Across the 86 banking systems that S&P Global Ratings covers, we expect credit losses will amount to around $1.8 trillion over the three years to end-2023 (see chart 1). China's commercial banking system accounts for just over half of this total, reflecting the sheer size of that system in a global context, the relatively greater reliance of the Chinese economy on bank financing relative to capital market financing, and stringent provision-coverage regulations for nonperforming loans (NPLs).

In aggregate, the picture for bank credit losses is better than we had previously expected (see chart 2). The latest data indicates that such losses amounted to around $635 billion in 2021, just under one-fifth below our previous forecast. Our revised forecast for 2022 is just under $585 billion, around one-third lower than our previous forecast. This publication also includes our first global forecast for 2023, where we expect losses of around $595 billion.

Compared to our earlier forecasts, this improved picture--and the regional variations we discuss later in this report--highlights that credit losses are moderating, even though they will likely remain somewhat elevated compared to pre-pandemic lows.

Chart 1

Chart 2

We use the term credit losses to refer to the income statement charge (in U.S. dollars) by which banks add to their balance sheet provisions or allowances for expected losses on domestic customer loans, plus any direct write-offs of domestic customer loans. Banks often describe credit losses as "provisions for" or "charges for expected credit losses," among other similar terms. Credit losses generally precede charge-offs, the actual write-down of loans that detract from the balance sheet allowances for credit losses. The cost of risk (or credit cost ratio) refers to credit losses as a proportion of customer loans.

The recent rapid spread of the omicron variant highlights the inherent uncertainties of the pandemic as well as the importance and benefits of vaccines. While the risk of new, more severe variants displacing omicron and evading existing immunity cannot be ruled out, our current base case assumes that existing vaccines can continue to provide significant protection against severe illness. Furthermore, many governments, businesses, and households around the world are tailoring policies to limit the adverse economic impact of recurring COVID-19 waves. Consequently, we do not expect a repeat of the sharp global economic contraction of second-quarter 2020. Meanwhile, we continue to assess how well each issuer adapts to new waves in its geography or industry.

Credit Cost Ratios Are Declining From Pandemic Highs

We estimate that on a global scale, banks' credit cost ratios will average around 65 basis points (bps) in 2022 and 2023. This is an improvement from the pandemic-related high of just over 100 bps in 2020 and the 2021 ratio of 77 bps. We see the 2021 ratio (which is comparable to the pre-pandemic level of 74 bps in 2019) as reflective of substantial writebacks of 2020 provisions in some regions, as the worst of the credit-negative effects of the pandemic did not materialize. This was largely due to government fiscal support measures to households and businesses and temporary borrower forbearance measures encouraged or mandated by the authorities across much of the world. We saw credit cost ratios fall across every region in 2021 except China, and even in that case the ratio remained broadly unchanged. The most significant fall was in the U.S. (where the ratio was negative) and Western Europe, where the ratio fell by around 40%.

By way of comparison, we estimate that the credit cost ratio was around 100-120 bps in the aftermath of the 2008-2009 global financial crisis (GFC). That said, a comparison with credit losses in the GFC is not a like-for-like assessment. This is because of factors such as current accounting rules, which require a more timely recognition of credit losses than the rules in place during the GFC, as well as the composition of global lending, which is now more weighted toward developing-market banking systems (including China) that tend to contend with higher economic risk factors and often have weaker asset quality. The nature of the support mechanisms is also very different from in the GFC, as governments moved rapidly to support households and businesses rather than needing to directly support banks. The GFC also had a more limited effect on loan asset quality in some regions, including Asia-Pacific, for example, than we expect to be the case now.

Our projections for credit losses vary widely among regions (see charts 3 and 4). Globally, we expect that the total increase in credit losses from their level in 2019 will amount to around $235 billion over the three years to end-2023. (By increase, we mean the sum of the differences in projected credit losses for each of 2021, 2022, and 2023 compared to the actual credit losses for 2019.)

Chart 3

Chart 4

China alone accounts for the vast majority of this increase. This largely reflects the sheer size of the Chinese banking system in a global context and stringent provision-coverage regulations for NPLs. In terms of customer loans, the Chinese banking system is approximately the same size as the U.S., Japanese, German, and U.K. banking systems combined. Moreover, the banking system in China is much more important in supplying credit to its economy than in the U.S., where borrowers typically benefit from a more mature bond market, as well as access to funding via the banking system.

While property is weighing more heavily on Chinese commercial banks' asset quality, this impact has been partially offset by COVID-19-related forborne loans performing better than expected. We think that credit costs should begin to stabilize this year, compared to 2019 pre-COVID levels as banks release provision coverage. Write-offs are likely to remain elevated to keep the reported NPL ratio stable. Our forecasts put annual credit losses at Chinese renminbi (RMB) 2 trillion on average over the four years to 2023, down from our previous estimate in August of RMB2.5 trillion. Even by end-2024, however, the nonperforming asset (NPA) ratio will still be about half a percentage point higher than the pre-pandemic metric. Over this period, the provision coverage on NPLs is likely to remain above 180% (197% coverage as at fourth-quarter 2021), and for NPAs above 65%. The decrease in this coverage takes some pressure off credit costs and is consistent with a dynamic provisioning framework.

Western Europe accounts for a further $64 billion of the increase in credit losses compared to pre-pandemic levels. We note, though, that credit losses in this region were at a historical low in 2019, and a return to those low levels isn't likely over our forecast period to end-2023. Overall, asset quality has held up better than we previously expected during the pandemic, thanks in large part to unprecedented fiscal support which has averted widespread problems. While some problem loans may still emerge, we are confident that credit losses will remain manageable, after peaking in 2020. As our forecasts indicate, we expect losses to remain around their current levels.

