Judgment in Financial Instruments (IFRS 9 / US GAAP)
Financial instruments accounting, governed by IFRS 9 on the international side (and a patchwork of standards in US GAAP, recently updated for credit losses as ASC 326), is a domain notorious for complexity and judgment. IFRS 9 introduced sweeping changes to how companies classify financial assets, when and how they recognize impairment (credit losses), and how they account for hedging activities. The goal was to be more principles-based and forward-looking than its predecessor (IAS 39), but in doing so, IFRS 9 placed greater reliance on management’s estimates and assumptions. Let’s explore a few key areas: classification of instruments, impairment (expected credit losses), and hedge accounting.
Classification of financial assets under IFRS 9 depends on two things: (1) the business model for managing the assets, and (2) the contractual cash flow characteristics (the so-called SPPI test – solely payments of principal and interest). Unlike the old rules that had many categories (held-to-maturity, available-for-sale, etc. with bright lines like “intent to hold to maturity” strictly defined), IFRS 9 uses principle-based categories: amortized cost, fair value through OCI, or fair value through P&L, determined by how the asset is managed and what its terms are. Judgment is needed to assess the business model, especially when it’s not clear-cut. For example, a bank might generally hold loans to collect interest, but occasionally sells some loans or securitizes them. Is the business model “hold to collect” (amortized cost) or “hold and sell” (which would be FVOCI)? IFRS 9 acknowledges there can be some sales in a hold-to-collect model (for example, selling a loan for risk management or due to credit concerns doesn’t taint the model). Management must analyze past sales and future strategy to decide the predominant intent. This involves judgment about future actions and interpretations of “infrequent” or “insignificant” sales. Two banks with similar portfolios might differ – one might designate certain portfolios as hold-to-collect-and-sell if it has a routine practice of liquidity-driven sales, while another with rare sales might still consider it hold-to-collect. The decision has big accounting outcomes (fair value changes hit equity vs not recognized). Regulators have generally allowed these differences if justified, but they do monitor if firms might be using business model classification to achieve desired accounting (which is not easy, since classification is done at initial recognition and not supposed to be changed arbitrarily). The SPPI test is another area of judgment: an asset must have cash flows that are only principal and interest on principal outstanding (interest being basically time value of money and credit risk compensation) to be eligible for amortized cost or FVOCI. Many instruments have quirky features – e.g. interest linked to a commodity price, or a leverage factor. IFRS 9 says such features likely make cash flows not “SPPI”, requiring fair value through P&L. But what if the feature is tiny or very unlikely to affect cash flows (what IFRS 9 calls de minimis or non-genuine features)? The standard does not quantitatively define “de minimis.” Therefore, significant judgment is needed to decide if a contractual term that could introduce variability in payments is so minor that it can be ignored. For instance, a loan might have an interest rate that bumps up by 0.1% if the borrower’s ESG score drops below a threshold. Is that a genuine risk factor altering cash flows more than de minimis? One might argue 0.1% is trivial and rare, so the loan still passes SPPI. Another might be stricter. IFRS 9 provides concepts (non-genuine means only trigger on an “extremely rare, highly unlikely” event; de minimis means impact is negligible), but no hard rule – thus banks and firms had to set internal thresholds and justify them. One consulting analysis noted, for example, that evaluating if a non-SPPI feature is de minimis “requires significant judgement as de minimis is not defined in IFRS 9.”. Many companies documented policies like “we consider a cash flow variability of less than X% of the base case to be de minimis” or similar, to consistently apply SPPI judgments. Auditors review these and often test a sample of new instruments to see if they agree with the SPPI conclusion. Generally, classification judgments happen at initial adoption and for new products, and once set, the accounting follows mechanically. But getting it right is crucial; a misclassified asset could mean later restating if, say, an instrument thought to be SPPI actually wasn’t (hence should’ve been fair-valued with P&L volatility). Fortunately, IFRS 9’s principles, while involving judgment, align with intuition in most cases (traditional loans, bonds = SPPI; equity-like or derivative-like = not SPPI). Still, the edges keep finance departments on their toes.
