It’s common for a bank of any size to have both commercial real estate (CRE) loans and construction and industrial (C&I) loans on its balance sheet. Another commonality: most small, midsize and large banks currently don’t have sufficient data to conduct an estimated loss analysis on these assets for the new Current Expected Credit Loss (CECL) standard, which goes into effect for publicly owned institutions in 2020.
At banks where data captures are not in place for all asset classes, 65% need benchmark data on CRE loans and 62% need benchmark data on C&I loans, according to the initial results of the ongoing CECL Implementation Progress Survey being conducted by MountainView Financial Solutions, a Situs company. The survey also shows that data problems are somewhat less severe at banks where data captures are in place, as 42% of those banks still have data issues with C&I loans and 40% have data issues with CRE loans.
In MountainView’s survey of CECL committee and project team members at banks nationwide, CRE and C&I loans were the top two asset classes where banks are either needing benchmark data or having issues with the data they currently possess. Residential mortgages ranked third in the responses to these two survey questions.
“CRE and C&I is where everyone is lacking data – the smaller institutions with just a few loans on the balance sheet, as well as the big guys – as they prepare for CECL implementation,” says Atul Nepal, a quantitative analyst at MountainView. Nepal says the responses to these two survey questions match the comments his team has heard in conversations with banks of all sizes over the last 12 months. Referencing one surprising example, he says his team recently talked to CECL staff at a midsize regional bank that’s heavy on CRE and C&I loans but is still having data issues.
The number of CRE and C&I loans on a balance sheet obviously varies from bank to bank, but even large commercial banks with sizable portfolios are saying they have data problems. MountainView is frequently hearing that while the banks have the loans on their books, they haven’t put the loan data into a database or data warehouse. In other words, they gathered a lot of information in the loan application and reviewed the information in the underwriting process, but they haven’t standardized and centralized the data for feeding into a CECL model.
A second general reason behind the data problems, according to Nepal, is that these loan amounts tend to be very big, so the actual number of loans held in portfolio by a bank may be very small and therefore not sizable enough to do any statistical analysis of risk. Providing an illustration of this point, he says that in looking at $20 million of CRE loans on a small bank’s balance sheet, it actually may just be one apartment complex, one small strip mall and one hotel. With only three loans and with no similar assets or benchmarks for comparison, the bank can’t develop any reasonable and supportable statistical estimations of future losses.
CECL committee and project team members at banks are encouraged to take MountainView’s survey. There are nine questions, participation takes approximately two minutes and all answers are anonymous so no response can be attributed to any person or bank. MountainView will publish a white paper with the survey results in the coming weeks.