Where the Double Count Enters
The specific problem in acquired-loan accounting is the non-PCD path: a loan is first recorded at a purchase price or fair value, then an allowance for credit losses is recognized through Day-1 credit-loss expense. The question is whether those two steps can describe the same expected shortfall twice. Tracing the entries and their later consequences makes the issue easier to judge than treating “double count” as a slogan.
The answer also requires a second distinction. A fair-value mark is not automatically an ACL, and the solution is not to estimate a credit portion of the mark and subtract it from the allowance whenever the numbers seem to overlap. The accounting path determines what each amount is allowed to do.
Fair value sets the starting measurement, not the allowance
An acquired loan enters the buyer’s books with a new measurement. In a business combination, the loan is recorded at acquisition-date fair value: the price that would be received to sell the asset in an orderly transaction between market participants. In other acquisition settings, the initial investment generally reflects the amount paid under the applicable guidance. This measurement replaces the seller’s previous carrying amount as the acquirer’s starting point.
Fair value is not a forecast of one borrower’s payment behavior written in a separate allowance column. It is a market-participant measurement. A market participant pricing a loan would consider the cash flows it expects to receive and the risks that could prevent collection, along with other relevant market assumptions. Credit expectations can therefore affect the price without being separately itemized as “the credit loss” in the valuation mark.
The ACL answers a related but different question. Under Topic 326, the acquirer estimates the amount of credit loss expected over the relevant life of the financial asset. The estimate is presented as an allowance against the asset, with changes generally flowing through credit-loss expense. The two measurements are connected because the price of an acquired loan can reflect expected collection. They are different because fair value establishes the acquisition measurement while the ACL is the credit-loss measurement required by the impairment model.
That difference is easy to lose when a loan is described only as “bought at a discount.” A discount has a reference point. It may be the difference between unpaid principal and the price paid, or between another contractual amount and fair value. The difference can contain credit expectations alongside other market-participant assumptions. The word “discount” describes an amount; it does not, by itself, identify the accounting function of every dollar in that amount.
The non-PCD model is where FASB says the overlap becomes a reporting problem. A financial asset acquired without more-than-insignificant deterioration in credit quality since origination is treated, under the model described in the Update, in a manner consistent with an originated asset. The asset is initially recorded at purchase price or fair value. The ACL is then recognized with a corresponding Day-1 credit-loss expense.
In abstract form, the sequence looks like this:
- Record the acquired loan at its acquisition measurement.
- Estimate expected credit losses under Topic 326.
- Record the ACL and a provision expense.
- Carry the loan net of the allowance and continue measuring income over time.
The first and third steps can contain overlapping information. Suppose a market participant would pay less for a loan because some contractual cash flows are not expected to be collected. That expectation affects the first step through the price. If the same expected shortfall is then recognized again as a Day-1 allowance that reduces the carrying amount, the net balance can be lower than the amount the buyer expects to collect. FASB describes this as double counting of expected credit losses already considered in fair value.
This is an economic critique of the presentation, not a claim that the two journal entries are literally identical. The fair-value measurement may include many inputs, and the ACL may be measured using a different method and set of assumptions. The concern is that the non-PCD model gives the same expected credit risk two accounting consequences: it is reflected in the acquisition measurement and also charged immediately through the allowance process. The overlap can exist even when no one has intentionally recorded the same invoice twice.
The distinction matters because the allowance is not merely an explanatory note. It changes the reported net amount of the asset. The Day-1 provision changes current-period income. The new carrying basis then becomes part of how future interest income and credit performance are interpreted. A transaction can therefore look unusually costly at acquisition and unusually productive later, even though the underlying cash flows have not changed simply because the accounting labels moved.
The sequence shows why the issue is about placement and timing. The market measurement does not disappear when the allowance is recorded. The ACL does not become a decomposition of fair value. Under the non-PCD route, both remain in the accounting record, and the allowance is charged through income at acquisition. FASB’s concern is the resulting combination, not the existence of either measurement in isolation.
The overlap changes what later performance appears to say
A loan’s cash economics and its reported income pattern are related, but they are not the same thing. The buyer pays cash at acquisition and later receives principal and interest, subject to collection risk. Accounting determines when the purchase-price adjustment, expected credit loss, and subsequent change in expectations enter the income statement. A reader who sees only the later yield may miss what happened on Day 1.