The impairment of financial assets under IFRS 9 – specifically the Expected Credit Loss (ECL) model – is arguably one of the most judgment-intensive accounting changes in decades. Gone is the old “incurred loss” model where you only recognized credit losses when there was objective evidence of impairment. IFRS 9 (and similarly US GAAP’s CECL model) require that from day one, entities estimate expected credit losses over the next 12 months, and if credit risk significantly increases, over the lifetime of the asset. This forward-looking approach means management must build models of the future – predicting credit defaults and losses, incorporating macroeconomic forecasts, and updating these at each reporting date. Determining what constitutes a “significant increase in credit risk” (SICR) is a pivotal judgment. IFRS 9 does not provide a bright line (like a specific drop in credit rating or fixed change in probability of default); it gives a framework (compare risk of default at reporting date vs initial recognition, consider qualitative and quantitative factors, and there’s a rebuttable presumption that if an asset is 30 days past due, credit risk has significantly increased). Each bank or lender had to develop criteria for SICR – often a combination of relative changes in credit scores/ratings, watchlist status, etc. As PwC observed, “Under IFRS 9, a significant increase in credit risk, not defined [explicitly], requires lifetime ECL recognition… and is a highly judgmental area.”. In fact, they called assessing SICR “a critical and highly judgemental area” of ECL. For example, one bank might decide that if a loan’s probability of default has doubled since origination (say from 1% to 2%), that’s significant; another might set a threshold of tripling, or use letter-grade notches (e.g. a downgrade of 2 rating grades). Regulators like the European Banking Authority have tried to foster consistency by comparing how banks define SICR, but some variation is inherent. Once assets move to lifetime loss (Stage 2), the estimation of lifetime expected losses requires further judgment: What will the economy look like next year, or 5 years from now? IFRS 9 expects use of multiple scenarios (good, base, bad) with probabilities. Determining those scenarios and weights is part art, part science. For instance, in an economic downturn, how much do you weight a pessimistic scenario vs a quick recovery scenario? During the COVID-19 pandemic, this became extremely challenging – no historical precedent matched, so banks had to overlay expert judgment on top of models. Many institutions added management overlays or post-model adjustments, essentially manual adjustments to model output to account for things the model might miss (e.g. government interventions, moratoria on loans, etc.). These overlays are pure judgment, though informed by data. An insight from 2020-2021 was that banks’ ECL estimates varied widely, leading to difficulty comparing their financial health. S&P Global noted, “The variability of management’s judgment when determining expected credit losses makes it difficult to compare banks’ health and has caused confusion in the market.”. In other words, even though all banks were following IFRS 9, one bank’s rosy view might yield much lower provisions than another’s cautious view, making it hard for analysts to discern which was actually sounder without digging into the assumptions. This is a classic consequence of a principle-based approach: flexibility allows tailoring to one’s situation, but at the cost of comparability and potential skepticism if judgment appears overly optimistic. To mitigate this, banks disclose qualitative and quantitative information about their assumptions (e.g. what GDP drop they assumed in the adverse scenario, or how sensitive ECL is to unemployment changes). Auditors too devote significant effort to credit loss models – often involving credit risk experts to re-perform calculations and challenge scenario designs. Yet, ultimately, ECL numbers are an estimate of an uncertain future, so they will never be “right,” only reasonable. IFRS 9 expects firms to update these judgments each period as more information comes – a dynamic process that truly tests professional judgment on an ongoing basis.
US GAAP’s CECL model (Current Expected Credit Loss), implemented in 2020 for many institutions, takes a similar forward-looking approach but without the staging of IFRS 9 (under CECL, it’s lifetime expected losses from day one for all instruments). In a way, CECL removed one layer of judgment (no need to decide significant increase in credit risk – everything is treated as having lifetime loss), but it still requires forecasting lifetime losses. For long-term assets, this arguably involves even more judgment (predicting say 30-year losses on a mortgage portfolio upfront). IFRS proponents argue their approach is more principles-based by tying recognition to credit deterioration, but either way, the estimation uncertainty is huge. Both frameworks rely on management judgment, and indeed both saw substantial management overlays in 2020 due to COVID uncertainty. It’s interesting to note that the banking regulators had to issue guidance encouraging consistency and use of realistic assumptions – essentially reminding management to exercise sound judgment and not overly delay recognizing losses or, conversely, not overreact either (since huge swings could harm lending unnecessarily). This underscores that professional judgment in accounting for credit losses also has real-world impact on behavior and stakeholder confidence.
Hedge accounting under IFRS 9 is another area improved in principle (more principle-based than IAS 39) but still reliant on judgment calls. IFRS 9 allows hedging strategies to be reflected in accounting more easily, as long as they meet risk management objectives and effectiveness criteria. One significant judgment is defining what risk is being hedged and how effectiveness will be assessed. While IFRS 9 removed the rigid 80–125% effectiveness bright line, it requires an economic relationship between hedging instrument and item and no systematic bias (no deliberate under-hedging or over-hedging). Companies must use judgment to designate hedges in a way that faithfully reflects their risk management – for example, deciding to hedge only the benchmark interest rate component of a debt (leaving credit spread unhedged) or hedging a net position of risk (allowed under IFRS 9). They also choose methods to assess effectiveness (could be statistical or qualitative). These choices determine if hedge accounting can be applied and whether mismatches in fair value will be deferred in OCI or hit P&L. IFRS 9 is more accommodating (e.g. you can hedge components of non-financial items, which was harder before), but it put the onus on companies to justify that the hedge is effective and appropriate. That includes forecasting the hedged item’s exposure (for cash flow hedges) – e.g. if you hedge forecasted sales in USD, you must assert those sales are highly probable. That probability assessment is a judgment about future business volumes. Especially in volatile times, auditors scrutinize whether “highly probable” was reasonable (e.g. airlines hedging fuel might claim a certain level of fuel consumption – if a pandemic hits and travel plummets, some hedges might no longer qualify because the purchase forecast wasn’t actually probable). In recent years, many firms had to discontinue hedges when their forecasts changed drastically, illustrating that judgment areas can have consequences like accelerated recognition of losses that had been deferred.
