The CECL Conundrum for Lenders: Which Loss Forecasting Methodology to Use?

By Regan Camp, Principal Consultant, SS&C Primatics

Vintage toned Wall Street at sunset, NYC.

The Financial Accounting Standards Board (FASB) issued the final Current Expected Credit Loss standard, or (“CECL”) on June 16, 2016. Many lending institutions are likely to experience a “shock to the system” because of significant changes to the end-to-end reserving process for financial instruments measured at amortized cost.

Unlike the incurred loss model, which is based primarily on historical loss information, CECL requires lenders to consider reasonable and supportable forecasts in the reserve estimate. The FASB does not specify a method for measuring expected credit losses; rather it is the individual institution’s responsibility to select a method or combination of methods to estimate expected credit losses.

Selecting a loss forecasting methodology will be one of the biggest decisions that lenders will make when transitioning to CECL. SS&C Primatics recently led a webinar, hosted by the American Bankers Association, to help lenders decide which method might be right for their portfolios. During the webinar, we discussed a number of methodologies that may be considered:

Open Pool Loss Rates: Lenders should be familiar with this methodology, since it is commonly used for incurred-loss models today. However, this method is heavily grounded in static historical data and will not be easily adapted to forecasting expected credit losses under CECL. Incorporating reasonable and supportable forecasts will entail significant judgment and qualitative adjustments in order to reflect a life of loan loss estimate. Our expectation is that many institutions will abandon this methodology during the transition to CECL, although it may remain popular for smaller institutions.

Vintage Analysis: This is a method of forecasting losses by analyzing historical losses of homogeneous loan pool sharing the same origination period or “vintage.” Performing a vintage analysis can be useful to the CECL estimate because the approach tracks losses over the entire life of the loan pool, from origination through payoff or charge-off. While this methodology will likely be popular (due in part to regulatory advice), questions remain about how to adjust the historical loss rates to take into account reasonable and supportable forecasts.

Migration Analysis/Roll Rates: This methodology leverages hindsight to predict losses based on historical movements between portfolio segments defined by risk criteria such as delinquency status or risk rating.  Migration analysis is a more granular approach than a traditional historical loss rate analysis and may offer a more predictive estimate of losses inherent in the portfolio.  This methodology is suitable for midsize institutions and scales well with bank size as there are highly sophisticated ways of generating forecasted transitions.

 Probability of Default/Loss Given Default: This methodology is used to estimate credit losses as a function of the exposure to credit risk by multiplying three components: Probability of Default (PD), Exposure at Default (EAD), and Loss Given Default (LGD). If lenders can accurately assign values to these variables, the math is straightforward: PD x EAD x LGD. The challenge will be in defining PDs in a manner that is appropriate for CECL.

Discounted Cash Flows: This is likely to be considered the “gold standard” for estimating expected credit losses because it replicates the expected collectability of contractual cash flows including the expected timing of losses and prepayments. The allowance is calculated as the difference between the recorded investment as of the balance sheet date, and the present value of expected cash flows discounted by the asset’s effective interest rate. The discounted cash flow method can be combined with other methodologies above (for example, it is very common in mortgage space to use forecast delinquency transitions to generate defaults that are used in the estimation of expected cash flows). Using cash flows requires more data to run the models and generally has higher system requirements than non-DCF methods. However, the benefits of using this approach may prove to be worth the added effort for some portfolios.

Identifying Practical Models for the Portfolio

Some of these methodologies are more appropriate for certain products types (for example, credit cards versus mortgages). We expect that lenders will consider multiple methods relative to the diversity of their portfolio. For a more in depth comparison of these loss forecasting techniques, listen to the full webinar. For a broader discussion of the operational impacts of CECL, check out our white paper, Preparing for CECL: Five Key Operational Challenges.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s