Current Expected Credit Loss (CECL)
HOW WILL CECL CHANGE THE ALLOWANCE FOR LOANS LOSSES CALCULATION?
Under current accounting standards, lenders use two to five years of historical net charge-off data to calculate an average annual net charge-off percentage which is intended to represent the expected net charge-offs to be incurred by the lender in the following 12 months on their performing loans. This portion of the allowance for loan losses (“ALL”) calculation is referred to as the FAS 5 portion of the ALL calculation. Many lenders also call it the general reserves within the ALL calculation.
CECL will require institutions to set aside an ALL accrual to cover the expected net charge-offs to be incurred over the estimated life of these nonimpaired loans. For most institutions, that means setting aside an ALL amount to cover the net incurred losses over a period of three to seven years depending on the type of loan.
CECL also changes the accounting for “problem” loans or loans where full repayment is in doubt. Under current accounting standards, a specific reserve is needed for any impaired loan when the unpaid balance of the loan exceeds either the net realizable value (NRV) of the collateral pledged against the loan (for collateral dependent loans) or the net present value (NPV) of the future cash flows expected to be received on the loan. A loan is considered impaired if it is more likely than not1 that the borrower will default resulting in the institution not collecting the total unpaid balance (principal and interest) due on the loan. A loan is collateral dependent if it is more likely than not that the institution will be repaid solely from the liquidation of the collateral pledged against the loan.
Under CECL, a loan is considered impaired when it is probable that the borrower will default. To reach this probable threshold, it must be likely that the borrower will default. In practice, that means the lender has started the foreclosure process (or soon will) and/or the institution has decided to charge off the loan. This will be a change from current accounting standards because most watch, special mention, and substandard loans will remain in the general CECL calculation instead of being subject to the specific reserve calculation. The specific reserve for CECL-impaired loans is calculated by comparing the unpaid balance of the loan to the net estimated liquidation value of the collateral.
The total ALL recorded on an institution’s balance sheet is equal to the total of the FAS 5 ALL (current methodology), or ALL on nonimpaired loans (CECL methodology), and the total of the specific reserves assigned to impaired loans.
 The phrase, more likely than not is accepted to mean that there is an over 50% chance the borrower will default.
WHEN IS CECL EFFECTIVE?
Banks that file with the Securities and Exchange Commission (SEC) and are considered a public business entity must implement CECL during their fiscal years beginning after December 15, 2019. CECL will become effective for all other institutions during their fiscal years beginning after December 15, 2020. There has been a recent proposal by the Financial Accounting Standards Board (“FASB”) to delay the effective date for non-public business entities until fiscal years beginning after December 15, 2021 but, as of the date of this article we are unsure if this proposal will be approved. Nevertheless, regulators expect all financial institutions to know how CECL will impact them well before the effective date of the standard.
Examiners expect non-public business entity banks and credit unions to have a plan on how they intend to implement CECL by year-end 2018. They further expect that you will be running side-by-side comparison calculations of CECL and the ALL under your current methodology starting in 2019. They will then expect your institution to finalize the methodology you will use in 2019 or early 2020 so that you are ready to go live when CECL becomes effective. Banks that file with the SEC will be required to disclose the expected impact of CECL in the footnotes to their audited financial statements starting with year-end 2019.
WILL CECL CHANGE ACCOUNTING FOR TROUBLED DEBT RESTRUCTURED (TDR) LOANS?
Yes. Under current standards, TDR loans are assigned a specific reserve which is determined like the specific reserves calculated for impaired loans under current accounting standards. Under CECL, these loans will be included in the general ALL calculation. The amount of TDR loans will still need to be disclosed in audit and call reports, so lenders will need to continue to identify and track these loans. Some institutions may also elect to treat TDR loans as a separate category in the general CECL ALL calculation.
WILL CECL APPLY TO CREDIT UNIONS AND SMALLER BANKS?
