Solving the CECL Challenge
Presented by
Doug WrightCFO, Mission Federal Credit Union
andBryan W. Mogensen CPA, CLA (CliftonLarsonAllen LLP)
Doug Wright
• CFO, Mission Federal Credit Union
• 31 years in the credit union, banking industry
• 16 year CFO
• Member, FASB CECL Transition Resource Group
• MBA, Gonzaga U., BA, Univ of California, Berkeley
Bryan W. Mogensen, CPA• 25+ years working primarily with credit unions
• Speaker/Presenter at: – AICPA National Credit Union Conference
– ACUIA National Conference
– Various State League and Chapter Meetings
– CUES/CUNA
• Graduate of University of Wisconsin – Milwaukee, WI
• Member of: – AICPA
– Arizona Society of CPAs
• On the AICPA National Credit Union Planning Committee
Objectives
Provide CECL updates & timelines
Provide data & modeling recommendations
Tackle common CECL technical challenges
Evaluate CECL transitioning issues
Understand audit, back-testing & validation requirements
Consider business implications
4
Agenda• FASB & Regulatory Updates
• General Planning & Timelines
• Data Collection Insights
• Modeling Insights
• Buy vs. Build (& Maintain)
• Key Technical Challenges
• Transition & Post-Implementation Considerations
• Audit Insights
• Capital & Other Business Considerations
• Q&A
FASB Updates
TRG Meetings
– Potential topics provided by AICPA Depository Institution Panel, regulators, auditors, individual FIs, & FASB staff
– TRG Topics Discussed Include:• Non-PBE timing
• Reversion
• Purchased Credit-Impaired loans
• Troubled Debt Restructures
• Credit Card Receivables – estimated lives & loss estimates
• Recoveries
• Loan Extensions
Staff & Board Action
Regulatory Updates
• Bank regulators focused on CECL– Extensive internal discussion– Stronger guidance for large institutions
• CCAR & D-fast similarities• Expected modeling expertise/rigor greater
– Still struggling with smaller institutions• Data validity, model validity are issues• Balancing reasonable expectations and resource
constraints
• NCUA grappling with same issues
FASB Timeline Extension
Fiscal years after December 15, 2021 & interim periods within those fiscal years
Pushes back implementation to 2022; however:– 2022 interim periods required
CUs don’t generally issue interim statements; but– Form 5300s include ALLL, so effect could essentially
be adoption as of January 1, 2022 If required, a full year of 5300s before audit is
conducted…5300 restatement possibility NCUA may modify
CECL Implementation Plan & Timeline CECL Implementation Team - Done Developing basic CECL understanding - Done
– External resources (associations, vendors, other CFOs, etc.) Initial data evaluation – Done Data gathering & cleanup efforts – Basic done, but ongoing
through implementation Methodology/Model evaluation – 2018 - 2020 Buy vs. build – 2018 - 2020 Initial capital evaluation – 2018 - 2019 Parallel runs – 2020 - 2021 Implementation – 2022 Refinements – 2022 & Beyond
Data Collection Insights Data required depends on methodologies used
Roll rate Vintage PD/LGD Discounted cash flow
Many segments may not have statistically valid sample sizes Reduce sample size by more statistically precise techniques (multi-variate
regression, iterative processing) Sample size/lack of data can also be tackled with industry data
Economic cycle “expectation” presents interesting questions Valid data back to pre-2007? If so, is 2005/6/7 - current reflective of a reasonable cycle? How does this change in the next 3 years?
It’s likely that additional refinements will be required regarding data for the first 2-3 years after implementation
Data Collection Recommendations
Identify the methodologies you are most likely to use Identify critical inputs required:Loan specific infoEnvironmental & economic info
Do your best with historical data, but develop capabilities to retain & evaluate required data on a go-forward basis
Collect as much “useable” info as possible on a loan-specific level
Identify how you are going to overcome small sample sizes
Modeling/Methodology Insights
• FASB committed to allowing flexibility, but largely “uninvolved” post-implementation
• Regulators profess flexibility, but may gravitate to more common approaches Field examiner knowledge Institution comparability Vendor concentrations
• Different methodologies may yield “better” results for different segments
Modeling/Methodology Insights
Model “platform” should include areas to input & describe:o Historical basiso Current adjustments: qualitative & environmentalo Reasonable economic forecastso Reversion to historicalo Other key assumptions: prepayments, contractual lives, extensions
Model platform built with future in mind; how is the program going to be maintained & updated?
