TwentyTwenty Analytics An Overview Dan Price, CPA, CFA President TwentyTwenty Analytics dprice@ttadata.com
What? Why? How? When?
New Credit Loss Methodology What is CECL? New Credit Loss Methodology Life of Loan Losses Leverages Existing Credit Management Practices Forward Looking Information The amendments in this Update require a financial asset (or a group of financial assets) measured at amortized cost basis to be presented at the net amount expected to be collected. The measurement of expected credit losses is based on relevant information about past events, including historical experience, current conditions, and reasonable and supportable forecasts that affect the collectability of the reported amount. An entity must use judgment in determining the relevant information and estimation methods that are appropriate in its circumstances. The amendments in this Update are an improvement because they eliminate the probable initial recognition threshold in current GAAP and, instead, reflect an entity’s current estimate of all expected credit losses. Previously, when credit losses were measured under GAAP, an entity generally only considered past events and current conditions in measuring the incurred loss.
What is CECL?
CECL Compared to Current ALLL What’s the same? Uses historical charge offs Pools homogenous loans Adjusts for Q&E factors Longer time horizon More historical data Greater potential for economic change In simple terms, think of it like this: If we never made another loan, and never funded an additional credit line except for those in which we’re contractually obligated to do so, how much would we lose before all of our loans went away? Expected credit losses of financial assets should be measured on a collective (pool) basis when similar risk characteristic(s) exist.
Why CECL? Current generally accepted accounting principles (GAAP) require an “incurred loss” methodology for recognizing credit losses that delays recognition until it is probable a loss has been incurred. Both financial institutions and users of their financial statements expressed concern that current GAAP restricts the ability to record credit losses that are expected, but do not yet meet the “probable” threshold. The global financial crisis underscored those concerns because users analyzed credit losses by utilizing forward-looking information to assess an entity’s allowance for credit losses on the basis of their own expectations. Consequently, in the lead-up to the financial crisis, users were making estimates of expected credit losses and devaluing financial institutions before accounting losses were recognized, highlighting the different information needs of users from what was required by GAAP. Similarly, financial institutions expressed frustration during this period because they could not record credit losses that they were expecting but had not yet met the probable threshold.
Why NOT CECL? Do you REALLY think CECL would have made a huge difference here? At about this point, credit unions would have started making the case that the worst was over and CECL could be adjusted to reflect an improving economic expectation (mean reversion) At this point, auditors would be telling you “You’ve been screaming that the economic condition would get better for four years now… No way”
First Mortgage Charged Off Other Real Estate Charged Off Why NOT CECL? First Mortgage Charged Off (Total, All CUs) Other Real Estate Charged Off Dec-2000 $10,775,622 $15,932,241 Dec-2006 $29,066,789 $45,201,879 Dec-2007 $38,286,581 $134,477,406 Dec-2008 $196,008,445 $577,306,288 Dec-2009 $551,709,267 $1,096,741,733 Dec-2010 $833,522,638 $1,200,422,589 Dec-2011 $956,274,054 $1,046,861,566 Dec-2012 $986,491,548 $828,740,092 Dec-2013 $633,735,148 $489,903,560 Dec-2014 $387,635,259 $287,985,737 Dec-2015 $260,151,461 $196,423,047 Dec-2016 $194,794,002 $148,491,694 For the eight years leading into 2008, credit unions charged off $154 million in first mortgages TOTAL. Then, in 2008, they charged off $196 million. The eight years following 2007, credit unions charged off $4.8 billion. The point here isn’t to explain that CECL is silly, it’s to drive home the concept that estimates can be gamed. CECL is more judgmental, which increases that potential. Your auditors are tasked with attesting to the fairness of your financial statements, but they’re still your responsibility, and you should commit to putting together the best representation of the standard as opposed to deciding what number is right up front and gaming your assumptions to hit that target. Source: Callahan
Why CECL? Data Analytics Performance / Profitability FASB Less Volatile Systematic Calculation FASB At worst, the Why? Is “because we have to” At best, it’s to create a less volatile calculation. A calculation that is data driven, and can be your stepping stone to building a more data driven culture at your credit union.
Timing Credit Unions are exempt from the definition of a Public Business Entity (PBE) and are not required to comply until 12/31/2021. The cumulative-effect adjustment is recognized as an adjustment to the beginning balance of retained earnings as of January 1, 2021.
