Dave Keever | Principal | Crowe Howarth
To comply with CECL, banks will need to access much more data, both volume and type, than they do calculating their reserves under the current incurred loss model. Dave Keever, a Crowe Horwath principal, helps attendees get a handle on their current data stockpiles and what they’ll need to forecast losses into the future.
Since 2012 Mr. Keever has directed the firm’s work with financial institutions relative to portfolio credit management and Dodd-Frank Act stress testing, and by extension, their data collection processes. His presentation addresses such issues as capturing loan vintages and identifying loan concentrations and the associated risks, and help attendees learn what data they’ll need for CECL and where in their organizations they might locate data no longer contained in their loan systems.
Data for CECL- and Extended Applications
- Get all your data in one system so you can aggregate. All those types of data are not in financial institutions’ systems. After gathering the data, ensure it’s clean.
- Get finance and credit together – and treasury, because they understand these issues. Under CECL, Q factors will be a tool people use to get a reasonable provision.
- Have to have policies, procedures and controls along with data. You will be moving toward DFAST and this is the same information you’ll need for Dodd-Frank; also most will move in coming years to Basel 3 and you’ll need this same information. So consider what’s coming down the road, and gather the data today that you’ll need tomorrow.
CECL’s scope includes all financial assets measured at amortized cost with the exception of:
- Loans to participants in defined contribution plans,
- Policy loan receivables for insurance companies,
- Promises to give (pledge receivables) of not-for-profit entities, and
- Receivables from entities under common control.
- Financial guarantee contracts that are not accounted for as insurance or at fair value with changes through net income
The key point is that CECL covers more than just loans. CECL starts to take a look at other financial assets, like insurance, wealth management, mutual funds.
Consider available relevant information. What is relevant – vintage? You might not be capturing all that information in a system that will enable you to review and make decisions from it.
CECL can apply to hospitals, medical practice groups – if those are your customers, you need to think about their risk – if they’re setting up loans they’ll have to comply and be an extension of you.
Need to consider remaining life exposure (funds people haven’t drawn upon).
Who tracks the actual date of change when a loan migrates to a new rating?
Considerations you need to look at for gathering CECL data:
- Pools: Can you look at borrowers holistically across different types of loans to judge risk? Look at the overall exposure.
- If a loan is not material an impaired loan could be in an impaired pool, unless its “significant,” then it needs to be viewed individually.
- Capture detailed location information on the borrower and on the collateral as well. Institutions are going to revisit their pooling structure.
- Another issue is the interface between systems. How do you bring all those data into a single warehouse for use. Decide what is critical and get those data in one system or warehouse. If you can get your data all in one location you can pull the data in a matter of minutes.
- Stratify by geography, risk rating, MDRM within states, NAICS codes by risk rating, collateral type by risk rating.
Macroeconomic variables: Use variables that the interagency group pushes out to DFAST banks. The agencies are pushing out domestic macroeconomic variables for three years and going back to 2001. Then see how your charge-offs relate.
Unemployment is the #1 macroeconomic driver of risk for commercial loans. For residential, it is the housing price index. For CRE, the biggest determining factor is the 10- year corporate bonds rate.
Compare your loss rates against other institutions in your footprint to get your comparable to their loss rates for an idea on future projections.