Yes. CECL will apply to all banks, credit unions and other lenders offering loans with an estimated useful life greater than one year. The Financial Accounting Standards Board (FASB) approved CECL in 2016 after a four-year evaluation. The Office of the Comptroller of the Currency (OCC), the National Credit Union Administration (NCUA), the Federal Reserve and the Federal Deposit Insurance Corporation (FDIC) have all publicly supported CECL. Industry-related organizations have attempted to persuade the FASB to exempt credit unions and smaller banks (with $1 billion or less in total assets) from CECL but their efforts have been unsuccessful. Some of these groups even reached out to Congress for help but, Congress chose not to debate the matter. It’s time for all lenders to focus on implementation versus hoping this standard will go away.
WHAT IMPACT WILL CECL HAVE ON MY ALL CALCULATION AND CAPITAL POSITION?
Many industry experts and regulators estimate that CECL will result in a 15% to 35% increase in ALL balances. Although the impact is likely to vary, it is advisable to estimate expected ALL balances sooner rather than later, so you can adjust your future growth and dividend plans. It is equally important to start considering the impact of CECL in the pricing of longer-term loans (loans that will still be outstanding when CECL becomes effective). I expect that CECL will require lenders to increase the interest rates they charge borrowers to cover the additional ALL needed on outstanding loans.
ARE REGULATORS PLANNING TO ADJUST REGULATORY CAPITAL CALCULATIONS FOR CECL?
In May 2018, the FDIC, OCC and Federal Reserve Bank issued an interagency statement that proposed allowing financial institutions to elect to “phase-in” the impact of the initial CECL adjustment in their regulatory capital calculations over a three-year period. Under this proposal, institutions could increase the amount of regulatory capital in their regulatory capital calculations by 75% of the initial CECL adjustment in the first year that CECL becomes effective. In Year Two, institutions could increase their regulatory capital by 50% of the initial CECL adjustment and in Year Three they could increase their calculated regulatory capital by 25% of the initial CECL adjustment. After Year Three, no further adjustment to regulatory capital will be allowed. The NCUA followed suit with a similar proposal.
It is quite likely that this proposal will become effective which will bring relief to many banks and credit unions. The comment period for this proposal ended July 2018 so I expect a final decision soon.
When you implement CECL, the Day One adjustment will be record using the journal entry:
- Retained Earnings – DR
- Allowance for Loan Losses – CR
The proposed phase-in of CECL will apply to the Day One adjustment only. Subsequent changes to the ALL under CECL will flow through the income statement as a debit or credit to the provision for loan losses and a corresponding debit or credit to ALL. CECL will result in more volatility in an institution’s reported earnings so it’s important to educate your board of directors, shareholders or members, and senior management to this fact now.
HOW IS CECL CALCULATED?
The CECL ALL calculation starts by separating the loan portfolio into segments. Each segment should have similar risk characteristics such as the term offered, type of collateral pledged against the loan and have been underwritten using similar criteria (e.g., maximum loan-to-value ratio, minimum required debt service coverage ratio, etc.).
More is better when deciding how many collateral types to use in the CECL calculation because:
- It helps to demonstrate that you have adequate diversity within your loan portfolio, allows senior management and the board to better understand the types of projects the institution is involved in, and could provide you with valuable insights into future opportunities. For example, it could highlight that you don’t currently offer a certain loan type that is in high demand in your market.
- It helps to isolate losses to a lower balance of loans. Assume that your institution incurs abnormally charge-offs in land development loans. If you considered all commercial real estate (CRE) loans as one segment, you would be projecting those losses onto all CRE loans held by your institution. On the other hand, if you separated this portfolio into 10 to 12 different segments and had loans in each loan type, you would be isolating these bare land charge-offs to a much smaller loan balance which in turn, would reduce the amount of the calculated ALL and help preserve much-needed capital. I would not recommend a model that only used a few collateral types. Although those models are less complex to use and typically cost less, they could hurt your institution down the road by causing you to report a higher ALL balance. This in turn could lead to regulatory action being taken against your institution or prevent your institution from growing because you have less capital. These lower cost models offered by some vendors sound good on paper but, they don’t provide their users with valuable information on risk within their loan portfolios, make it almost impossible to accurately project net losses to be incurred or prepayment percentages, and don’t provide any meaningful support for the subjective adjustment portion of the CECL calculation (see discussion below).