Implementation Account 12/31/2020 Call Report 1/1/2021 CECL Effective Date 9/30/2021 Call Report 12/31/2021 Call Report Allowance for loan and lease losses (Current Method) $150,000 (A) $175,000 Allowance for credit (CECL) $200,000 (B) $235,000 (C) Cumulative-effect adjustment to the January 1, 2021, beginning balance of retained earnings $0 $50,000 (B – A) (NOT C – A) Net YTD Charge Offs $65,000 $85,000 Provision for loan and lease losses $90,000 Provision for credit losses $120,000 The following table compares the amounts reported by the institution in its Call Reports for December 31, 2020; September 30, 2021; and December 31, 2021 as a basis for illustrating the journal entries the institution would make to reflect the effects of adopting the new credit losses standard. Assume the institution records provision expense entries only as of quarter-end. Journal entries are easy, you can discuss them with your auditor to be sure you’re doing the right thing. You don’t have to memorize the mechanics here. The big takeaway is: You don’t have to present your CECL reserve until 12/31/2021, BUT You have to calculate it as of 1/1/2021 (12/31/2020), and that calculation will likely be audited as part of your 12/31/2021 audit This is something that, to my knowledge, has not been widely circulated/explained about CECL and you need to take note of.
Impact at Implementation ALLL Impact will vary Initial estimates indicate an increase in allowance for loan and lease losses (ALLL) of 25-50 percent for well-diversified portfolios. Longer average time to maturity (either resulting from younger portfolio or longer term) Greater Impact Strong CURRENT Economic Conditions Greater Impact
Impact to Provision (in 000s) Consistent Portfolio Current Balance ALLL - BoY Charge Offs Provision Expense ALLL - EoY Retained Earnings 12/31/2020 100,000 1,000 -1,000 1,000* 10,000 12/31/2021 1,500 9,500 12/31/2022 *Pre CECL Growing Portfolio Increased Risk
Probability-of-default How Determining credit loss Discounted cash flow Vintage / Static Pool Loss rate Roll-rate Probability-of-default Cohort The Final CECL ASU states in section 326-20-30-3: The allowance for credit losses may be determined using various methods. For example, an entity may use discounted cash flow methods, static pool, vintage analysis, loss rate methods, roll-rate methods, probability-of-default methods, or methods that utilize an aging schedule.
Create a Data Process Ensure data usable in future periods Automated (or scheduled) file creation Data reliable & standardized Ensured consistency going forward Captures relevant data Accessible data
Why Create a Data Process? I assume there will be acceptable methods that don’t require detailed loan level data. However, I also can’t see a scenario in which taking that easy out is in your long term best interest
Potential Data Fields – Loan Level Loan Identifier Current Risk Rating Borrower Name Date of Risk Rating Change Borrower Location Loan Type Origination Date Collateral Type Loan Balance as of Report Date Collateral Location Unfunded Commitment Original Collateral Value Renewal / Maturity Date Most Recent Collateral Value Loan Term Date of Most Recent Valuation Interest Rate Type of Collateral Valuation Fixed / Variable Original Debt Service Coverage Lien Position Original Credit Score Original Risk Rating Current Credit Score Loan Identifier – This should be truly unique to the loan record. Understand how the data comes out of your system. Is this ID truly unique? Am I generating multiple records for the same loan (e.g. if the loan is secured by multiple pieces of collateral)? Make sure you understand this for each system Loan Type – What field am I using to define my portfolio segments? There is a difference between a field existing in your system and the information within that field being valuable. If you’re not populating your loan system with information you need, change those processes today.
Potential Data Fields – Loan Level Date of Most Recent Credit Score Co-Borrower / Guarantor Name Past Due History Co-Borrower / Guarantor Location Past Due Status at Report Date Co-Borrower / Guarantor Original Credit Score SIC or NAICS Code Co-Borrower / Guarantor Current Credit Score Gross Charge-off Co-Borrower / Guarantor Current Credit Score Date Date of Charge-off Payment Frequency Gross Recoveries Payment Amount Date of Recoveries Loan Branch How does data come out of your system? Are you looking at all active loans only? All active loans and also those charged off? All active loans, those paid off and charged off? Best case If you don’t have a static pool, think about how you might be able to construct one (e.g. building using active loan files over various points-in-time, or supplementing data from your loan origination system)
Errors due to Model Risk Bad data Bad model Sound Practices for Model Risk Management (OCC Bulletin 2011-12) details the potential that financial institution may experience adverse consequences based on a decision reached by using a model, and it originates due to fundamental model errors or due to incorrect or inappropriate use of an otherwise sound model. https://www.occ.treas.gov/news-issuances/bulletins/2011/bulletin-2011-12a.pdf These errors may occur as a result of: Risk of Bad Data (Data / Calculation Accuracy) Data Mining Data Manipulation Risk of Bad Model (Foolish Assumptions) Data Analysis and Interpretation Using Conclusions to Improve Performance http://ttadata.com/model-risk/ Sound Practices for Model Risk Management (OCC Bulletin 2011-12)
