Industry Insights

CECL: Is Time on Your Side?

September 2017
Author:  Gordon Dobner

Gordon Dobner

Partner

Audit

Financial Services

2800 Post Oak Boulevard, Suite 3200
Houston, TX 77056-6167

Houston
713.499.4600

The issuance of the Current Expected Credit Loss (CECL) standard in June 2016 represents a major change in how financial institutions will account for credit losses on financial assets carried at amortized cost, and most substantially, loan portfolios. The standard will require institutions to use an expected lifetime-loss model, rather than today’s incurred-loss model. The standard requires institutions to consider historical lifetime-loss experience, current conditions and reasonable and supportable forecasts. It’s expected institutions—specifically smaller and less complex institutions—should be able to leverage some of their current processes and systems. However, the inputs and assumptions to create an expected lifetime-loss rate model significantly differ from today’s incurred-loss rate model. Therefore, institutions will need time to make the appropriate changes to their current systems.

In addition, the standard isn’t prescriptive on what models or methods should be used to meet CECL principles. This will require institutions to decide what models and methods are reasonable based on the risk inherent in their own portfolios. To determine how much of their current process and models may be leveraged, institutions need time to understand how their current models will change under an expected lifetime-loss approach. Understanding other available models, and the pros and cons of all models, will be crucial in selecting what model(s) meet the risks and characteristics inherent in their loan pools. Institutions should have robust documentation supporting their ultimate model choices and how they’ve applied CECL principles. Institutions will need time and adequate internal and external resources to make sound, well-documented judgments.

The Financial Accounting Standards Board understood the implementation of this standard would take time and significant judgment. Therefore, it provided a longer lead time than normal for final implementation dates. Institutions should focus on developing an internal process that will provide ample time for successful implementation. BKD has developed the following high-level road map to help guide institutions through the implementation process:

Because each item will take time and is crucial to creating a sound, well-documented implementation process, institutions need to start the Planning & Readiness phase today to have adequate time to work through potential issues. This is especially important for data and documentation processes supporting model(s) considered based on the credit risks in the institution’s loan portfolio.

The rest of this article focuses on the pool segmentation and credit risk identification and data inventory and gap analyses phases, as these are the building blocks to a sound and flexible implementation. Tackling these steps early will be important to give the institution time for the necessary adjustments for model implementation best suited to its portfolios. Although this article won’t address all planning phases, creating a CECL committee, developing a timeline and educating critical personnel affected by the changes will be vital to all institutions, regardless of size or complexity. Institutions should have these phases completed or have actionable items for completion before 2017 ends.

Pool Segmentation & Risk Identification

One of the CECL standard’s key principles is the requirement to pool assets based on similar risk characteristics, unless an individual asset doesn’t share similar risk characteristics. Pooling should be based on similar risk characteristics and be aligned with how institutions monitor and manage credit risk in their institutions.

Examples of risk characteristics per the CECL standard include:

  • Internal or external (third-party) credit score or credit rating
  • Risk rating or classification
  • Financial asset type
  • Collateral type
  • Size
  • Effective interest rate
  • Term
  • Geographic location
  • Industry of the borrower
  • Vintage
  • Historical or expected credit loss pattern
  • Reasonable and supportable forecast period

Although institutions are pooling loans based on similar risk characteristics in their incurred loss models today, there are two critical reasons institutions need to re-evaluate their pooling characteristics under CECL. First, and most simply, documentation of current pooling and related similar risk characteristics is lacking. Many institutions select call report data, as it’s considered the most reliable or comparable (for peer purposes) characteristic. Second, there will be important loan characteristics in an expected lifetime-loss model that may significantly affect estimates not critical under an incurred-loss model. Consider characteristics such as payment structures and remaining life. These loan pool characteristics matter significantly under a CECL model, while they have little effect in an incurred-loss model because the measurement is only losses incurred as of today.

For example, if a loan pool today includes loans that substantially vary in contractual term, amortization, payment structure or fixed versus variable rate, those characteristics may materially affect the remaining life cycle and credit risk of that pool under CECL. Institutions also may need to consider further granularity in pools to break out loan terms, amortization structures (fully amortizing, balloon, interest only) or fixed versus variable rates. In addition, certain models work better for certain risks. For instance, if a loan pool contains mostly interest-only or balloon-payment structures in which principal reductions don’t reduce risk over the loan’s life cycle, then a vintage model wouldn’t make sense.

There’s a regulatory expectation that smaller, less complex institutions could use current pooling methodologies. However, there’s no current definition of “smaller and less complex.” Therefore, institutions should document current pooling methodologies and consideration of risk characteristics and their effects on estimated pool life and credit risk. A best practice would be to perform a historical analysis of trends or patterns for the most significant risk drivers and indicators for each pool to support pooling levels. This step is fundamental to a successful implementation, as it will provide documentation of the credit loss drivers and indicators for individual pools, which will then lead to preliminary model consideration. Without this information, it will be very difficult to perform a quality data inventory and gap analysis.