- To fully understand the inherent risk in your loan portfolio, you must understand who and what types of projects or assets you are lending on. It is advisable to break down your loan portfolio by as many segments as you can to understand the weighted average yields, weighted average risk ratings / FICO scores, average loan-to-value ratios, and geographic concentrations for each loan type. What you know, you can effectively manage!
The next step in calculating the ALL under CECL is to determine the historical charge-off percentages to be used over the estimated remaining life of the loans within each segment. The table below illustrates this concept using a small portfolio of used auto loans with a total unpaid balance of $10,000,000 as of December 31, 2018.
Further assume the following for this portfolio:
|Year loans were originated||Unpaid balance||Weighted estimated remaining average life of loans per the vintage year|
Further assume the institution’s historical net charge-off rates (C/O’s) were as follows:
|Year net charge-off incurred||Net C/O rates – Loans originated within 12 months or less||Net C/O rates – Loans originated within 13 to 24 months||Net C/O rates – Loans originated within 25 to 36 months||Net C/O rates – Loans originated over 36 months|
|Average Net C/O Percentage||0.23%||0.76%||0.51%||0.23%|
Ignoring prepayment calculations and subjective adjustments discussed below, the CECL-calculated ALL for these loans would be:
|Year loans were originated||Unpaid balance||Weighted estimated remaining average life of loans the given year of origination(A)||Years to use in CECL calculation||Estimated net C/O % over the life of the loans (B)||Calculated CECL ALL
 Institutions may choose to use weighting in this calculation so that more recent years have a greater influence on the charge-off calculations. In this case, simple averages were calculated.
 To illustrate how these were calculated, for loans originated in 2019 that have weighted remaining average life of 38 months, the first-year percentage of 0.23%, the second-year percentage of 0.76%, the third-year percentage of 0.51%, and the fourth-year percentage of 0.23% were totaled to arrive at the estimated average losses over the loans of 1.73%.
|2018||$4,500,000||38 months||1, 2,3, 4||1.73%||$77,850|
The calculation which separates the loan portfolio by loan type then year of origination or vintage illustrates the CECL calculation which is commonly referred to as the vintage method. Another approach is to segment the loan portfolio first by loan type, then by vintage and finally by risk rating for commercial loans and FICO credit score for consumer loans. This approach is commonly referred to as migration analysis. Another approach is to segment the loan portfolio by loan type and vintage year and then break the historical net charge-off percentage calculation into two components: probability of default and severity of loss once default occurs. This method is referred to as either the probability of default / severity of loss (PD/SL) method or the probability of default / loss given default (PD / LGD) method.
Regardless of the method used, the unpaid balance of the loans used in the CECL calculations must be adjusted for both expected prepayments for term loans and additional advances that can be utilized by borrowers for line of credit loans.
Borrowers often pay off their loan prior to maturity by making additional principal payments or refinancing the loan with another lender. Therefore, the unpaid balance of term loans should be reduced by expected prepayments. The prepayment percentages are calculated using the formula:
Prepayment percentage = Total principal payments received less contractual principal payments expected to be received / Unpaid principal balance of the loan
These prepayment percentages need to be calculated over the estimated useful life of the loans much like the net charge-off percentage calculation demonstrated above.
CECL calculations must also be adjusted for line of credit loans (LOC). In this case you can either calculate the weighted average principal balance outstanding for each vintage year in the estimated life of these loans or, assume (like we did in our model) that the fully extended balance of the LOC was outstanding in the CECL ALL calculation. This approach is more conservative and considers the fact that borrowers who are experiencing financial difficulties often maximize the amount they borrow on their LOC.
CECL ALL calculations are required to be forward looking and consider economic and other conditions that may affect the level net charge-offs to be incurred in the future. Institutions should adjust the historical loss percentages for each segment for subjective adjustments. This is done the same way it is done today under the current ALL methodology. The most common subjective adjustments used are:
- Expected changes in the economy
 This assumes an institution doesn’t risk rate its consumer loans.