Frequently Asked Questions Will the agencies require institutions to reconstruct data from earlier periods that are not reasonably available in order to implement CECL? No The following information are some of the highlights of the jointly issued FAQ by the financial institutions regulatory bodies on the New Accounting Standards – December 19, 2016 The entire document can be found at: https://www.ncua.gov/regulation-supervision/Pages/policy-compliance/communications/letters-to-credit-unions/2017/05-enclosure.pdf Question 26. Will the agencies require institutions to reconstruct data from earlier periods that are not reasonably available in order to implement CECL? [August 2017] Answer: No. The agencies will not require institutions to undertake efforts to obtain or reconstruct data from previous periods that are not reasonably available without undue cost and effort. However, an institution may decide it would be beneficial to do so to more effectively implement CECL. An institution may find that certain data from previous periods relevant to its determination of its historical lifetime loss experience are not available or no longer accessible in the institution’s loan system or from other sources. The institution should promptly begin to capture and maintain such data on a go-forward basis so it can build up a more complete set of relevant historical loss data by the effective date of the new credit losses standard or as soon thereafter as practicable. This is not to say that you should do nothing – If you don’t have good quality historical data, you can build the data going forward. If you need to have information from pervious periods, you can estimate it based on a reasonable methodology and relevant third-party information when it is available. CUs should promptly begin to capture & maintain such data on a go-forward basis so it can build up a more complete set of relevant historical loss data
Frequently Asked Questions How will the change impact a regulator’s assessment of my capital adequacy? It won’t 18. Will adoption of the new accounting standard impact U.S. GAAP equity and regulatory capital? [December 2016] Yes. Upon initial adoption, the earlier recognition of credit losses under CECL will likely increase allowance levels and lower the retained earnings component of equity, thereby lowering common equity tier 1 capital for regulatory capital purposes. For credit unions, implementation of CECL will impact retained earnings and will likely lower regulatory net worth. However, it will not impact the measurement under the NCUA’s risk-based capital rule that becomes effective in 2019. Under this new rule, the entire allowance balance will be reflected in capital for purposes of the new risk-based capital calculation. CECL will impact retained earnings and will likely lower regulatory net worth. However, it will not impact the measurement under the NCUA’s risk-based capital rule that becomes effective in 2019.
Facts, Timing, & Implementation Liability for Unfunded Commitments Under CECL, a liability for estimated losses on unfunded commitments should be calculated unless they are cancellable. Generally this does not include unfunded credit cards or lines of credit because they have the ability to unconditionally cancel. Liability for Unfunded Commitments Under CECL, a liability for estimated losses on unfunded commitments should be calculated using the contractual period for unfunded lines unless unconditionally cancellable. While on the surface this is concerning, generally a financial institution would not record an allowance for unfunded commitments on the unfunded credit cards or lines of credit because they have the ability to unconditionally cancel the lines of credit.
Frequently Asked Questions Should institutions use third-party vendors to assist in measuring expected credit losses under CECL? Not required – If done, follow agencies’ guidance on third-party service providers. The following information are some of the highlights of the jointly issued FAQ by the financial institutions regulatory bodies on the New Accounting Standards – December 19, 2016 The entire document can be found at: https://www.ncua.gov/regulation-supervision/Pages/policy-compliance/communications/letters-to-credit-unions/2017/05-enclosure.pdf Question 16. Should institutions use third-party vendors to assist in measuring expected credit losses under CECL? [December 2016] Answer: The agencies will not require institutions to engage third-party service providers to assist management in calculating allowances for credit losses under CECL. If an institution chooses to use a third-party service provider to assist management with this process, the institution should follow the agencies’ guidance on third-party service providers. Ultimately, the CU is responsible for maintaining the data, all the supporting information, and calculation of the expected losses. Given the amount of time to implementation it may be too early to engage a vendor solely on the basis of a CECL calculation. Right now, we believe CUs should be considering the models (internal or externally provided) and saving the data that they will need. In the market today, I have seen several vendors offering their products solely for the production of CECL estimates. These models can be extremely expensive to purchase and to get in place. Also, these models are generally based on one or two of the methodologies and the CU may ultimately decide on a different model. Ultimately CU is responsible for maintaining data, all supporting info & calculation of expected losses.
Vocabulary Disconnect Words you Won’t Find Cohort Static Vague Guidance Probability of Default Roll Rate This is meant to serve as a primer for your assessment of methodologies and best practices. Generally speaking, if you don’t understand what a colleague, vendor or practitioner is saying, ASK! Determining the best model for you and your portfolio is ultimately up to you. FASB doesn’t say you absolutely must use one of their listed methods, or the methods that they don’t cover in detail are less valid. Most likely it’s simply that the ones they illustrated were easier to illustrate
Vocabulary Disconnect Static Pool Analysis Vintage Analysis
TwentyTwenty Analytics Resources Interagency FAQ Transition Resource Group Landing Page IFRS 9 Implementation Guidance Deloitte Insights Landing Page Dan Price, CPA, CFA President TwentyTwenty Analytics dprice@ttadata.com (352) 634-0042 https://www.ncua.gov/regulation-supervision/Pages/policy-compliance/communications/letters-to-credit-unions/2017/05-enclosure.pdf