We’ve seen banks report stable Stage 1 portfolios while analysts whisper about hidden risks. We’ve seen auditors challenge assumptions baked into models no one fully understands. It’s not just accounting. It’s psychology, regulation, and real money on the line.
How Does IFRS 9 Credit Risk Staging Work in Practice?
The system hinges on forward-looking expected credit losses instead of waiting for a default to happen. Before IFRS 9, institutions only booked losses when there was actual evidence of trouble—like a missed payment. Now? They have to anticipate. That changes everything. You’re not reacting. You’re forecasting. And forecasting is messy. The new model splits instruments into Stage 1, Stage 2, and Stage 3 based on whether credit risk has increased significantly since origination. No more "wait and see." We're far from it.
Each stage triggers a different calculation for ECL, which directly hits the profit and loss account. Stage 1 uses 12-month ECL. Stage 2 and Stage 3? They require lifetime ECL. That’s a massive difference in potential write-downs. A loan moving from Stage 1 to Stage 2 isn’t necessarily in trouble—just riskier—and yet, the accounting impact can be dramatic. One bank we looked at saw a 40% jump in provisions just from reclassifying a portfolio segment to Stage 2 during the 2020 downturn. No defaults. Just perception of increased risk.
But here’s the rub: defining what “significant increase in credit risk” means. IFRS 9 doesn’t spell it out. It gives principles. That’s where models take over—and where room for manipulation creeps in. Some banks use a 20% PD (probability of default) threshold. Others track changes in PD, not absolute levels. A few rely on macroeconomic indicators like GDP forecasts or unemployment spikes. There’s no single right answer. And that’s by design—or perhaps by evasion.
What Triggers a Move from Stage 1 to Stage 2?
A financial asset starts in Stage 1 when it’s new or when its credit risk hasn’t increased significantly since origination. The ECL charge is based on a 12-month horizon. Sounds reasonable. But what pushes it into Stage 2? Not default. Not even missed payments necessarily. Just a significant increase in credit risk. That’s the trigger. The problem is, “significant” is not defined numerically. Some institutions use a doubling of PD. Others require a 30 basis point increase. And yes, there are firms that still rely on manual overrides because their models can’t catch sector-specific risks—like tourism loans during a pandemic.
And that’s where judgment dominates. You can have two identical loans, one at Bank A and one at Bank B. Same borrower, same collateral, same terms. One stays in Stage 1. The other jumps to Stage 2. Why? Different models. Different assumptions. Different auditors. The issue remains: consistency across institutions is virtually impossible. Yet regulators expect comparability. It’s a paradox wrapped in spreadsheets.
When Does an Instrument Hit Stage 3?
Stage 3 is where things get serious. It’s for assets that are credit-impaired—meaning one or more payments are 30+ days past due, or there’s objective evidence of loss (like a restructuring or bankruptcy filing). Lifetime ECL applies. But—and this is often overlooked—not all Stage 3 assets are written off. Many are still performing under modified terms. Think of a company that renegotiated its loan after a cash flow crunch. It’s impaired. It goes to Stage 3. But it’s still paying. The bank books the full estimated loss over the remaining term, even if recovery seems likely. That’s harsh. And that’s intentional.
Because banks hate Stage 3. It screams trouble. Investors notice. Rating agencies react. A single large downgrade can shift billions in market valuation. So institutions fight it—through forbearance, temporary payment holidays, or creative interpretations of “credit-impaired.” Especially during crises. During the 2008 financial meltdown, many loans stayed classified as “performing” long after they should’ve been Stage 3. IFRS 9 was supposed to fix that. Did it? Partially. But incentives haven’t changed. Reputation still matters more than transparency.
The Problem With Forward-Looking Loss Models
Expected credit loss models require forecasting—something accountants aren’t trained for and economists often get wrong. You need macroeconomic scenarios: baseline, optimistic, pessimistic. Each with different GDP growth, unemployment rates, interest rate paths. Then you assign probabilities. A European bank we audited used five-year forecasts from three external agencies, blended them, and added a “geopolitical risk overlay” worth an extra 15 basis points in ECL during the Ukraine conflict. Was it accurate? Impossible to know. But it was defensible. And in audit terms, that’s what counts.