 The fourth most common method is the net present value of discounted cash flow method (NPV). This method is not discussed here because most institutions will not have the necessary information.
 Our model uses the unpaid principal balance of credit card and home-equity lines of credit loans because most lenders restrict future advances on these loan types when a borrower’s credit score declines or the borrower becomes delinquent on their loan.
- Expected changes in the real estate market (only affects loans collateralized by real estate)
- Actual or expected changes in lending policies
- Changes in the experience or expertise of lending staff
- Changes in the volume, delinquency status, or the relative dispersion of the risk ratings / FICO credit scores within a loan portfolio. In other words, are there trends within your loan portfolio that may affect the relative inherent risk within that portfolio?
These adjustments will require supporting documentation which is likely to be more extensive than what occurs in current practice.
WHAT DATA IS NEEDED TO IMPLEMENT CECL?
I have attended over 20 different CECL seminars offered by accounting firms, CECL software vendors, consultants and regulatory bodies. The question about data is answered in a variety of ways. Some organizations list everything conceivable to ensure they don’t leave anything out while others state they will cover it all and no change in practice is required.
The good news is that you probably already have the data you need to calculate CECL in your current loan subsidiary ledger (with the possible exception of the additional information needed to calculate prepayment percentages).
In order to use your institution’s historical data to calculate CECL, you will need the following information:
- Customer / member number
- Loan number
- Loan type, including separating commercially-oriented and consumer loans
- A separation between term loans and lines of credit loans
- Origin date of loan
- Maturity date
- Original amount
- Current interest rate
- Unpaid balance at month-end
- Additional amount available to draw on if it’s a LOC loan
- Category (e.g., 0 to 30 days vs. over 90 days delinquent)
- Amount of contractually due principal payments received by vintage year
- Amount of total principal payments received by vintage year
Note: If you use migration analysis (track net charge-off percentages by risk rating / FICO credit scores) you will also need to enter the risk rating / FICO credit into the subsidiary ledger. If you want to use the PD / SL method, your loan subsidiary ledger must also track the number of loans that default, number of loans originated for each loan type, and the severity of loss (net charge-off as a percentage of the unpaid balance of the loans outstanding) for each vintage or year of origination.
Most smaller lenders (e.g., less than $10 billion in total assets) will not have the data to use the NPV method so the additional data needed for this method is not discussed.
Additional data will be required to justify subjective adjustments to the CECL historical charge-off percentages. To help with this, we provide our clients with tables and graphs that segment their loan portfolios by:
- Collateral type
- Ranges of the loan-to-value ratio
- Ranges of the debt service coverage ratio for commercial loans and debt-to-income ratio for consumer loans
- Risk rating for commercial loans and FICO credit scores for consumer loans (assuming the institution doesn’t risk consumer rate loans)
- Separating the loans located inside and outside of the normal trade area
- Loans acquired through participation
- Loan officer responsibility codes (to determine if there are any trends in loan officers’ individually-managed portfolios)
- Delinquency status
- Spec versus presold loans for commercial construction one-to-four family loans
We are adding stress testing to our process in the next 60 to 90 days so that the effect of various changes in the real estate market, interest rates, or borrower cash flows can also be considered in the establishment of the subjective adjustments. Our team of actuaries has also developed a separate model to measure the impact of changes in national unemployment rates, gross national product, and other leading economic indicators on charge-off rates experienced by lenders. The results of this process will also be provided to you to assist in the creation of the subjective adjustments. To our knowledge, no other vendor has actuaries assist their clients in projecting future net charge-off rates.
