Analysis becomes more difficult for investors if financial statements fail to faithfully represent the underlying economics of a business. Sometimes this reflects the limitations of accounting; however, sometimes the accounting is simply misleading – for example, US GAAP accounting for loan losses.
The FASB has recently issued an amendment to US GAAP to correct an anomaly where an accounting loss is recognised when loans are acquired, even though no economic loss has occurred. We explain why accounting and economic reality can diverge, and how the US GAAP amendment will help investors.
Financial statements are most useful when they faithfully represent the underlying economics of a business. A failure to do so, at best makes financial statements confusing, and may necessitate adjustment, and at worst can result in incorrect analysis and poor investment decisions.
Businesses can be enormously complicated and summarising their financial position and performance within the confines of financial statements is difficult. Accounting generally does a decent job, but not always. In our Footnotes Analyst articles we often focus on situations where we think accounting is incomplete or deficient in some way, given that these can present the greatest financial reporting challenges for investors. Problematic situations we have discussed include:
- Not reporting an asset in the balance sheet even though a resource controlled by a business clearly provides future economic benefits.
- Reporting a loss in a period when no economic loss has been incurred. This is the problem FASB has recently corrected in respect of loan impairments – more on that later.
- Measuring assets and liabilities in a way that mis-states the true economic benefit or burden.
A failure of accounting to describe the economics does not necessarily an indicate accounting is deficient; sometimes the accounting is limited by other factors and sometimes there can be disagreement over exactly what is the underlying economics that should be described.
Economic reality – The limitations of accounting
For financial statements to be useful to investors there needs to be some confidence in the data. If, for example, the most relevant measurement of an asset is so subjective that investors cannot rely on what is reported, it may be preferable that an alternative accounting approach is used, even if this does not fully represent the underlying economics.
(Lack of) intangible asset recognition creates a disconnect between accounting and economics
Intangible asset accounting is a good example. Few internally generated intangibles are recognised in the balance sheet, largely due to the difficulties in identifying the assets, measuring their cost or value, and determining when and whether the investment will be recovered. The extent of the restrictions on recognition can be debated (and are presently being reconsidered by both the IASB and FASB), but it is unlikely that we will ever see all intangible resources as assets on company balance sheets simply because it is just too difficult.
One of the consequences of limited recognition of intangible assets is that return on capital measures are distorted and become less useful in equity analysis. Any investor using ROIC and similar metrics needs to understand this.
For more on intangible asset accounting, and some of the resulting challenges for investors, see our articles ‘Missing intangible assets distorts return on capital’, ‘Intangible asset accounting and the ‘value’ false negative’ and ‘DCF terminal values: Returns, growth and intangibles’.
Undiscounted deferred tax assets can overstate value
Another example of where measurement in accounting can fail to reflect underlying economic value is deferred tax. Although the measurement of some deferred tax, in effect, allows for the time value of money and closely matches economic value, this is not the case for deferred tax recognised in respect of carried forward tax losses. The tax loss deferred tax asset is simply equal to the amount of those losses multiplied by the tax rate (if recovery is deemed probable). If recovery is not likely, zero is recognised.
There are two problems with this approach:
- If recovery through reduced tax payments is spread over many years the present value of this benefit will be lower; and,
- The all or nothing recognition may not reflect the reality of there being a probability that the tax will be saved.
Ideally the measurement should be the present value of the expected (probability weighted) tax savings arising from the losses – a measurement approach used successfully in accounting for other assets and liabilities.
It is very difficult to estimate the timing and likelihood of recovery of tax losses, which makes it difficult to obtain a reliable measure of economic value. Accounting therefore goes for a simpler but less relevant undiscounted measurement basis. Although this provides some help, investors need to consider carefully whether using this asset without adjustment in equity valuation is the best approach.
We discussed this topic in our article ‘Deferred tax fails to reflect economic value’ using UK telecoms stock Vodafone to illustrate. The data in the article is from a few years ago, but the current position is similar, with a £17.5bn deferred tax asset recognised in the 2025 balance sheet and a further $13.8bn unrecognised.