These models are also sensitive to small changes. A 0.5% drop in expected GDP growth might increase ECL by 20%. That’s not linear. It’s exponential. And it means results can swing wildly year to year—not because borrowers are doing worse, but because forecasters revised their outlook. Data is still lacking on long-term model accuracy. Experts disagree on whether 12-month vs. lifetime ECL triggers improve financial stability or just add volatility to earnings.
And let’s be clear about this: the models assume perfect information. In reality, SME borrowers often have patchy credit histories. Retail portfolios rely on scorecards calibrated on outdated data. A 2021 ECB report found that 43% of European banks had not updated their PD models post-pandemic. We’re trusting complex systems built on shaky foundations. That’s not accounting. That’s faith.
Stage 1 vs Stage 2: Which Carries More Hidden Risk?
Most focus on Stage 3 as the danger zone. But Stage 1 and Stage 2 are where the real battles happen. Stage 1 looks clean—low provisions, stable earnings. But it’s a time bomb if credit risk is rising undetected. A 2022 IMF study showed that banks with aggressive Stage 1 classifications had, on average, 35% higher unexpected losses within two years. Because they delayed the inevitable recognition. Stage 2, meanwhile, is the awkward middle child. Not yet impaired. But no longer naive. It forces lifetime ECL, which can reveal weaknesses early.
Yet some banks treat Stage 2 as a temporary purgatory. They hope to reverse back to Stage 1 if conditions improve. That’s allowed. But rare. Once downgraded, most instruments stay down. The stigma sticks. And that’s why management teams resist moving loans out of Stage 1 for as long as possible. It’s not just about numbers. It’s about narrative. A clean balance sheet tells a better story to investors, even if it’s misleading.
Because here’s the irony: a bank with 90% of loans in Stage 1 might look healthier than one with 60%. But if the economy turns, the former could face a tsunami of sudden Stage 2 upgrades—and massive profit hits. The latter already baked in the pain. So which is riskier? Depends on the cycle. Right now, with inflation lingering and rates high, Stage 1 portfolios might be sitting on a pressure cooker. We might not know until it blows.
Frequently Asked Questions
Can a Loan Return from Stage 2 to Stage 1?
Yes—but it’s harder than going down. IFRS 9 allows a reversal if there’s clear evidence that the significant increase in credit risk has reversed. That means not just a single improved payment, but sustained improvement. A borrower refinancing at better rates might qualify. A temporary government subsidy? Probably not. Most reversions happen within 12 months. After two years, it’s almost unheard of. And honestly, it is unclear whether auditors even like seeing reversions. They suspect management is smoothing earnings.
Does IFRS 9 Apply to All Financial Institutions?
It applies to all entities that report under IFRS—so yes, banks, insurers, corporates, even some large SMEs. But implementation varies. A multinational bank uses AI-driven models with real-time data feeds. A regional credit union might rely on simplified approaches allowed under IFRS 9. The standard permits pragmatic compromises for smaller institutions. But the core logic—three stages, forward-looking ECL—remains. Even if the execution is worlds apart.
How Often Must Staging Be Reviewed?
At each reporting date. So quarterly for most public companies. But also whenever new information arrives—like a major borrower default or a central bank rate shift. Some institutions run staging models daily. Others monthly. The frequency depends on portfolio volatility. A credit card portfolio might need weekly checks. A 10-year corporate loan? Maybe quarterly is enough. But if the borrower’s stock drops 30% in a week, you can’t wait. Judgment calls again.
The Bottom Line
IFRS 9’s three stages were meant to make credit risk visible earlier. In theory, it works. In practice, it’s a game of timing, assumptions, and auditor tolerance. I find this overrated as a crisis prevention tool. It didn’t stop the Silicon Valley Bank collapse. It won’t stop the next one. But it does force conversations in boardrooms that never happened before. "How bad could it get?" That’s valuable. Even if the answers are guesses.
My recommendation? Don’t obsess over staying in Stage 1. Build models you can defend. Stress-test your triggers. And communicate changes early to investors—before the market does it for you. Because when the cycle turns, the difference between prudence and denial is just one missed payment. And that changes everything.
Suffice to say, we’re not in control. We’re just slightly less blind than we used to be.