We recommend that our clients also enter the following data into their loan subsidiary ledgers:
- Collateral type (To do this correctly most lenders will need to add significantly more loan types to their loan subsidiary ledgers)
- Risk ratings for commercial loans
- FICO credit scores for consumer loans
- Cash flow generated from on-going operations (commercial loans)
- Principal and interest payments due to the institution (commercial loans)
- Principal and interest payments due to other lenders (commercial loans)
- Estimated market-value of collateral pledged against the loan
- Name of the master or general contractor for commercial construction loans
- Debt-to-income ratio for consumer loans
- Number and type of policy exceptions
- Number and type of technical exceptions
- Zip code for real estate loans (this information is already in the loan subsidiary ledger)
- Whether the loan is on nonaccrual status or a TDR (this information is likely already in your loan subsidiary ledger)
The good news is that your data processing system is already set up to store this additional data. An interagency statement issued by the FDIC, OCC, and Federal Reserve Bank in 2006 required financial institutions to provide this type of data for major loan concentrations. This statement hasn’t been enforced well but, based on my observations of examiner actions in my client base over the last 12 months, this is likely to change. Banks and credit unions are also being asked to do a more effective job of stress testing their loan portfolios. To satisfy this requirement, institutions will need information like the above in their loan subsidiary ledger.
HOW LONG WILL IT TAKE TO ENTER THE ADDITIONAL DATA NEEDED FOR CECL?
Based on my experience helping clients segment their loan portfolios, it takes two loan processors two weeks to enter the data for 1,000 loans if they spend approximately five-hours per day on the project. (In most institutions, it’s not realistic for a loan processor to spend their entire day on this data scrub.)
After the initial data scrub, entering additional data can be built into the process for booking new loans or updating the system after annual reviews are completed for larger commercial loans.
WHAT QUESTIONS SHOULD I ASK CECL VENDORS?
- Who designed the software and what are their professional qualifications? Do they have a team staffed with CPAs, lenders, quants or actuaries?
- How easy is it to validate the CECL calculations? Auditors and regulators will want to validate the CECL calculations. If this isn’t fully transparent, your annual audit could cost more. Calculations that are not transparent could also result in regulatory criticism if examines cannot validate your model or your senior management team can’t readily explain how it works.
- Does the vendor’s solution use internal data or national net charge-off data in the CECL calculations? If the former, have they considered statistical credibility impacts? If the latter, where do they obtain this data and how is it verified? Can you, your auditors or examiners validate the national data? How does the vendor adjust this data to your institution’s individual situation? If the model doesn’t consider the lending practices or net charge-off history unique to your institution, it may unjustly apply the loss history from the areas of the country that typically experience higher losses to your portfolio resulting in a higher ALL balance than what is necessary. Examiners have stated that the use of peer or published data is acceptable to begin with if you don’t have the data internally. However, over time they expect all institutions to document the assumptions they use and correlate the data used in the models to their individual institution’s net charge-off history and lending practices.
- How many loan types does the vendor use in their CECL model? The more loan types you use the better because it better segments your loan portfolio. This is especially true if your institution experiences heavy losses for a loan type as discussed above.
- How does the vendor protect the confidentiality of customer/member information? Is the host site subject to an annual SSAS #16 report, either a SOC 1 or 2 type report or both?
- How does the vendor calculate the CECL compliant ALL? Do they use one method or multiple methods? Ask for side-by-side comparisons of CECL using various methods so that over time, the ideal method can be determined.
- What is the cost? I have heard costs ranging for a $100-million-dollar institution from $4,000 per month plus on-boarding fees of $32,000 to as low as $5,000 per year with no to minimal on-boarding fees. Be cautious with a model that is too cheap. Although they may be good to use in estimating the impact of CECL, they may not be reliable in the actual implementation of CECL. Remember, the adage, if it’s too good to be true, it probably is. Some of the fees quoted by the more expensive firms are just too high for what they offer!
- What additional reports or information do you receive with the vendor’s solution?
- How much time will your staff spend each month on CECL? If your staff need to do a lot of work to produce the CECL calculations, that cost should be considered in the evaluation process.
CECL is a major change but, with proper planning and a structured approach to implementation you will get through it and most likely, know more about the risk within your loan portfolio than you ever did before. CECL should result in most institutions improving their risk management practices over their loan portfolios.