Lifetime expected credit losses for companies that are not banks: what is in scope, how the allowance gets built and supported, and the elections that simplify it.
What ASC 326 requires, and how the allowance gets built
Key takeaways
What it is. CECL reserves lifetime expected credit losses on financial assets at amortized cost, booked at origination and updated every period, built on historical experience, current conditions, and reasonable and supportable forecasts.
Where it breaks. Scope discovered late (this is more than a bank standard), methods the data cannot support, forecasts without documentation, and qualitative overlays that read as plugs instead of evidence.
How we help. We scope the balance sheet, build the allowance model and the forecast framework, document the elections including the ASU 2025-05 practical expedient, and support the estimate through audit.
CECL reads like a bank standard, and most of the commentary around it is written for lenders. But the scope is financial assets measured at amortized cost, which puts every company's trade receivables, contract assets, notes receivable, and held-to-maturity securities inside it. For an operating company the exposure is usually concentrated in receivables, and the work is proportional: a pooled aging model, a documented view on conditions and forecasts, and disclosure that ties to it.
The standard's flexibility is the hard part. No method is prescribed, no forecast horizon is mandated, and no bright line separates a supportable adjustment from a plug, which means the documentation carries the estimate. This page covers where the work actually sits for non-banks, and the elections, including the new ASU 2025-05 practical expedient, that keep it proportional.
The standard
How CECL changed the allowance model
CECL replaced the incurred-loss model, which waited for a loss event to become probable, with a lifetime expected-loss model that books the reserve on day one. The estimate covers the asset's contractual life, adjusted for expected prepayments, and rests on forecasts that must be reasonable and supportable before reverting to historical experience. The change sounds bank-shaped, but it applies to any company holding trade receivables, contract assets, loans, or held-to-maturity debt.
The move from waiting for probable to reserving expected losses on day one. Illustrative and not exhaustive.
Measurement
Building the CECL allowance estimate
The allowance draws on three inputs: historical loss experience, current conditions, and reasonable and supportable forecasts that revert to history beyond the supportable horizon. No estimation method is prescribed; loss-rate and aging approaches fit trade receivables, while PD times LGD and discounted cash flow fit loan portfolios. ASU 2025-05 adds a practical expedient for current trade receivables and contract assets from ASC 606 revenue, letting an entity assume current conditions persist instead of forecasting, with entities other than public business entities also able to consider cash collected after the balance sheet date.
The inputs, the method menu, and the new ASU 2025-05 practical expedient for trade receivables.
This is for you if
Your auditors have asked how the allowance reflects CECL, and the answer is still a percentage from memory.
Receivables, notes, or a loan book are growing faster than the process that reserves against them.
You hold held-to-maturity securities or contract assets with no documented expected-loss position.
You want the ASU 2025-05 elections evaluated before the next fiscal year locks the policy.
What you get
The scope memo Every in-scope asset class identified, with the available-for-sale and PCD boundaries documented.
The allowance model Pooling, loss rates, and a method fit to your data, built to update rather than rebuild.
The forecast framework The reasonable and supportable horizon, the reversion approach, and overlay governance in writing.
The elections The ASU 2025-05 practical expedient and the collection-activity election evaluated and papered.
The disclosures Rollforward, credit-quality, and policy footnotes generated from the model output, not drafted apart from it.
How We Help
What we deliver
A CECL estimate built as a standing workpaper: scoped, modeled, documented, and refreshable at every close.
The scope memoEvery in-scope asset class identified, with the available-for-sale and PCD boundaries documented.
The allowance modelPooling, loss rates, and a method fit to your data, built to update rather than rebuild.
The forecast frameworkThe reasonable and supportable horizon, the reversion approach, and overlay governance in writing.
The electionsThe ASU 2025-05 practical expedient and the collection-activity election evaluated and papered.
The disclosuresRollforward, credit-quality, and policy footnotes generated from the model output, not drafted apart from it.
When companies bring us in
Your auditors have asked how the allowance reflects CECL, and the answer is still a percentage from memory.
Receivables, notes, or a loan book are growing faster than the process that reserves against them.