Consider the direction of the effect without assigning a universal valuation formula. If expected losses have already contributed to a lower acquisition measurement, and a separate Day-1 ACL reduces the asset again, the loan begins its life in the accounts with a net amount that may be below expected collection. FASB says that this “double count” can artificially inflate the subsequent yield on acquired loans. The later yield can look stronger because the accounting basis has absorbed the consequences of two overlapping reductions, while the cash to be collected has not increased.
This is why the issue affects comparability. Two acquirers could purchase economically similar loans and report different Day-1 expense and later yield depending on whether their transactions are classified in the PCD or non-PCD model, and on how they document that classification. FASB’s basis for conclusions says stakeholders found the distinction difficult to understand and inconsistently applied. A reader comparing provision expense, net interest income, or post-acquisition performance may need to know whether an acquisition-date expected loss was expensed immediately or built into the gross-up structure.
The effect should not be described as a free gain or loss created by accounting. The buyer cannot collect more cash merely because a lower net carrying amount produces a different yield calculation. Nor does a gross-up entry make a distressed loan sound. The issue is the timing and location of recognition. FASB notes that cumulative income over the loan’s life reflects cash collected minus cash paid, while the timing and classification of income differ between the PCD and non-PCD models.
That distinction also helps explain why the Day-1 expense can be misleading in both directions. A large provision at acquisition may suggest that credit quality deteriorated after the deal, even when the expected loss was already part of the acquired asset’s economics. A strong later yield may suggest that the loan is outperforming, even when part of the apparent return reflects accretion or amortization of an acquisition adjustment. Neither conclusion is safe without identifying the acquisition model and the components of the starting basis.
The correct question is not “Did the bank recognize an ACL?” Every acquired loan within the relevant credit-loss framework needs an allowance. The better questions are: What did the acquisition measurement already reflect? Where was the acquisition-date ACL recorded? Which portion of the purchase adjustment is noncredit? What changed after acquisition? And which income line reports each change? These questions follow the transmission path from valuation to balance sheet to earnings.
There is a related reason not to infer the answer from a single ratio. A lower net loan balance can affect a yield calculation, but it does not reveal which assumptions generated the lower price or how the allowance was estimated. A provision ratio can rise because of the initial recognition convention rather than a new deterioration in the acquired portfolio. A later reversal can reflect changed expectations, not cash recovery. The source-backed lesson is to read the acquisition entry and the subsequent measurement together.
The OCC’s answer to the tempting “credit portion” shortcut makes this boundary explicit. For acquired loans that are not PCD or purchased seasoned loans, the entire fair-value mark is treated as a purchase premium or discount that will ultimately be accreted or amortized into interest income over the remaining lives of the loans. The unaccreted discount or unamortized premium remains part of amortized cost, against which the ACL is evaluated. The full amount of credit loss must be recognized through PCL.
Why not separate the mark after acquisition? Because the non-PCD/non-PSL accounting path does not authorize that reconstruction. The fact that credit risk helped determine fair value does not create a new ledger category that management may use to reduce the ACL. The mark is handled through the yield mechanism, while the credit-loss estimate is handled through the allowance and PCL. Treating part of the mark as an offset would not correct the model FASB criticized; it would create a different, unsupported accounting treatment.
This does not contradict the gross-up approach. Gross-up is a specified model for PCD assets and, under ASU 2025-08, qualifying PSLs. In that model, the acquisition-date ACL is added to the purchase price or fair value to establish initial amortized cost, and the remaining noncredit discount or premium is handled separately. The model itself supplies the boundary. A non-PCD/non-PSL loan does not acquire the gross-up treatment merely because its fair-value mark appears to include credit risk.
The distinction between models can be expressed as two different answers to the same acquisition question:
- Non-PCD/non-PSL treatment: record the acquisition measurement, recognize the ACL through credit-loss expense, and keep the purchase discount or premium in the amortized-cost and yield mechanics.
- Gross-up treatment: add the acquisition-date ACL to the purchase price or fair value to establish initial amortized cost, then separate the noncredit discount or premium from later changes in expected credit loss.
The labels are not the conclusion; they tell the reader which recognition path applies. That is why classification evidence matters, and why PCD deterioration, PSL seasoning, and boundary cases outside the new route remain essential to the analysis.
For now, the essential mechanism is complete. Fair value can already reflect the expected reduction in cash flows from credit risk. A separate non-PCD Day-1 ACL can recognize that expectation again in the presentation, leaving the net asset below expected collection and distorting later yield. The answer is not to call the entire purchase discount an ACL or to net an invented credit portion against the allowance. The answer depends on the permitted acquisition model. The next step is to see the gross-up approach in a worked entry, where the ACL and the noncredit discount are assigned different jobs from the start.