Many CECL allowances are a historical loss rate applied to a pool, plus an adjustment. Two decisions dominate the result: how the pools are drawn and which history feeds the rate. A pool that mixes unlike loans, or a history that omits write-offs, produces a number that looks precise and is wrong. This post sets out what ASC 326-20 requires on those points, shows a loss-rate calculation with a data completeness error, and lists the audit procedures. For the broader picture, see CECL for private companies and CECL audit procedures: model validation and qualitative factors.
When must assets be pooled?
ASC 326-20-30-2 (from ASU 2016-13) requires expected credit losses to be measured on a collective (pool) basis when similar risk characteristics exist. An asset that does not share risk characteristics with the others is evaluated individually, and an asset evaluated individually is not also included in a collective evaluation, so nothing is counted twice.
Paragraph 326-20-55-5 lists characteristics that may be used to aggregate assets, and says the list is not all inclusive: internal or external credit scores or ratings, risk ratings or classification, financial asset type, collateral type, size, effective interest rate, term, geographical location, industry of the borrower, vintage, historical or expected credit loss patterns, and reasonable and supportable forecast periods.
What must the historical information reflect?
Paragraph 30-8 says historical credit loss experience of assets with similar risk characteristics "generally provides a basis" for the assessment, and it can be internal, external or a combination. The entity considers adjustments for differences in current asset-specific risk characteristics, such as underwriting standards, portfolio mix or asset term within a pool at the reporting date, or when the history does not reflect the contractual term.
Paragraph 55-3 adds that the historical data, after adjustment for current conditions and forecasts, should be applied to pools defined consistently with the pools in which the experience was observed. If the pools were redrawn, the history must be re-cut.
What adjustments does the standard require?
Under 30-9, an entity does not rely solely on past events. It considers the need to adjust historical loss information so it reflects the extent to which management expects current conditions and reasonable and supportable forecasts to differ from the conditions during the historical period. Adjustments can be qualitative and should reflect changes in relevant data such as unemployment rates, property values, commodity values or delinquency. Paragraph 55-4 lists factors, including the borrower's financial condition, remaining terms, the volume and severity of past-due and adversely rated assets, changes in lending policies and underwriting, the quality of the credit review system, and environmental factors such as economic conditions in the area of concentration.
What is reversion?
An entity is not required to forecast over the contractual life. For periods beyond which it can make or obtain reasonable and supportable forecasts, it reverts to historical loss information that is reflective of the contractual term, and it does not adjust history for existing conditions or expected future conditions beyond the reasonable and supportable period (30-9). It may revert at the input level or based on the entire estimate, and immediately, on a straight-line basis, or on another rational and systematic basis.
Paragraph 55-6 lists the judgments management makes, and each is an audit area: the definition of default, the approach to measuring historical loss, the historical period, how to adjust for conditions and forecasts, how to treat recoveries, prepayments, the reversion approach, and whether assets share risk characteristics.
Worked example: a vintage loss rate with a data gap
A lender pools unsecured term loans. Management uses fully seasoned vintages to compute a lifetime loss rate:
| Vintage | Original balance | Lifetime losses | Loss rate |
|---|---|---|---|
| 2021 | $2,000,000 | $52,000 | 2.60% |
| 2022 | $2,200,000 | $61,600 | 2.80% |
| 2023 | $2,400,000 | $50,400 | 2.10% |
| Total | $6,600,000 | $164,000 | 2.4848% |
The pool's current balance is $3,500,000. Management adds 0.35 points for a worsening unemployment forecast, giving 2.83485% and an allowance of $99,220 ($3,500,000 x 2.83485%).
The auditor reconciles the loss history to the allowance rollforward. The rollforward shows $171,500 of charge-offs, net of recoveries, on these vintages for the period, but the loss file shows $164,000. In this illustration the $7,500 difference came from loans transferred to a collections system that the loss file did not read. Corrected history: $171,500 / $6,600,000 = 2.5985%. With the same 0.35-point adjustment the rate is 2.94849% and the allowance $103,197. The gap in data understated the allowance by $3,977. The fix is to the data, not to the qualitative layer: a larger overlay would hide the error and carry it forward.
Audit procedures
- Test the pooling design. Obtain management's documentation of why the pools were chosen. Compare to the characteristics in 55-5 and to how losses actually vary. Do the pools hide a segment with a materially higher loss rate? Test that individually evaluated assets are excluded from pools (30-2).
- Test the completeness of the historical population. Reconcile loan counts and balances in the loss file to the subledger and general ledger at several dates. Include loans that were sold, transferred, restructured, charged off, or migrated to another system.
- Test the accuracy of losses. Reconcile charge-offs and recoveries in the loss file to the allowance rollforward, as above. Trace a sample of write-offs to approvals and loan files. Check that losses are attributed to the right vintage, and that the definition of default or loss is applied consistently.
- Test period selection. Ask why the historical window was chosen and what it includes. A window that omits a downturn, or includes only a benign period, needs support (55-3, 55-6(c)).
- Evaluate adjustments for current conditions and forecasts. For each adjustment, identify the data used, the link between the data and credit losses, and the amount. Compare forecasts to external sources. Test whether the same factors are already in the history (55-4 says adjust for conditions "not already reflected").
- Evaluate the reasonable and supportable period and reversion. Obtain the policy, test that it is applied consistently, and recompute the reversion. A forecast period of several years with no reversion invites bias.
- Test contractual term and prepayments. Confirm that the loss rate matches the life of the assets in the pool (30-6, 30-8).
- Test the calculation. Recompute the allowance from the inputs, and agree the pool balances to the subledger at the reporting date.
- Retrospective review. Compare last year's allowance to realized losses, using the approach in retrospective review of prior-year estimates.
- Conclude on bias under AU-C 540 paragraph .32, individually and in aggregate with other estimates.
Frequently asked questions
Can an entity use another entity's loss data?
Yes. Paragraph 30-8 allows internal or external historical loss information, or a combination. The auditor evaluates whether the external data reflects assets with similar risk characteristics and similar terms.
Does the standard require a forecast over the life of the asset?
No. Beyond the reasonable and supportable period, the entity reverts to history (30-9).
What if a pool is too small to give a reliable loss rate?
Combine pools only where risk characteristics are similar, or evaluate the assets individually. The standard does not permit ignoring risk of loss because history is thin, since the estimate includes a measure of expected risk of loss even if remote, unless history adjusted for conditions and forecasts gives an expectation of zero nonpayment (30-10).
Sources
- FASB, Accounting Standards Update 2016-13, Financial Instruments, Credit Losses (Topic 326), paragraphs 326-20-30-2, 30-6 to 30-10 and 326-20-55-3 to 55-7: https://storage.fasb.org/ASU%202016-13.pdf
- AICPA, AU-C Section 540, Auditing Accounting Estimates and Related Disclosures, paragraphs .22 to .24 and .32: https://www.aicpa-cima.com/resources/landing/aicpa-professional-standards
CORAA prepares ready-to-review audit files for US CPA firms from client books; you review and sign. See /us.
Part of CORAA University · Evidence and audit procedures: related guides, calculators and templates.
AU-C references are to AICPA Professional Standards, paraphrased here; read the standard itself before relying on a requirement. This page is guidance, not professional advice. Report an error