Vodafone deferred tax balances
Vodafone financial statements 2025
The company states it only recognises a deferred tax asset in respect of losses that will be utilised within 60 years, indicating that the lack of adjustment for the time value of money has a material effect, and that the economic value of these tax losses is likely to be much lower than the balance sheet asset. Considering the current market capitalisation of Vodafone is only £22.5bn, the market seems to take the same view.
Economic reality – More than one perspective
Some transactions can be viewed from more than one perspective, such as using a different measurement basis. One accounting method must be selected for presentation in financial statements, possibly with supplemental information provided from the other perspective. Sometimes the reporting basis selected may not be the best at depicting the economic effects that matter in a particular situation. A good example of this is the choice of cost or fair value for the measurement of financial instruments.
Does cost or fair value best describe the economics of lending?
The accounting for bank deposits and loans is generally based on amortised cost measurement, with net interest margin as a measure of performance. The fair value of loans is also disclosed, but not what performance would be if fair value accounting were fully applied. While the net interest margin approach may adequately reflect the underlying economics in most situations, sometimes it may be deficient, with fair value being more relevant for investors. We covered this topic in our 2023 article ‘Fair values and interest rate risk – Silicon Valley Bank’.
Silicon Valley Bank failed in 2023, even though up to that time it had been reporting what seemed to be healthy positive profits, a decent interest margin and a respectable return on equity. The economic problem was an unhedged exposure to interest rate changes, but this did not appear as a loss in the financial statements due to historical cost measurement. The loss would have eventually shown up in the historical cost based income statement as a reduced or negative interest margin. However, the market worked it all out before the negative profit and loss impact materialised and the ensuing liquidity crisis precipitated the collapse. It was an example of the financial statements faithfully reflecting one economic perspective but not faithfully reflecting the underlying economics that really mattered.
It was possible for investors to estimate the underlying economic position of Silicon Valley Bank prior to its collapse. Below was our attempt in 2023 (it was impossible to be accurate) at recreating quarterly profit on a fair value basis.
Recognised net interest income and unrecognised fair value changes – SVB
Silicon Valley Bank quarterly reports and The Footnotes Analyst estimates
There can also be disagreements over exactly what are the economic effects of a transaction. An example we recently covered in ‘Stock-based compensation: Transparency, timing and EPS’ concerns the timing of the recognition of a stock-based compensation expense.
When should stock-based compensation be recognised as an expense?
Under IFRS 2 and US GAAP, the grant of equity and options to employees that vest at the end of a qualifying period is gradually recognised as an expense over the vesting period (often 3 or 4 years) and not at the time of grant. This amortisation approach applies even if employees automatically receive the equity or options simply by remaining in employment, with no other conditions to satisfy. The accounting approach is based on the belief that the grant of equity is compensation for future services to be provided over the vesting period.
An alternative view is that stock-based compensation is primarily a reward for services performed in the preceding period. We think this is a more realistic view of the economics and that this approach provides other benefits in terms of transparency in reporting. Nevertheless, we recognise that the different views (and resulting different recognition of a stock-based compensation expense) are each valid and it is a matter of opinion and judgement as to which should be used for financial reporting.
Sometimes, though we think the economics is clear, with no realistic alternative interpretation, the accounting and economics diverge without justification. The current FASB approach to the recognition of loan credit losses is an example of where we think the accounting makes no sense, which is why we think investors should welcome the recent amendment.
Economic reality missing – Loan loss allowances under US GAAP
We first discussed this topic in 2023 after being alerted to an unrealistically large credit loss in the quarterly results of Citizens Bank. Below we briefly summarise the issue and how the accounting diverges from the underlying economics, before explaining the recent change to US GAAP that is designed to fix the problem. For more about the background to credit impairment accounting, how the current approaches were influenced by the 2008 financial crisis, and the relevance of this for IFRS reporters, follow these links:
- ‘Expected credit losses: Beware the day 2 effect’ also includes a downloadable spreadsheet model to illustrate the different calculations.
- ‘Comparability is crucial for informed investment decisions’ includes our response to FASB when the proposed change was first published.
US GAAP loan impairments (or loss allowances) in respect of bank loans are measured using the ‘current expected credit loss approach’ (CECL), but with one exception: it is the extent of this exception that has just been changed by FASB.