Global example – Banking: A European bank implementing IFRS 9 in 2018 had to overhaul its credit risk systems. Let’s say they historically had an internal credit grading and used that for loan loss triggers under IAS 39 (incurred loss). Under IFRS 9, they created criteria such as: if a loan’s internal rating drops by X notches, or if the 12-month PD exceeds a threshold relative to origination PD, then SICR is triggered. They also decided any forbearance (loan restructuring for borrower trouble) is a SICR. These judgments were informed by regulatory guidance and back-testing (they wanted to capture most loans that eventually default in Stage 2 earlier). Another bank might have set different notch thresholds or relied more on delinquency. When COVID hit, suppose this bank quickly overlaid an additional adjustment: they knew government payment holidays were masking which loans were truly in trouble, so they used macroeconomic models to estimate how many loans would have been past due if not for holidays and moved some portion to Stage 2 preemptively. Another bank in a different country might have taken a wait-and-see approach. The outcome: Bank A had a big increase in Stage 2 loans and provisions early in 2020, Bank B smaller. By end of 2021, Bank A perhaps releases some provisions as the situation clarifies, whereas Bank B increases theirs once government support ended. Both might argue they used reasonable judgment given what they knew – and both could be compliant with IFRS 9’s principles. However, investors following these banks needed to understand their differing philosophies. This example underscores how judgment in accounting can lead to different timing of recognition, even if ultimately both cover the losses that occur. It also highlights that transparency about assumptions is key; IFRS 7 requires disclosures of credit risk assumptions, and regulators in Europe (ESMA, etc.) asked for enhanced disclosures during COVID to allow users to compare and trust the numbers.
Global example – Corporate (Treasury): A manufacturing company might use forward contracts to hedge forecasted purchases of copper (a key raw material) which are priced in USD. Under IFRS 9, they can hedge the combination of FX risk and commodity price risk or just one of them. Let’s say they decide to hedge only copper price risk in USD terms, not the FX separately, because they have natural USD exposure elsewhere. IFRS 9 allows hedging that component (copper price) even though it’s part of a broader purchase contract. The company needs to document its risk management objective and how it will assess effectiveness – perhaps by showing that changes in copper forwards correlate highly with changes in their copper purchase cost. They also must judge that the forecast purchases of copper are highly probable (maybe they forecast volumes based on sales projections). If later demand falls, they may have “excess” hedges. Under principle-based hedge accounting, the company can discontinue hedge accounting for the excess – but determining when a forecast is no longer highly probable is a judgment call. Some might cut off early (as soon as trouble signs appear), others might wait until nearer the delivery date. Both IFRS 9 and ASC 815 require continuous monitoring of forecast transactions’ probability. That experienced treasury accountants and auditors apply significant judgment to these determinations is evident in practice – especially in 2020, many airlines and fuel hedges were in question when flights were grounded (was the forecast fuel consumption still expected?). It took careful consideration and discussions with auditors/regulators to decide at what point those hedges had to be deemed ineffective because the underlying forecast shrank.
In conclusion, IFRS 9 (and analogous GAAP standards) present a more principles-driven approach to financial instruments that grants companies flexibility to reflect economic substance (like expected losses, risk management activities) more faithfully. But with that flexibility comes heavy reliance on models, assumptions, and judgment. The standard explicitly requires this judgment and expects companies to use all reasonable and supportable information available – a phrasing that again highlights the breadth of consideration (including forward-looking info) that judgment must encompass. We can clearly see that poor judgment or bias in this area has serious implications: it could mean underestimating credit losses (and getting a nasty surprise later, possibly drawing regulatory ire or even threatening the company’s stability), or overestimating them (which might unnecessarily alarm investors or restrict credit unwisely). Indeed, banking regulators globally have been vigilant that ECL models not be used to hide problems nor overshoot to create cookie jar reserves – essentially, reminding that professional judgment must be neutral and grounded in evidence. Accounting for financial instruments thus epitomizes the need for sound judgment: it’s complex, it’s forward-looking, and it deals with large dollar amounts that can swing based on subtle changes in assumptions. The next section will discuss impairment of non-financial assets, another area where estimates and judgment are paramount.