Posted: Mar 6, 2018
Categories: Regulations, Consulting
Comments: 0

ThePaymentsReview continues a new feature that occasionally highlights regulatory topics important to credit unions.

Accounting for loan losses is at the heart of credit union accounting. Setting aside reserves for loan losses is an important accounting component, but an increase in allowances reduces a credit union’s capital. Under current accounting standards, a credit union recognizes losses when they reach a probable threshold of loss. This is called an incurred loss accounting model. In practical terms, incurred loss accounting is a backwards-looking model, measuring a pool of loans against historic annualized write-offs. This method can drastically underrepresent potential future losses when a loan portfolio is exposed to a financial crisis, especially after a run of several years with lower losses. And this is exactly what happened following the financial crisis of 2008 in which some credit unions found themselves under reserved and unprepared for losses in their loan and mortgage portfolios while losses to their investments, and in many cases, shares declined. In the rising economy of the early 2000’s, losses were not being accounted for as “probable”.

In response (and in hindsight) the Financial Accounting Standards Board (FASB) approved a new standard in 2016, which becomes mandatory for all U.S. financial institutions in 2021. Instead of the “incurred loss” model which looked at losses in terms of “probable”, and is open to interpretation, the new standard is based on “expected loss”. In this model credit unions need to estimate lifetime credit losses on day one, and those losses don’t have to be “probable”. This change has the potential to increase allowances for losses to increase. This increase was estimated to be as high as 30-50% when first discussed in 2015, but as forecasts of future economic conditions brighten, most analysts have lowered that number. Although it is not required until 2021, early adoption of the new standard is permitted beginning in 2019, allowing for credit unions to phase in the changes in a more orderly manner.

The new standard is called CECL, Current Expected Credit Losses, and is one of the most significant accounting changes affecting credit unions. Not only will this cause allowances for losses to increase in most cases, but the incremental costs to calculate CECL estimates may be higher as well.


A new way to do business


Credit unions currently have loan origination criteria. But with CECL, decisions made at loan origination will immediately impact reserves. Borrowers that present a lower credit risk will mean less allowance for loan loss at origination, increasing the relative amount of deployable capital. And given that the longer the contractual term of the loan, the higher the probability of loss, lending officers may provide incentives for borrowers to take shorter term loans to reduce the credit union’s CECL allowances. Most importantly, credit unions that are better at monitoring loss expectations by credit rating and term will be able to better price these factors into their loan products over time. To accomplish this means retaining the data, starting now.


A data warehouse is now necessary


The changes required by CECL calls for a much deeper level of modeling, analysis and reporting than what was previously required. To determine what is acceptable allowance for loan and lease losses, and report and defend compliance to auditors and examiners, it is necessary to retain and analyze loan origination data. Currently, credit unions track historical data such as loans that have been paid off and annual charge-off-rates, making the data stored very manageable; even by using simple spreadsheets.

For CECL, more data, over a much longer period, is needed, such as:

  • Data elements collected at loan origination including credit scores, residential real estate values, initial risk rating, segmentation for the loan, and loan-to-value ratios.

  • Dates associated with the loan including renewal and maturity dates.

  • Ongoing data elements associated with each loan including loan balance, current risk rating, current fixed versus variable rates, current collateral values, changes to credit risk status (including delinquency, credit ratings, nonaccrual status), and any charge-offs and loan recoveries associated with the loan (partial and full).

  • Forecasts associated with the loan including forecasts of the local real estate market, unemployment rates, and other indices that examiners will be looking for.

In addition to storing and analyzing the data, many roles within the credit union will either be sources or receivers of data elements. Underwriting, origination, loan processing, modification, loan review, forecasting, and real estate all need to be in sync and “audit ready”, requiring all data associated with each loan be available and reportable.


With Great Flexibility Comes Great Challenges


CECL does not specify which model should be used, giving each credit union the flexibility to choose the analytical model that works best for their organization. While many credit unions may see this flexibility as a positive aspect of the standard, it does create challenges. Credit unions need to test each model to determine which is best for their credit union. This requires resources and knowledge that a credit union may not have in house.

There are several models and each presents tradeoffs. For perspective, the graph below shows 6 different CECL reporting paradigms across the top, and criteria associated with CECL reporting. The Historic Average model has low accuracy (red dots), both short term and long term, which results in increased allowances for loan losses (CECL Estimate), but also is low complexity and minimal computation, making it more robust, and easy to justify to auditors. On the other extreme, the APC Score model has high accuracy resulting in lower required loss allowances, but is higher in complexity and computation.

CECL Model Comparison. Source: Deep Future Analytics, LLC


Steps for credit unions to prepare for CECL


Get educated on CECL. There are many consultants who are claiming specialization in CECL. Some have the credentials, and some are merely capitalizing on a very complex and stressful change. The best defense is for the credit union’s management team and board of directors to become educated, independently, on the impending changes before hiring, outsourcing, purchasing or making decisions.


Start collecting the data now. The process of collecting and storing the historical data elements denoted above can be exhaustive and daunting initially. But once a process is put in place, and credit union staff appreciates why it is necessary, it becomes easier over time. Most credit unions are collecting at least a subset of the data elements that will be needed for compliance. However, it remains a potentially overwhelming task to validate and correlate the data; the process of “normalizing” the data can take months or longer. Credit unions should not waste time delaying this important first step. After establishing a process to collect the data, it’s time to decide how to store and retrieve it.


Pick a data warehouse solution. For years, many credit unions have depended on Excel or Microsoft Access for a data warehouse, and this has sufficed to date. But the complexity of reporting on the data, and justifying compliance to auditors, necessitates a more sophisticated method. There are two important components to look at – the data warehouse layer and the reporting layer, which can also include a data visualization tool. There are many commercial solutions that do all of these, usually with some compromise to one or all layers. There are many excellent data warehouses, and separately there are many great reporting tools and data visualization tools. For a credit union that hasn’t already embarked upon the search for a data warehouse, consider a solution from a CUSO that specializes in serving the specific needs of credit unions. While the data will not be needed for allowances for loan loss for a year or two, building up the history of detailed, normalized data provides flexibility in the next step, exploring, testing, and choosing the right model.


Pick a CECL model. As explained above, there are several models, with tradeoffs on complexity for accuracy, all of which can impact the amount of working capital. One approach is to test each model based on existing data to determine its impact. This is an iterative process as more data is collected, and more knowledge is gained. The credit union’s board of directors should be kept apprised; they may also offer their input based on experience from similar efforts at other credit unions. At the same time, share the outcome of testing efforts with auditors to discuss appropriateness for the financial institutions.


What’s most important is to start now. The more time a credit union has to understand what it faces and what its options are, the better their decision will be.

Rate this article:
No rating
Dave Chojnacki

Dave ChojnackiDave Chojnacki

Dave has more than 10 years of relationship management experience and holds a Credit Union Compliance Expert (CUCE) Designation from CUNA. He currently serves as Director of Consulting Services at Trellance.

Other posts by Dave Chojnacki

Full biography , Contact author

Please login or register to post comments.


Featured Stories