Expected credit losses recognised in full in the period a loan is originated or purchased
The CECL method requires that the full lifetime expected credit loss be recognised immediately a loan is recognised, either when it is originated by a bank and cash is first lent to a customer, or when the loan is purchased if it was first originated by another party. The loss is shown as an expense in profit and loss and as a reduction in the loan asset in the balance sheet. If the loss from actual loan defaults turns out as expected, there is no further credit loss expense to be recognised, and the contractual interest earned is credited to profit and loss. If loss expectations change, the allowance is progressively adjusted.
To illustrate, suppose a bank purchases a portfolio of 5 year loans for $1,000, which is the same as the principal amount outstanding. The loans carry an interest rate of 6% p.a., and the bank expects 1% of the loans to default each year, with no recoveries. Immediately after initial recognition an impairment loss of $50 ($1,000 x 1% x 5) would be recognised and the loans reported at a net $950 in the balance sheet. Interest income of $60 would be recognised in the first year, declining slightly over the 5 years as loans default and some interest remains unpaid.
You may think this approach is perfectly sensible, after all if a loss can be anticipated surely that loss be recognised as soon as it becomes apparent? The problem is that the expected loss is effectively already built into the carrying value of the loan and recognising it again is double counting.
Assuming the bank is happy with a net 5%p.a. return from these loans, if we discount the contractual future cash flows, including the 6% interest, at 5% we (approximately1It does not work out exactly at this amount in our example, but it is close enough to illustrate the point. Our approximations do not alter the message that CECL double counts credit losses.) get to $1,050. In effect this would be what the loans would be worth if there were no defaults. However, the carrying value of the loans is $1,000 (before the CECL allowance is deducted) which means that the expected lifetime credit losses are already included and deducting a day-2 provision to reduce the carrying value to $950 is double counting. The result of applying the CECL method is an unrealistically high expense, and low net interest income, in the period of purchase (or origination).
This is not an easy concept to explain. Here is the explanation given by two of the 7 FASB board members at the time the standard was issued in 2016, who opposed the CECL model:
Messrs. Kroeker and Smith dissent from the issuance of this Accounting Standards Update because they disagree with the requirement to recognize a credit loss at origination or purchase, at an amount equal to the “lifetime expected credit loss” for financial assets. ….
this conceptual shortcoming of this Update thereby results in financial reporting that does not faithfully reflect the economics of lending activities. …
Recording a credit loss at initial recognition (along with a related allowance for loan losses) results in a balance sheet presentation that reflects the credit risk twice; it is reflected in the price paid (which is based on the terms of the instrument, including the stated interest rate) and it is reflected in the allowance for loan losses. In an arm’s-length transaction …. an entity would not be expected to incur an economic loss on the day a loan is made or the security is purchased.
This is an extract from the full dissenting view which can be found at page 235 of US GAAP ASU 2016-13.
Like the two dissenting board members, we believe that US GAAP CECL lifetime credit loss allowances recognised in day-2 fail to reflect the underlying economics of a lending business.
Alternative gross-up method removes the misleading day-2 loss for some loans
The day-2 loss distortion under the CECL approach is most noticeable when credit impaired loans are purchased. These are loans that are already in default or where default is more likely than when the loans were first originated (loans with more than insignificant credit deterioration, to use the US GAAP terminology). For this group of loans, FASB required a different approach called the ‘gross-up method’ where the estimated losses are still recognised, but the allowance is added to the loan carrying value and not immediately expensed.
In our example above this would mean initially recognising the loans at $1,050 less an allowance of $50. The premium of $50 added to the loan amount would reduce the recognised interest income over the life of the loans. In effect, the loss of $50 is spread over the term of the loan in the form of a reduced amount of interest income rather than being recognised up front.
This alternative approach works well. The problem is that it only applies to a subset of loans. In a purchase of loans (such as in a business combination), most will still be under the CECL model with a day-2 loss recognised, which can be highly distorting.
We used a similar example in the model included in our previous article about loan impairments – click here to view the example and the detailed workings for the CECL and gross-up methods.
