IFRS 9 and the loan loss lottery

As reserves for bad loans balloon, banks grapple with measuring Covid-era credit risk

  • Covid-19 has wrought havoc on banks’ attempts to estimate lifetime loan losses, with models struggling to compute the effect of government support packages.
  • IFRS 9 requires banks to increase provisions when loss estimates worsen. Regulators are telling them to look past those projections and keep credit flowing.
  • Provisions for Covid-19 vary greatly between banks and across regions. Five of the largest eurozone banks have set aside $9 billion to cover loan losses. In the US, the biggest eight banks expect $35 billion of losses.
  • In its latest results, Barclays gave greater weight to its extreme scenarios – both good and bad – due to increased economic uncertainty.

When global standard-setters were crafting their new accounting regime for expected credit losses, little did they know that an economic shock would turn the rules on their head so soon.

Under IFRS 9, banks must set aside provisions to cover losses over the life of a loan when the likelihood of default increases. The regime has a simple philosophy: when credit risk rises, loan reserves rise.

But the coronavirus crisis has prompted a rethink. Regulators in Europe and Asia are now pressuring banks to ignore core parts of the rules in order to free up capital to support new lending. Their message is that banks shouldn’t assume credit risk is rising even if a borrower is unable to meet repayments, because economies will rebound later this year or early next year.

Lenders are not convinced. They are wary of steeper than expected losses if government relief efforts miss the mark, and the anticipated recovery fails to materialise.

“[Regulators] are trying to say, ‘don’t automatically assume that there’s been a significant increase in credit risk in what otherwise would be a good business going forward’,” says Trevor Derwin, a partner at Deloitte, and an IASB advisory council member. “The problem with that is, it’s easy to say – but it does mask the impact of solvency issues beyond liquidity issues.”

Not every regulator is on the same page. The Danish Financial Supervisory Authority, for instance, has cautioned against watering down loss provisions under IFRS 9. “We believe that losses should be recognised early, as they appear on the horizon, in order to deal with them in a timely manner,” says Jesper Berg, director general of the Danish FSA.

US banks are subject to a different version of IFRS 9 known as the current expected credit loss (CECL) standard. Observers say the US Federal Reserve has been less explicit about how banks should apply the standard during the coronavirus crisis, although the regulator has allowed banks to delay the capital impact of higher provisions until 2022.

With so much uncertainty over the shape of the economic recovery, banks have reported varying levels of loss provisions. UK bank Barclays, for example, booked £1.6 billion ($2.1 billion) in provisions in the second quarter of 2020. This represents 1.79% of its total loan portfolio.

Germany’s Deutsche Bank, on the other hand, set aside provisions of €761 million ($890 million) in the second quarter, 0.69% of its loan book.

There are not many forecasts that are valid because we are still in the beginning of what can happen in the downturn. To update the IFRS 9 models is the biggest challenge we have

Louise Lindgren, Lansforsakringar Bank

Part of the difficulty in calculating provisions is that the pandemic has thrown banks’ models into disarray. Risk modelling uses previous economic scenarios to inform default predictions. Covid has exceeded many of the worst-case scenarios, forcing banks to resort to best guesswork to retune their model parameters.

Models also rely on macroeconomic factors like GDP and unemployment figures as inputs. However, a key driver of credit risk today is state support, which is hard to incorporate into model projections.

“When we do IFRS 9 modelling no one was thinking about this situation. No one had this in their models,” says Alexander Petrov, head of corporate and internal ratings-based models at Nordea Bank.

Balancing act

IFRS 9 differs from its predecessor, IAS 39, in introducing three stages of loan impairment. In stage 1, banks must set aside 12 months of provisions for all loans at point of origination. If the loan suffers a “significant deterioration” in credit quality, it moves to stage 2 and must have lifetime provisions. Loans in stage 3 are fully impaired, with lifetime provisions and a reduction in expected interest payments.

As the default risk of a loan portfolio changes, banks must decide which loans to move between stages of impairment, increasing or decreasing provisions accordingly.

