The painstaking regulatory and reporting burden that has plagued consumer and mortgage lending for decades is coming to the world of commercial lending fast—and its name is CECL (Current Estimated Credit Losses, pronounced like the name “Cecil”).
The name Cecil means “blind,” which is ironic, because FASB’s upcoming guidance will push FIs to clarify the future performance of their loan portfolios by using models to predict CECL of all loan portfolios. Achieving this objective for commercial portfolios is virtually impossible, however—in the short term, at least—due to the lack of useable data to produce any model with statistical relevance.
The key word here is “useable.” Banks have data on their commercial portfolios, it’s just not useable because:
Data is captured in Word documents and Excel spreadsheets. To make this type of information useable means hiring a bunch of interns to pore through the mountains of paper documents, Word and Excel files to convert into specific data elements.
Metrics are not standardized. You may want debt service coverage to be 1.25, but how did you modify the formula to define it from company A to company B?
Current risk ratings aren’t helpful. How helpful are risk ratings as a risk assessment tool if 80%+ of the portfolio is in a single category? Additionally, the overall accuracy of the individual risk ratings given the subjective criteria such as “generally performs in line with industry” or “adequate collateral support” is suspect in many banks today.
Past charge offs/loss data is unavailable. How far back did the bank retain this data?
Banks need to establish plans to ensure that they maximize the value of the CECL effort. Approaching CECL from a singular vision of estimating credit losses on commercial loans for just Allowance for Loan and Lease Losses (ALLL) reporting is a waste of time. While banks will work to comply with the FASB guidance, this major effort has the opportunity to be much more far-reaching and provide greater vision and overall benefit in risk assessment and forecasting than ever before.
Maximizing the Value of CECL Models
The typical formula to estimate future credit losses is as follows:
PD (Probability of Default)
X EAD (Exposure at Default)
X LGD (Loss Given Default)
= Current Expected Credit Loss
In addition to CECL, this model can be beneficial for:
Loan Pricing. With greater granularity and future accuracy of the risk assessment from this model, banks can provide greater accuracy in the risk premium portion of the pricing model. Instead of just the operational cost of originating a loan and/or maintaining the relationship along with the cost of funds, bankers can layer the potential loss factor to the loan and relationship to determine pricing. Instead of just negotiating for more deposits (assuming the deposits are actually realized), banks can modify the risk factors to reduce the probability of default or improve the collateral coverage to reduce the loss given event of default.
Portfolio Monitoring. For credits with a relatively high probability of default, but low loss given event of default, managing the credit should always focus on the collateral first. If a bank discovers that the relationship risk has increased, the institution’s first action should be to ensure it retains appropriate perfection of its security interest and that the value of the collateral has not deteriorated. On the flip side—with a low probability of default but high loss given event of default—a credit officer’s first priority is to make sure nothing has changed with the repayment capability of the borrower, because should payments stop, the bank has major problems.
Risk Rating Updates. Banks have an opportunity to get rid of narrative and judgmental-based risk ratings, and more systematically update the risk assessments as new information is received. They can: 1) spread a statement, 2) update collateral value, 3) note changes in industry performance, and 4) update the value of the current estimated credit loss.
Stress Testing. How will the impact of declining rent/square footage rates in office space impact a bank’s commercial real estate portfolio? If the collateral value drops, impacting our loss given default, what does that do to the expected loss? For those credits where the expected loss drops too far, banks can assess their options to lower probability of default to compensate.
Getting Started with CECL
The bank’s credit group needs to lead two concurrent efforts: 1) collect the data, and 2) develop the model framework.
It’s a framework because without sufficient data, banks are left to make educated guesses as to which factors to include and the overall weight of each of the factors in calculating future estimated losses.
As banks start the data collection process, they will quickly realize that, in the commercial line of business, data is stored in Microsoft platforms (Word, Excel, Access), financial statement spreading platforms (Moody’s, Baker Hill), ticklers in core accounting system, and in paper files.
This makes it virtually impossible to export the data into a single platform or data warehouse. To make matters worse, banks need to link the information not only to individual borrowers but to related entities (such as guarantors, owners, and affiliated companies).
In order to incorporate the model into the daily operations, banks need to associate the CECL model with a commercial loan origination platform. That’s no easy task, either.
Loan origination platforms differ greatly in terms of functionality regarding CRM, statement spreading, loan documentation, and other functions. CECL adds another level of required functionality.
Challenge the Vendors
Banks need to challenge their vendors to develop capabilities for CECL data collection, modeling, and analytics. Banks undergoing commercial loan origination system selections should make sure CECL is part of the requirements. For banks not planning on a new commercial LOS, 2016 might be a good year to start the process given this could become the foundation for all things CECL.
The worst thing a bank can do about CECL is wait. Don’t wait for CECL to become adopted; don’t wait for some point in the future where you might have more capacity. Don’t wait for some miracle technology platform to appear that will suddenly have all the data you need. Open your eyes and get started on the hard work of data standardization and model development! -jp