CECL Implementation: Where Banks Are Now and Where They Need to Be

By Ryan Barrow and Carrie Connell

CECL, or the Current Expected Credit Loss model, is designed to improve how institutions that issue credit account for balance sheet reserves and potential losses. Today, bankers use an “incurred loss model” that requires the recognition of credit losses on loans when it becomes probable the contractual amounts due will not be collected. With CECL, however, bankers calculate future losses using an “expected loss model” that considers forward-looking information such as current economic conditions and reasonable and supportable forecasts. Overall, CECL will result in the recognition of lifetime expected credit losses immediately when a financial asset is originated or purchased.

While the new rule was issued in 2016, many in the financial services industry are still grappling with how to implement CECL without it becoming a burden on their overall operations and methodology. Not to mention, there is a time limit at play. Public business entities filing with the Securities and Exchange must adopt CECL for interim and annual periods in fiscal years beginning after December 15, 2019, while PBEs not filing with the SEC must likewise adopt CECL beginning after December 15, 2020. According to a proposal expected soon from the Financial Accounting Standards Board, all other entities will need to adopt CECL for fiscal years beginning after December 15, 2021.

While these dates may seem somewhat distant, when it comes to introducing such fundamental changes, it’s the tiniest details that can matter the most. It is expected that with the implementation of CECL, banks will be required to allocate additional capital for potential losses, which may force some banks to reconsider their capitalization and investment strategies. Few institutions are currently far enough into the implementation process for there to be a clear consensus, but for reasons like this, bankers must be prepared for CECL well ahead of their respective deadlines.

Where to begin: gathering data

The first step to any bank’s CECL implementation is gathering data on its loans and leases, which will then be used to group and categorize loans based on level of risk. This data can range anywhere from loan origination date to loan type, charge-off date, borrower credit scores, loan-to-value ratios and other relevant data points.

TOOLKIT > To access ABA’s many resources for bankers preparing for CECL, including ABA’s members-only CECL Network discussion group, visit aba.com/CECL.
This is the step where most banks are today. In a survey of nearly 50 financial institutions by our firm, Porter Keadle Moore, 80 percent of respondents were “gathering information” to implement CECL. This is similarly in line with the 53 percent of respondents who indicated they had “given CECL serious consideration, but had not put the wheels in motion,” and 37 percent who have made “significant progress.”

That’s not to say this is a poor position to be in, particularly for those banks implementing in 2021. On the other hand, for those with a 2020 deadline, the survey results indicate a need for additional urgency on their part. Because a bank’s loan data is scattered across their entire organization and typically not in a centralized location, it will take time and effort to compile and categorize it all. This is especially true for those larger organizations with non-homogenized or uniform loan types. For institutions that issue consistently similar loans, the data gathering process should be a much simpler task.

The difficulty of the data-gathering process is also one of the reasons why implementing CECL is a team-centric effort. While PKM’s survey indicated that 20 percent of respondents created a CECL implementation team, it’s likely this number will rise as the implementation date approaches. Keep in mind: CECL is not solely an accounting issue, and the bank’s team should include lending, IT and even risk management representatives.

Step 2: Parallel models and methodology selection

After a bank compiles its loan data into a central repository, it must then create various loan pools to categorize them based on level of risk. This is where methodology selection comes into play. CECL does not specify a single, individual type for measuring expected losses but rather leaves it up to the institution to select a well-documented and verifiable estimation method that can be applied consistently over time.

During this step, banks should ensure their loans are segmented properly with accurate data that is as granular as possible, while still maintaining statistical significance. In doing so, banks give themselves additional flexibility in testing various methodologies to determine which approach best suit their needs.

After segmenting its loans, a bank should begin the process of running parallel models with its current allowance for loan and lease losses model. This will give the bank a clearer picture of what strategic changes to make. For those companies with an implementation deadline of 2020, parallel models should start being run at the latest in the fourth quarter of 2018, with a similar timeline for those implementing in 2021.

According to PKM’s survey, only 8 percent of respondents have run parallel models to this point. One respondent also indicated that its parallel model resulted in a 10 percent increase in ALLL.

What about third-party vendors?

Many bankers are considering whether or not they should contract with an outside vendor to implement CECL. The answer largely depends on a bank’s individual needs, but data shows that more and more banks are electing to go with a third-party solution. According to PKM’s survey, almost 45 percent of respondents were deploying a vendor product.

Implementing CECL does not require a vendor. However, few community banks have the time or resources to properly and efficiently manage implementation in-house. This is particularly true for the data collection process—with loan data stored in various digital repositories, there may tremendous value for organizations with non-uniform loan pools to engage a vendor capable of capturing this data in a central database and putting it to better use.

Then there is the issue of running parallel models and exploring or switching methodologies. Many banks still use spreadsheets for their ALLL calculations, and while this may continue for smaller institutions, for those that require more flexibility, third-party vendors may provide an efficient method to analyze methodology options.

The CECL deadline is fast-approaching, and bankers must begin implementing this new standard now in order to ensure future calculations are in-line with the bank’s expectations. While it’s widely believed CECL will result in an increase in the ALLL and a charge to capital upon adoption, with proper preparation and testing, banks will be well-prepared. This will all depend however, on the steps bankers take now to prepare for the impending change.

Ryan Barrow is senior audit manager, and Carrie Connell is senior risk advisory manager, at Porter Keadle Moore, an accounting and advisory firm serving public and private organizations in the financial services, insurance and technology industries.