby to get enough wood? What tools will you need to fell the trees? You’ll have to assess your needs and determine if you have the resources at your disposal.
It is the same when deciding on a CECL methodology. If you are looking to run a vintage analysis, for example, you’ll need origination and maturity dates, loss rates on the loans in your pools, prepayments, life-of-loan data, and so on. Can you access all that data easily? Do you have it at pool level or loan level? The institution must assess its situation and the feasibility of the methodology it wants to implement. It must be understood whether it can withstand the weight of its auditors’ and examiners’ probes. Establishing a methodology that will hold up under scrutiny will require the input of representatives of multiple departments, each with its own area of expertise.
WHEN: How far back do you need to look for data?
To continue our bridge building analogy, imagine having available only trees that were planted within the last three years. Is that enough wood to build your bridge? Could you get what you need from three years of growth?
You need to consider historical and future timeframes when assessing data availability and prospective methodologies. For a CECL-compliant model, how extensive should your look-back period be? If your residential mortgage pools average 25 years per loan, do you have or need loan-level data going back 25 years to account for the entire life of a loan? Can your historical data account for an economic cycle? If not, what changes in your data retention policy should you make to begin gathering all the historical data you need?
How will you prove the integrity of your historical loss data? “Reasonable and supportable” estimates are required under the new standard and institutions will have to be able to defend their allowances with data that is accurate and consistent with their estimate timeframes.
And how far ahead can you forecast and how well can you defend your expectations?
WHERE: Is our data accessible?
Are the trees for the wood you need for your bridge a few feet away or do you need to transport wood from miles away? For CECL, where does your data reside and how easy is it to access? Is it all in your core system or spread among multiple databases? What departments will contribute what data? Will you need to use external sources to fulfill your data needs?
WHY: Is this the best methodology for our institution?
The kind of bridge you can build will be decided by multiple factors, including the amount and quality of wood you have, the length of bridge needed to span the gap, and the skills of your builders. Can you build something elaborate like a suspension bridge or are you limited to building a simple beam bridge?
Similarly, institutions need to assess their portfolios and data to determine an appropriate CECL-compliant methodology. What kind of methodology will your data allow you to build? Do you use Excel to estimate your allowance or an allowance software platform? Institutions across the country are finding that the manual effort involved with Excel and its propensity for errors make it insufficient for the more sophisticated methodologies. Each financial institution has a unique portfolio and underwriting guidelines, which means each methodology must be a unique design. Financial institutions must evaluate their portfolio regularly to ensure they are using an appropriate methodology.
While there is no one-size-fits-all solution to CECL, answering these self-assessment questions will help you find a CECL-compliant methodology tailored to your institution and its portfolio. The breadth and depth of data required will be greater than ever before, and institutions need to address their data needs proactively today, so they can be prepared for tomorrow.
Learn more about how MST Advisory Services can help your financial institution.
About the Authors
Zach Langley is a Data Analyst with MST Advisory Services. He aids financial institutions in assessing their data and offering solutions in bridging data gaps in the transition of their allowance methodology to CECL. He is passionate in helping clients develop a CECL-compliant model that best fits the institution.
Zach Englert is a Data Analyst with MST Advisory Services. He works with financial institutions across the United States to implement efficient and effective CECL-compliant methodologies. His goal in his work is to approach CECL with a smile and encourages clients along the way to do the same.