Collect, correct and correlate: CECL standards and the need for data
In June 2016, the Financial Accounting Standards Board (FASB) issued a new standard for the timely reporting of report credit losses on loans and other financial instruments—and in the process, created one of the most significant changes in recent years for financial institutions. The final version of the current expected credit loss standard (CECL) not only affects banks and credit unions, but firms with instruments exposed to potential credit loss as well.
The diligence needed to prepare, understand and analyze the data you need to comply with CECL shouldn’t be taken lightly. In January, we explored the issue in a BAI Banking Strategies piece, “Where credit is due: Why and how you need to prepare for CECL now.” In the second of this three-part series, we discuss how to identify and analyze CECL’s data requirement.
By intention, CECL requirements are not a quick “year in review.” In fact, the FASB calls for significant historical data—and estimated ranges vary, starting at three years all the way up to up to 12. While financial institutions already capture data on an ongoing basis, the majority still lack the precise data needed for CECL compliance.
Each member from your internal team (we detail how to build that team in the January article) will have a unique perspective and understanding of data within the organization. Identify internal data components first: Which aspects allow you to aggregate loan and investment types? And within each category, which additional elements allow for sub-pools?
Organize loans by collateral or type such as mortgages, HELOCs, autos or credit cards. Then create subgroups based on whether the loan types are fixed or adjustable, followed by term. Most likely, your organization’s risk/treasury department already analyzes these levels.
As you segment each category, check to see whether you have other instrument-specific data such as original and current credit scores, original loan to value (LTV), geographic information, employer base, etc. If not, this might signal some of the first data chunks to start collecting. And with this information, financial institutions can monitor concentration risk between loan type, geography and employer information, for example.
Additional data points will allow you to refine the pools and track credit loss with greater detail. With this information, you will see your cost to capital for a given strategy due to credit exposure—which enables stronger strategic planning to bolster profitability and competitiveness.
The team should also evaluate loan and investment types for any other data points that might prove useful. Every financial institution differs, and loan data that might serve as key indicators of credit quality at one institution might prove less essential at another. No one exhaustive data list fits all.
Next, the team should identify external data. What economic factors indicate credit changes and quality? Many common factors can be easily obtained; these include historical unemployment rates, GDP, CPI and housing values. Forecasts can be found just as easily. As your team gathers this external data, it should determine (a) whether national economic values suffice, and (b) if the customer base is concentrated in a way that makes regional metrics useful.
What data matters – and what doesn’t
As you identify external data, do your best to determine which factors can indicate credit changes and which might not. A key component of CECL is to ensure reasonable, supportable assumptions. Therefore, it is important to find which historical internal factors correlate (or don’t) with the right historical external factors.
CECL auditors not only look for the historical data pull, but also your clear understanding of the data itself. Data manipulation comes further in the CECL preparation but with this initial prep, graphs prove extremely helpful in identifying the forces that affect data.
For example, graph one shows loss rate and unemployment over the past 15 years. There is an obvious correlation between these two factors, and this may indicate your institution’s portfolio performance.
Meanwhile, no correlation exists in graph two between inflation and those loss rates. Thus graph two would not be helpful for CECL, since it neither points out a pattern nor tells a story. Different data sets will of course show different correlations that vary by institution.
As you review the data pull, be careful to analyze for any potential flags as your data might be incomplete or unreliable, or include inconsistencies with the time periods measured (i.e., a monthly versus annual basis). Data might also lack governance or accountability if the output isn’t supportable. Graphs are extremely supportive in the quest to avoid inaccuracies.
As data points are identified, the team will need to ask where and how they can obtain consistent, reliable data. Since the team consists of representatives from various groups, it will be easy to see who already owns which pieces. The IT representative should provide insight as to how to organize and centralize all the data for reliable use. The team should also be able to create a plan to consistently gather additional data identified as needed.
We will discuss how your financial institution can use this data to make strategic decisions in the final article of the series. As much as you need to prepare for CECL, you can also prepare to capitalize on the insights revealed by the data you collect.
Tom Caragher is product manager, Enterprise Performance Management, Fiserv. Mr. Caragher has more than 25 years of direct asset liability and balance sheet management experience. He first joined Fiserv in 2005 working in product development and product consulting.