The extraordinary circumstances of the coronavirus crisis have presented central banks and regulators with a dilemma. If they allow banks to follow the IFRS 9 rules to the letter, loans will migrate en masse into higher stages of impairment, resulting in huge spikes in provisions. This will insulate banks from the ill effects of future defaults but choke off capital that could be lent to cash-starved companies.

Governments see lending as key to preventing a deep recession and aiding a fast recovery. So regulators are asking banks to be conservative with stage 2 and stage 3 movements and use optimistic recovery scenarios in their modelling. The European Central Bank and the Bank of England, for example, are urging banks not to use payment moratoria as a trigger to automatically move loans to stage 2.

In effect, banks are being told to gloss over current economic realities. According to economists at the International Monetary Fund, global growth is projected to decline 4.9% in 2020 before rebounding by 5.4% next year. However, this would still leave global GDP in 2021 around 6.5 percentage points lower than projected before Covid-19.

Regional economic projections are variable, reflecting the different approaches governments have taken to combatting the economic harm of the pandemic.

In the US, the unemployment rate stood at 11.1% in June, up from a long-term average of around 4%. By comparison, in the eurozone, where governments took stronger action to keep workers in jobs during lockdowns, unemployment rose to 6.7% in May.

Economists at JP Morgan expect the eurozone economy to contract 6.4% in 2020, compared to a 5.1% decline in the US. However, the eurozone is projected to bounce back more strongly, with 6.2% growth in 2021, compared to 2.8% growth in the US.


The variations in economic outlooks may help to explain differences in provisioning between the two regions’ banks. Loss provisions at the eight US systemically important banks topped $35 billion in the second quarter of the year. By contrast, five of the eight systemically important banks in the eurozone set aside $8.9 billion in the same period. The remaining three had not yet reported their latest results by time of publication. The UK’s five largest banks provisioned $12.4 billion in the second quarter.

The gap may, in part, be attributed to differences between IFRS 9 and CECL. The US regime lacks the three-stage approach of IFRS 9, and instead banks must provision for lifetime expected credit losses at the point of origination.

“US banks have had fairly significant increases in their credit loss provisions. In Europe, there are also significant increases, but not to the extent of the US banks,” says Deloitte’s Derwin, speaking before the latest second-quarter earnings announcements. “The ECB has been fairly vocal whereas I haven’t seen as much out of the Fed, so maybe European banks feel they have support to be less conservative than US banks.”

Even within Europe, provisions vary sharply by bank. Barclays increased its credit impairment charges from £523 million in Q4 2019 to a whopping £2.1 billion in Q1 2020. Of this, £1.2 billion resulted from a revision to its baseline scenario to reflect Covid, including peak unemployment rates of 17% for the US and 8% for the UK.

In its latest Q2 results, the bank lowered its provisions to £1.6 billion. It notes that its new baseline scenario recognises continued government support measures. But it has given greater weight to its most extreme scenarios – both positive and negative – in recognition of what it describes as “the significant range of uncertainty in the economic environment”.

Deutsche Bank’s set-asides have climbed less spectacularly since Q4 2019 to reach €761 million by July. To arrive at this figure, the German bank shifted its economic forecasting horizon from two years to a three-year smoothed average, with a base case V-shaped recovery. The effect is to dampen the increase in provisions that would have otherwise been required under IFRS 9.

Impaired vision

Regardless of what scenarios they develop and apply, credit experts concede their current forecasts are unlikely to be accurate.

Although regulators are encouraging banks to consider the effect of government programmes on long-term creditworthiness, it’s difficult to estimate the impact given the uncertainty around the timing and strength of the economic recovery.

Some banks are erring on the side of caution. A credit risk manager at one international bank says the lender took substantially higher provisions than its model predicted. “We don’t subscribe to the view that just because payment holidays and moratoriums are announced, there will be no defaults, no need to increase provisions dramatically, and no need for stage 2 transfers,” the manager says.

In order to estimate the impact of government stimulus programmes, banks are using management overlays, a form of expert judgment. Cynics might call it guesswork.