Also during this phase, institutions should attempt to calculate an estimate of individual loan pool lives. Under CECL, the remaining life is the contractual life adjusted for prepayments but excluding renewals and extensions, unless expected to be in conjunction with a troubled-debt restructuring. Different models have different needs regarding this assumption. However, at a minimum, institutions should perform an analysis to understand historical weighted-average contractual life and weighted-average remaining life of loan pools at various points in time. These would be compared to the actual material disposition of those pools, giving consideration to renewals and extensions and how they’re captured in the loan systems. System tracking of renewals and extensions of loans at maturity tend to be a consistent source of complication in CECL implementation. Tracking historic loan balances until they’re materially disposed of, performing attrition calculations or determining periodic prepayment rates are all exercises that should be considered. The conclusions that result from this process will directly affect the volume of historical data institutions may need. For example, if an institution is considering a loss-rate model and assumes the loan pool’s CECL life is approximately five years, it would need to go back five years or more to determine the historical lifetime loss rate.

Data Inventory & Gap Analysis

The purpose of the data inventory and gap analysis phase is for an institution to determine the data needs and availability for each potential CECL model being considered. In addition, an assessment of the completeness and accuracy of the data each model needs will be important. The ultimate result of this phase is to ascertain gaps institutions may have in their current and historical data and develop an action plan to overcome those data gaps. An exhaustive data inventory is essential to providing institutions with flexibility in their approach.

Institutions should consider core loan applications, including any report-writer interfaces, as a preliminary data source. However, there may be other areas where historical loan data is stored, including asset-liability management systems, current allowance for loan and lease losses spreadsheets or software and network drives. In addition, some institutions can restore data from backups to create a data warehouse. As loan applications and other historical data are queried, consider whether the system overwrites any historical information or inputs such as risk ratings, FICO scores, loan to value, etc., causing historical information to be unavailable.

As noted above, how systems capture renewal and extension information at maturity has been a consistent source of complication. Under CECL, a loan’s life ends at maturity whether or not it was renewed or extended, unless it’s a troubled debt restructure. Therefore, understanding how an institution’s system captures renewals and extensions is critical. For example, if an institution has a pool of nonowner-occupied commercial real estate loans with amortization periods of 15 to 20 years, but maturities of five years (related to the underlying tenant lease terms), the contractual life is no more than five years, but could be shorter when considering prepayments. If at maturity the loan is renewed for another five years, but a new loan or origination date isn’t created in the system, an institution could have more difficulty understanding the CECL life of that loan pool.

Overall, it’s expected almost all institutions will deal with some level of data gaps through the CECL implementation process, which makes it important to start this phase early. Examples of possible action items that may come from this phase include:

  • Change data retention policies and practices
  • Create a data inventory to overcome data gaps before implementation
  • Modify or increase the loan attributes captured at origination or on an ongoing basis
  • Modify controls over loan attribute input accuracy at origination or on an ongoing basis
  • Consider the need for external data where internal data is lacking
  • Remove certain models from consideration due to lack of available data

One common question we receive is, “What data should institutions be capturing historically for CECL?” There’s no perfect answer, as each model has unique data needs. However, many models build on similar data. The internal data institutions should inventory for historical availability and format usability includes, but isn’t limited to:

  • Historical periodic (monthly or quarterly) loan-level trial balances
  • Historical loan-level charge-off and recovery activity
  • Historical periodic (monthly, quarterly or annual) loan-level origination activity
  • If considering probability of default (PD)/loss given default (LGD) or discounted cash flow (DCF)—default events (must first define what a default is)
  • If considering PD/LDG or DCF—balance of individual defaulted loans at the default date
  • If considering DCF—periodic (monthly or quarterly) payment activity
  • Periodic (monthly or quarterly) funding activity on off-balance commitments

For each of the above items, institutions should aim to extract this data with all basic loan-level information. This includes payment structure and terms, pooling segmentation characteristics and key risk characteristics developed in the pooling segmentation phase. Lastly, institutions should determine sources for their forecasts based on key economic factors for each pool. Regional versus national sources and current sources used for economic data in their allowance calculations today should be considered.

Conclusion

Implementation of the CECL standard will take time and effort. With adequate time and appropriate internal and external resources, implementation can be managed without creating a significant strain. However, failure to appropriately plan and manage this process could place additional pressure on institutions’ internal resources. If you’d like to learn more about the CECL standard or how BKD can help your institution on the road to CECL implementation, contact your trusted BKD advisor.

BKD LinkedIn BKD Twitter BKD Youtube BKD Google Plus