Below is the Citizens Bank data we used in our 2023 article to illustrate the problem. Notice the large impairment loss in Q1 2023 that is significantly out of line with previous quarters. This (and the similar loss in Q1 2022) is mainly due to the misleading day-2 effect for loans acquired in the period. The exact amount of this day-2 effect is disaggregated in final table below.
Citizens bank net interest income and provisions for credit losses
First Citizens Bankshares 10Q filings
Citizens Bank accounting for non-PCD and PCD loans
Pages 16 and 21 of First Citizens BankShares Q1 2023 10Q. (UPB stands for unpaid principal balance.)
The first table from the Citizens Bank financial statements shows the loans acquired in the quarter. The purchased credit deteriorated (PCD) loans have a fair value significantly lower than the principal amount outstanding. Most of this will be due to the expected credit losses. However, this amount is not the cause of the misleading day-2 loss because of the application of the gross up method. These loans would be recognised at their fair value on the date of acquisition.2Under the gross up method, the carrying amount of the PCD loans would be stated at fair value plus the loan loss allowance. A separate loss allowance is then deducted so that the net carrying amount is fair value, and no day-2 loss is recognised.
It is the non-PCD loans that cause the problem. Although these are initially recognised at their fair value (which is not dissimilar from the principal amount outstanding), the lifetime expected loss is immediately recognised as a reduction in carrying value and an expense, even though this would have already been built into the fair value measurement. It is this day-2 loss that FASB has (mostly) removed through its recent amendment.
A better alignment of accounting and economics – the FASB amendment
The recent amendment to US GAAP expands the application of the gross-up method to purchased ‘seasoned’ loans in addition to purchased credit-deteriorated loans. It seems the concept of seasoned loans3A purchased seasoned loan, according to the amendment, is one that was originated at least 90 days before the date of purchase and where the purchaser was not involved in its origination. is introduced as an anti-avoidance measure to prevent banks structuring loan originations to apply the gross-up method in place of CECL. Subject to this restriction, the gross-up method will, from 2027, apply to all purchased loans, with CECL only applying to originated loans. Had this been in place in 2023 the large loss shown by Citizens Bank in Q1 2023 would not have been reported and the results far more meaningful.
Uneconomic day-2 losses are only partially eliminated by the amendment
However, CECL and the problematic day-2 loss recognition still applies to the purchased ‘non-seasoned’ loans and to all originated loans. The distorting loss is much less noticeable for originated loans considering these take place continuously. For a bank that is not expanding rapidly, the profit and loss effect of CECL and the gross up method will be similar. Nevertheless, in our view the accounting is still at odds with the economics and the balance sheet loan assets and shareholders’ equity are both understated, with a distortion to profit potentially problematic for lending businesses that have high growth.
While we welcome the change to US GAAP, and the removal of some of the misleading day-2 loan impairment losses, we think the amendment does not go far enough and that the gross-up method should be applied to all loans. We also think that the US GAAP accounting for loan losses is still unnecessarily complex, and that one bright line (the definition of PCD) has been replaced by another (the definition of ‘seasoned’).
A final comment on IFRS … The CECL day-2 loss approach does not apply to IFRS reporters, instead, IFRS 9 requires the use of a 3-stage build up approach. There is still a day-2 loss, but it is generally much smaller than for US GAAP reporters and we think the accounting more closely matches the underlying economics. Like US GAAP, IFRS also has a different approach for purchased credit impaired loans because the day-2 loss becomes more material for these, but the IASB has not made the same change in respect of other purchased loans.
In our article ‘Expected credit losses: Beware the day 2 effect’, we provide a downloadable interactive model which compares IFRS with US GAAP.
Insights for investors
- Accounting does not always reflect the underlying economics of a business in financial statements. Most commonly this is due to the inherent limitations of accounting and challenges in identifying the most relevant economics to portray.
- Sometimes the accounting seems to fail to reflect the economics for no good reason. In our view, the recognition of expected credit losses under US GAAP is a prime example.
- Prior to the recent change to US GAAP the purchase of a portfolio of loans could result in a day-2 accounting loss even though no economic loss was incurred. This has been amended, and from 2027 this distortion should be largely eliminated from financial statements.
- Loan loss accounting will remain problematic under US GAAP (although less obviously so) for originated loans, where the day-2 loss effect will remain.