“We don’t have a precedent to use as a benchmark for either the stimulus programme, or the economic recovery that will hopefully follow. We are in uncharted waters,” says a credit executive at a large North American bank.

Even ignoring the difficulty of modelling the effectiveness of state support, banks lack historic data of a similar global pandemic in modern times to crunch the numbers. “There are not many forecasts that are valid because we are still in the beginning of what can happen in the downturn. To update the IFRS 9 models is the biggest challenge we have,” says Louise Lindgren, chief risk officer at Lansforsakringar Bank in Sweden.

We don’t have a precedent to use as a benchmark for either the stimulus programme, or the economic recovery that will hopefully follow. We are in uncharted waters

Credit executive at a large North American bank

Regulators have acknowledged the problem. The European Banking Authority and European Central Bank, as well as local competent authorities, have issued guidance that the flexibility which is baked into IFRS 9 should be used to the full extent. The framework makes allowance for qualitative criteria, such as override capabilities, and regulators want banks to use that to navigate the Covid crisis.

“In that sense, the communication of regulators has turned around 180 degrees,” says Daniela Thakkar, methodology and member executive at industry trade body Global Credit Data. “Before the crisis, they all said ‘we see too much qualitative overrides, too much outside of the model’ – and now they’re saying ‘use these elements in order to avoid procyclicality’.” 

Based on peer group discussions, Thakkar says banks are adjusting more frequently their risk model outputs and applying model overlays based on expert judgment. The overlays attempt to take into account the specific impact of the pandemic on sectors at risk as well as the balancing positive effects of governmental support measures.

Even the Danish FSA, which has taken a stronger line on sticking to IFRS 9, has cautioned against taking a mechanistic approach to loan loss provisions. In case of forbearance and payment holidays, while it requires banks to assess the state of loans, “there is no automaticity to move the loan to stage 3,” says Berg.

He adds: “We require the banks to use scenarios that include the impact of Covid-19 on their loan book. Specifically they can use the scenarios that the Danish central bank published after the outbreak of Covid-19. These are quite harsh scenarios, although they include an eventual recovery.”

What regulators are saying

European Banking Authority The EBA says the application of public or private moratoria, aimed at addressing the adverse systemic economic impact of coronavirus, should not be considered by themselves as an automatic trigger to conclude that a significant increase in credit risk has occurred.

UK Prudential Regulation Authority The PRA expects firms to reflect the temporary nature of the shock, and fully take into account the significant economic support measures already announced by global fiscal and monetary authorities. Eligibility for the UK government’s policy on the extension of mortgage repayment holidays should not automatically be a sufficient condition to move participating borrowers into stage 2 of IFRS 9.

Bafin offices in Bonn
BaFin offices in Bonn

Germany BaFin In the event of delays to payment due to the coronavirus crisis, institutions should adopt a through-the-cycle perspective that takes into consideration measures by the state to mitigate the economic impacts of the crisis.

France AMF Covid measures such as payment holidays or additional loans do not automatically constitute an indicator of a significant increase in credit risk. Issuers should assess whether relief measures are enough to offset the presumption of an increase in credit risk in the case of payment delays. French companies should take into account the positive effects of relief measures in their forward-looking macroeconomic data used to determine longer-term estimates of expected credit losses.

Austria Financial Markets Authority The FMA recommends that banks apply the transitional rules for IFRS 9. Banks should take a medium-term perspective. In the case of delayed payments due to coronavirus, a through-the-cycle perspective should be taken, which also takes into account government measures for mitigating economic consequences.

Hong Kong Monetary Authority The HKMA says estimates of expected credit losses should reflect the mitigating effect of economic support and payment relief measures. The provision of relief measures to borrowers should not automatically result in exposures moving from a 12-month ECL to a lifetime ECL measurement. Additionally, institutions are expected to exercise informed judgment and to use the flexibility inherent in HKFRS/IFRS 9 to give due consideration to long-term economic trends in estimating ECL.

Editing by Alex Krohn

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact info@risk.net or view our subscription options here: http://subscriptions.risk.net/subscribe

You are currently unable to copy this content. Please contact info@risk.net to find out more.

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here