You hold held-to-maturity securities or contract assets with no documented expected-loss position.
You want the ASU 2025-05 elections evaluated before the next fiscal year locks the policy.
Our Experience
Where we have done this work
Engagement Notes
A software company that thought CECL did not apply
A growth-stage SaaS business heading into its first audit had reserved receivables with a flat percentage and assumed CECL was a bank problem. We scoped the balance sheet, found trade receivables and a book of multi-year financing arrangements in scope, built a pooled aging model segmented by customer tier, documented the reasonable and supportable forecast and reversion, and elected the ASU 2025-05 practical expedient for the current receivables. The allowance memo went into the audit as a standing workpaper rather than a year-end scramble.
Engagement Notes
A specialty lender's model, governed and defensible
A non-bank consumer lender had a PD and LGD model its auditors kept challenging on the qualitative overlays. We rebuilt the framework: pools tied to origination vintages, a forecast horizon anchored to variables that actually correlated with the portfolio's losses, and an overlay policy with quantified factors and documented governance instead of a single judgment number. The rollforward and vintage disclosures were generated straight from the model, and the following audit closed the CECL area without a comment.
The Detail
The gaps, and how we close each one
Issue 01
Scope: this is more than a bank standardASC 326-20
CECL covers financial assets at amortized cost, and companies discover the scope late: trade receivables and contract assets from ASC 606, notes and loans receivable, net investments in leases, held-to-maturity debt securities, and off-balance-sheet credit exposures such as loan commitments and financial guarantees. Available-for-sale debt sits outside, under its own model.
The treatment
Start with an inventory of the balance sheet against the scope list, because the allowance is asset-class by asset-class, not one number. Trade receivables and contract assets carry the day-to-day work for most operating companies. Held-to-maturity securities need an expected-loss view even when the portfolio is treasuries and investment grade; a zero-loss conclusion is permitted but must be reasoned and documented, not assumed. Available-for-sale debt follows ASC 326-30, where impairment runs through an allowance but the trigger and measurement differ, so mixing the two models is a common error. Purchased credit-deteriorated assets get a gross-up at acquisition rather than day-one expense, which changes deal accounting for portfolios bought at a discount.
How we handle it: We inventory the balance sheet against the ASC 326 scope and paper the boundaries, including AFS and PCD.
Issue 02
Trade receivables: the pooled aging buildASC 606 / 326
The legacy habit, reserving for specifically identified bad accounts plus a round percentage, does not comply. CECL requires expected losses on current and not-yet-due balances too, estimated on pools that share risk characteristics, with the estimate refreshed every period.
The treatment
The compliant build is still proportionate: pool receivables by customer type, geography, or product where loss behavior differs; compute historical loss rates by aging bucket from writeoff experience; then adjust for current conditions and the forward view. ASU 2025-05 makes the forward view electable away for current trade receivables and contract assets arising from ASC 606 revenue: assume conditions as of the balance sheet date persist for the assets' remaining life. Entities other than public business entities that elect it may also elect to consider cash collected after the balance sheet date, which for a private company with quick collections can shrink the estimate to arithmetic. Effective for annual periods beginning after December 15, 2025, with early adoption permitted, and the election still needs policy documentation and disclosure.
How we handle it: We build the pooled aging model and evaluate the ASU 2025-05 elections, documented as policy.
From our engagements: The expedient covers current receivables and contract assets from ASC 606 transactions. Long-term notes, lease receivables, and related-party loans stay on the full model.
An audit question on the allowance, or a fiscal year about to lock the policy? Talk to us before the estimate hardens.
ASC 326 prescribes no method, which means the wrong choice is available: a PD-times-LGD model without default data to calibrate it, a discounted cash flow build the close cannot refresh, or a loss-rate method applied to a portfolio whose mix has shifted away from its history.
The treatment
Fit the method to the asset and the data: aging and loss-rate methods for trade receivables, vintage or roll-rate where origination cohorts behave differently, PD/LGD or discounted cash flow for loan books with the data to support them, and the collateral-dependent measure where repayment rests on the collateral. Then define the forecast discipline: how far the forward view is reasonable and supportable, what happens at the horizon (reversion to history, immediately or on a slope), and which economic variables actually correlate with the portfolio's losses. Document the choices once, as policy, so each quarter is an update rather than a rebuild.