Discounted cash flow specifics:o Usually based on another methodology (e.g. PD/LGD)o Given other factors equal, will often result in lower estimated allowance
because future losses are discounted
Modeling & Methodology Recommendations
• Explore industry whitepapers/webinars on specific methodologies• Engage with several vendors to view approaches• If building one internally, consider:
– Platform requirements as noted previously– Ongoing maintenance from quarter to quarter– Validity, backtesting & “auditability”– Regulatory “defense”
• Applicability for other uses– Credit management– Credit stress testing– Loan/relationship pricing
Buy vs. Build Insights Vendors have committed significant resources to developing
models Platforms are still under development Most have worked with auditors, but none has been audited Pricing varies, depending on model complexity and institution
asset size Build extends far beyond initial platform development:
o Ongoing collection of data, maintenance & updating of model for each quarter, and model refinement
o Validation and “auditability”o Regulatory scrutiny
Buy vs. Build Recommendations
Contact and consider several vendors Beware of “CECL Compliant” claims Look closely at validation/explanation capabilitiesConsider cost vs. “performance” tradeoffsDon’t rush into a decision If build, note prior insights If build, work closely with auditor, regulator and
other independent parties for feedback
Technical Challenges
1. Credit cards
2. Loans formerly known as TDRs
3. PCI loans
4. Debt securities
Credit Cards
Unfunded credit lines No allowance if unconditionally cancellable
by issuer• Provision for unconditionally cancellable
• Most CUs have this provision
If not cancellable, record liability for unfunded commitments
Credit CardsEvaluate portfolio into:Transactors
• Pay off monthly
• Earn rewards
• Lower charge-offs
Revolvers• Retain monthly balance
• Most net interest income earned by these
• Higher charge-offs
Credit Cards
• Evaluate weighted average life– Payments based on activity (FIFO)
– Payments applied based on CARD Act• Higher APRs paid first
– Either method acceptable
• Determine loss emergence period– Period from loss-triggering event to charge-off
Troubled Debt Restructures (TDRs)• Individually impaired concept removed from ASU
2016-13
• TDR concept retained
• Initial indications – TDR credit losses measured using various CECL models
• But, FASB indicated that allowance must still consider full “economic loss”?
• Discounted cash flow measures economic loss
• Pending further clarification
PCD/PCI Loans
• Purchased Credit Deteriorated – loans that have incurred “more than insignificant deterioration since origination”
• More loans qualify as PCD compared to Purchased Credit Impaired (PCI)
• Discounts associated with credit loss added to reserve upon acquisition
PCD/PCI Loans
• PCD adjustments applied prospectively– Adjust amortized cost basis of loan to reflect
addition of the ALLL
• Not required to reassess current PCI loans meet definition of PCD
PCD/PCI Loans
Can elect to maintain pool of current PCI
Noncredit discount/premium, after ALLL adjustmento Accreted to interest income using interest method
on effective interest rate after ALLL adjustment for credit losses
Debt Securities
AFS:• Replaces OTTI impairment concept
• Determine if decline in fair value due to credit or other losses– If credit – post as allowance through income
statement (provision)
– If other (interest) – post through OCI
Debt Securities
AFS Use CECL model
Use an allowance instead of direct write-off (permits reversals)
Credit risk – post as allowance through income statement (provision)
HTM Treatment similar to loans held in portfolio
CECL reserve
Debt Securities
• Complex models not expected with plain vanilla portfolios
• For those with credit risk, requires reserve– Some municipals, private-label MBS
– Reassess allowance monthly
– Can reverse credit allowance up to zero
– Cannot have negative credit loss
Transition Considerations
Model and ALLL align with loan growth• Loan and related PLL booked on Day 1
Forecasting/Q&E• Should align with your ALM models
Transition Considerations
• Possible effects to:– Timing of loan promotions
– Loan discounts/offers
– Defer all loan origination fees and costs• Direct and indirect
Transition Considerations
• Net income challenges?– If book loan and PLL in Q1 then have
entire year to recoup income through loan interest
– If book loan and PLL in Q4 then have more PLL in CY and lower loan interest
Post-Implementation Considerations
Disclosures Back-testing & Validation Auditor Preparation & Testing
Disclosures
Detailed methodology & assumptions explanation
Credit quality “vintage table” (now likely including chargeoffs/recoveries by vintage)
Explanation of changes from prior periodo New loan volumes
o Historical factor changes
o Q factor changes
o Economic forecast changes
Back-Testing and Validation
Possible scenarios:– During testing phase
• Back-test charge-offs vs. projections
• Expected vs. actual correlations
• Ensure your model works as expected
– Use of another third-party vendor• Annual validation
Auditor Testing
• DIEP is drafting audit guidance practice aide– Currently 50 pages long
Auditor Testing
• Current issues being discussed:– Data challenges
• Current data – is it coded properly
• Old data – not all key data points captured by system
– Industry data• How can you audit/rely peer data when no access to
underlying data
Auditor Testing
• Reliance of a specialist– If you use third-party software model
– What kind of report can auditor use to rely on
• Not a SOC 1
Auditor Testing
• Risk based pools of loans– Test of controls over loan coding at origination
• Contractual terms– Test of controls over loan terms at origination
Auditor Testing
• Forecasts/Q&E factors– Quantifying forecasts
• Varied opinion by other institutions
• Varied opinions on where we are in economic cycle
• Who is correct
• Can both be correct
– ALLL Q&E align with ALM models• See auditors reviewing both to determine if they match or
align
Auditor Testing
• Reasonable and supportable estimate– Documentation must be enhanced to support
conclusions
• Testing of assumptions– Use of a specialist
– Use an internal/external validation model
Auditor Testing
• Data validation– Enhance tests of controls at origination to ensure
loans coded into correct segments, etc.
– Enhanced data analytics used to test portfolio
– Varied opinions on loan pools with similar risk characteristics
Capital & Other Business Considerations
• Regulatory Capital Impacts– Net worth/assets
– Total risk-based capital
• Consider earlier modeling, if:– Sub-prime concentrations
– Longer-lived assets
– Relatively close to regulatory net worth/asset minimum (9% or below)
• Cyclicality− Financial crisis− Recession in early
2020s?• Applicability for other uses
− Credit management− Portfolio concentrations− Credit stress testing− Loan/relationship
pricing
Any Questions?
If you have any questions regarding these guidelines please contact Josie Collins
703/842-2275