Skip to main content

Looking for Valuant? You are in the right place!

Valuant is now Abrigo, giving you a single source to Manage Risk and Drive Growth

Make yourself at home – we hope you enjoy your new web experience.

Looking for DiCOM? You are in the right place!

DiCOM Software is now part of Abrigo, giving you a single source to Manage Risk and Drive Growth. Make yourself at home – we hope you enjoy your new web experience.

Looking into the Shadows for CECL Clarity: Shadow Loss Analysis

April 28, 2018
Read Time: 0 min

On April 17, Chris Emery, MST senior advisor – engineering and director of special projects, walked webinar attendees through a discussion of testing and experimenting with potential CECL methodologies and the Shadow Loss Analysis feature of the Loan Loss Analyzer allowance automation software that streamlines the process. Following are some of the highlights of Chris’s presentation.

Regardless of when you’re going to implement CECL, there will be a time, which for most financial institutions is now or very soon, when you will have to run your incurred loss methodology – and potentially even continue to enhance or respond to regulator or auditor critiques – and test and experiment with expected loss methodologies. That’s a challenge: without having to double your efforts, how do you do both at the same time? 

Consider that scrutiny over the allowance is only increasing, so you will likely need to continue tweaking your current process. At the same time most institutions have a goal of running parallel CECL methodologies for at least four quarters. Some want to run parallel for five, six or more quarters, meaning that for the rest of 2018 and through 2019, SEC filers will be running parallel.   

CECL is principles-based, not prescriptive, so institutions have a variety of choices of how to implement CECL. You might choose to use different methodologies for different segments of your portfolio or different methodologies within a segment or pool, or even segment by different criteria to accommodate a methodology. 

You’re not going to come up with a single, clear choice of your methodology as you do your parallel testing. As you learn that a part or parts of a methodology won’t work for your institution, you’ll move on to test something different. You’ll more than likely evaluate various options before you find your best choice. You might even choose to combine a part of one methodology with a part of another to arrive at the CECL production methodology you use at implementation.   

After CECL implementation

Another concern in transitioning to CECL is what will happen beyond your CECL implementation date. Will the quarter before implementation be the last time you run parallel methodologies? Or is it something that has ongoing usefulness? Consider that FAS 5 was released in 1975, and the number of times you have changed your allowance approach since then – even in the last ten years, or five years. The allowance is a continually evolving process. We believe CECL will look much the same, evolving considerably in the initial years after implementation, but even thereafter as best practices shift and require you to refine your methodology, adjustment techniques and calculation methods. So you are likely to have a production CECL methodology and be running a separate parallel one, and that you will use different pieces from different approaches to get where you eventually want to be. Perhaps you don’t have data today to run the methodology you eventually want to use, so you have to choose an appropriate approach for day one, then work toward stockpiling data for a probability of default/loss given default (PD/LGD) or vintage production methodology. 

The point is that your CECL methodology is not a set-it-and-forget-it model. It needs to be continually refined and back tested. You might find a better methodology as you compile more data; it could be changes in the economy that are making your methodology react different from what you want. It will be a continually moving target.   

You will also likely want to use different methodologies for different categories of loans. For example, a bank has developed two parallels: a risk-rated cohort approach, much like loss migration with a tail for life-of-loan risk, as a starting point to move to a CECL methodology. That’s a small shift, a variation of the loss migration methodology they have been using for incurred loss estimates, tweaked based on loan type. And they elected to do a second parallel: a cohort might be an approach they could take if nothing else works – it’s something they know, are familiar with. But they are looking to something more sophisticated for different loan pools: a transition matrix for C&I and CRE to look at how risk ratings move over time, and as a transition to PD/LGD to get a lifetime expected default for those loans. Meanwhile they’re testing a vintage methodology for their consumer loans, looking at losses on a year-by-year basis from origination to establish a loss curve. Vintage wouldn’t apply to their CRE or C&I loans; unless you have truckloads of data, you’ll have a hard time getting the full vintage curve you want. But it might work for their large volume of auto loans. All in all, having decided on methodology for each loan segment, they will continue to run a parallel methodology for each segment to see if they get divergent answers, including transition matrix for all CRE loans combined, vintage for residential, and a delinquency-based cohort for consumer loans. They will continually back-test to see which is performing the most accurately over time.


Note: For more on CECL methodologies, read a recap of MST Advisory Director Regan Camp’s February 2018 webinar or access the webinar recording here

Shadow Loss Analysis: The Loan Loss Analyzer Solution

The Loan Loss Analyzer (LLA) Shadow Loss Analysis module addresses the need to test methodologies continually to ensure your production methodology is operating as expected. Once you reconcile your data, you can configure multiple parallel methodologies. We’re defining “methodology” as an aggregation, including quantitative and qualitative factors and economic assumptions, allowing for a mix-and-match CECL approach, as at the very least, there is going to be something in each loan pool requiring a different approach, however slight, due to a different type of risk, pool size or other factors.

Shadow Loss Analysis gives you the opportunity to use the same data but slice it in different ways to see how it acts under different scenarios. This allows you to understand what your total reserve will be under each scenario, subsequently enabling you to choose the most appropriate, advantageous methodology for your institution.

Webinar Attendee Questions Answered

Att: If you have had no losses in last five years, how do you prepare for CECL?

Emery: A five-year snapshot wouldn’t be a sufficient look-back period. The gold standard for the amount of data to establish a baseline historical loss is an economic cycle. Institutions without that much data have to make assumptions about what those losses might have looked like; some are using call report data to see what losses looked like. If you don’t have the data, you’ll use more judgmental, qualitative-type factors to determine your allowance for future expected losses.

Att: With CECL can I use different methodologies for different pool segments?

Emery: Yes, but you cannot use more than one production methodology for a pool segment. If loans in a pool don’t accommodate the methodology, they need to be removed from the pool and formed into or added to another pool, or addressed individually. The idea of shadow loss analysis is to give you an idea of how results will differ using different methodologies.

Att: In reference to the delay period under the cohort method, how long do you go back to establish a cohort for a reasonable judgment of total expected losses?

Emery: Usually you can get to a point where the cohort has become exhausted, which is a reasonable point from which we can determine expected loses. It is not always practical to have all the losses experienced to determine expected losses.

Learn more about Shadow Loss Analysis.

About Abrigo

Abrigo enables U.S. financial institutions to support their communities through technology that fights financial crime, grows loans and deposits, and optimizes risk. Abrigo's platform centralizes the institution's data, creates a digital user experience, ensures compliance, and delivers efficiency for scale and profitable growth.

Make Big Things Happen.