How we handle it: We fit the method to your data, define the forecast horizon and reversion, and write it once as policy.
Issue 04
Qualitative overlays that read as evidence, not plugsGovernance
History rarely captures the current quarter's conditions, so qualitative adjustments carry the difference, and they are the first thing auditors challenge: the direction, the magnitude, the support, and whether the overlay quietly reverses the model whenever the output is inconvenient.
The treatment
Give the overlay the same discipline as the model: a defined factor framework (customer concentration, macro conditions, portfolio mix, operational changes) rather than a single judgment number; quantified anchors tying each factor to a measurable input; consistent direction, so the factors move both ways as conditions change; and governance, meaning a documented review and approval that treats the overlay as part of the estimate, with back-testing against actual losses feeding the next quarter. An overlay that only ever moves the reserve up, never down, and never ties back to actual losses is the pattern auditors flag.
How we handle it: We build the overlay framework with quantified anchors and the governance that survives audit.
Issue 05
Disclosures that tie to the modelDisclosure
The disclosure package is where a thin allowance process becomes visible: the policy description, the methodology by portfolio segment, credit-quality information, and the allowance rollforward all have to reconcile to each other and to the model that produced them.
The treatment
Build the disclosures from the model outputs rather than drafting them separately: the rollforward ties beginning balance, provision, writeoffs, and recoveries to the ledger; credit-quality indicators present the portfolio the way management actually monitors it; public business entities add vintage disclosures for financing receivables; and the policy footnote describes the method, the forecast horizon, and the reversion approach in the same terms as the workpapers. Where the ASU 2025-05 expedient or the collection-activity election is taken, say so. Consistency between footnote, model, and MD&A is most of what a reviewer checks.
How we handle it: We generate the rollforward and credit-quality disclosures straight from the model output.
FAQ
Frequently asked questions
We are not a bank. Does CECL actually apply to us?
Yes. The scope is financial assets at amortized cost, which includes trade receivables, contract assets, notes receivable, net investments in leases, and held-to-maturity debt securities. For most operating companies the work concentrates in receivables and is proportional to the portfolio, but an allowance policy, a documented estimate, and the related disclosures are required.
What did ASU 2025-05 change?
It added a practical expedient for current trade receivables and current contract assets arising from ASC 606 revenue: an entity may assume current conditions as of the balance sheet date persist for the assets' remaining life, instead of developing forward-looking forecasts. Entities other than public business entities that elect the expedient may also elect to consider cash collected after the balance sheet date. It is effective for annual periods beginning after December 15, 2025, and early adoption is permitted.
Our receivables collect in thirty days. How much process do we really need?
A short-duration, well-collected portfolio needs a proportionally small process: pooled aging analysis, documented loss rates, the ASU 2025-05 election if it fits, and a policy memo. What it cannot be is zero: even a conclusion that expected losses are immaterial has to be reasoned and documented, and auditors ask for that support.
Which estimation method should we use?
The standard does not prescribe one. Aging and loss-rate methods fit trade receivables; vintage, roll-rate, PD/LGD, and discounted cash flow methods fit lending portfolios with richer data. The governing constraint is that the data has to support the method through audit, which usually argues for the simplest method the portfolio allows.
What do auditors test in a CECL estimate?
The data lineage from ledger to model, the support for loss rates and the forecast horizon, the governance over qualitative overlays, the consistency of elections with the policy memo, and whether the disclosures tie to the model output. Most findings are documentation findings, which is why the estimate is built as a standing workpaper rather than a quarterly scramble.
Sources & authorities
Primary sources for this page
ASC 326, Credit Losses. The current expected credit loss (CECL) model: a lifetime loss estimate recognized at origination.
The CECL standard. FASB ASU 2016-13, which replaced the incurred-loss model.
This page summarizes FASB accounting standards and SEC guidance for general information, and is not accounting or legal advice. Standards change; confirm the current text before you rely on it.