In contrast, we expect that the run rate of credit losses in North America over the three years to 2023 will be $97 billion lower in aggregate than losses in 2019. This includes negative credit losses for the U.S. banking system in 2021, thanks to large writebacks of 2020 provisioning in 2021. We expect positive credit losses in the U.S. in 2022 as reserves relative to loans have already fallen near pre-pandemic levels, leaving limited room to lower them further. Accelerating loan growth will also require banks to provision more for credit losses.

Most Major Banks Will Be Able To Comfortably Absorb Credit Losses From Earnings

We estimate that the top 200 rated banks represent about two-thirds of global bank lending. For these banks, we estimate that credit losses will absorb around 25% of their 2022 pre-provision earnings on average--this is the same level as the average in 2019. Moreover, on a regional basis we estimate that the median level of credit losses to earnings will range between just 12% in North America to 35% in Latin America (see chart 5). These figures illustrate the relatively healthy level of capital across the top 200 banks in aggregate and their capacity to absorb further losses without depleting capital.

Chart 5

Still, the pandemic is not fully over yet, and even as it recedes the path to recovery could be worse or take longer than we assume in our base case (see table 1). Some of the key downside risks for asset quality include real estate asset bubbles forming in some markets or a disorderly reflation scenario that exposes asset and market risks.

Such setbacks could lead to higher credit losses as more borrowers become distressed, as well as lower pre-provision earnings from reduced economic activity. This combination of events would inevitably affect many banks across the world.

Table 1

What's On The Horizon For Bank Credit Losses?

Below, we outline our credit loss forecasts for regional banking systems across the world to end-2023.

China: As COVID-19 asset quality pressures subside, new risks emerge

We forecast credit losses of $530 billion (RMB3.4 trillion) for the two years to end-2023 for China.

Defaults are rising among China's property developers and we estimate that roughly one-third are in some form of financial trouble. This means that just as COVID-19-related stresses are easing, property-related issues are keeping banks' NPAs elevated.

Authorities have fine-tuned credit-tightening requirements to provide temporary relief. While there is contagion within this sector, the spillover effects to general loan quality are so far limited, though this is still a developing story.

While property weighs more heavily on bank asset quality, this has been partially offset by COVID-19-related forborne loans performing better than expected. We think that credit costs should begin to stabilize this year as banks release provision coverage. Write-offs are likely to remain elevated to keep the reported NPL ratio stable. Property development loans are typically well collateralized with real estate. With our view of a moderate 3% nationwide decline in residential prices for next year, the pressure to write off loans will be limited. Having said that, weaker regions could experience larger price drops. Banks with aggressive risk appetites or high geographic concentration could be left with thinner collateral buffers, especially those with more recent vintages. We do not expect challenges related to property development loans to materially affect residential mortgages and note that the timely delivery of homes under construction is a priority for authorities.

We also note that, regardless of ultimate loss expectations, banks have to post sizable provisions if a loan is classified as nonperforming. As a general rule, this is about 150% of new NPLs, and could water down profitability.

Asia-Pacific excluding China: Losses return toward pre-pandemic levels

We forecast credit losses of $175 billion for the two years to end-2023 for the Asia-Pacific region, excluding China. We estimate that 2023 annual credit losses will be well down from the 2020 peak and will be approaching 2019 pre-pandemic levels. We expect some variability in the forward trajectory of credit losses across the region. Negative economic trends affecting the Malaysian, Indonesian, and Thai banking systems will weigh on credit losses. Further, high loans under moratorium in these jurisdictions remain a latent risk to asset quality. In broad terms, however, the intervention by public authorities spanning all banking systems in Asia-Pacific, including fiscal policy and monetary policy support, has worked. Banks should continue to demonstrate some resilience at current rating levels even as extensive relief measures are progressively removed. Further, our current base case is that likely higher interest rates across many jurisdictions in 2022 and 2023 will be broadly manageable at current rating levels. Sturdy provisioning during the onset of the pandemic has helped banks to manage their credit standing and will likely continue to underpin credit ratings at current levels in most jurisdictions as the economic rebound takes hold.

North America: Credit losses to rise following 2021 reserve releases, but remain benign

We forecast credit losses of around $104 billion for the two years to end-2023 for North America. In the U.S., we expect those losses to rise from last year's negative levels given that reserves for credit losses have already fallen close to pre-pandemic levels. With loan growth accelerating, banks will have to add to those reserves. Still, credit losses should remain fairly benign in 2022--below 2019 levels--before normalizing to around historical averages as a percent of loans in 2023. Similarly, in Canada, we expect provisions to rise in 2022--but to remain below 2019 levels--and then to inch higher in 2023.

Western Europe: Credit losses to stabilize

We forecast credit losses of $150 billion for the two years to end-2023 for Western Europe, evenly split between the two years. After peaking in 2020, in anticipation of higher delinquencies which have not yet materialized, credit losses declined by 40% in 2021 and are set to remain at similar levels in 2022 and 2023. While higher than pre-pandemic levels, credit losses will likely be benign overall, accounting for about 35 bps of loans. Our forecasts take into account that, although asset quality deterioration has been minimal so far, some problem loans may yet emerge. Borrowers--largely corporate--with pre-existing weaknesses may still experience difficulties, particularly in the likely scenario of interest rates rising. We expect lending growth to remain contained, so unlikely to drive meaningfully higher provisions in absolute terms. Equally, we don't expect to see a general release of provisions like we saw for U.S. banks in 2021.

Latin America: Credit losses are returning to a new normal

We forecast credit losses of $100 billion for the two years to end-2023 for banking systems in Latin America, with the bulk of these in the region's largest economies, Brazil ($50 billion) and Mexico ($17 billion). Asset quality metrics remain stronger than expected. While we believe that they will weaken slightly--due to expected flagging economic growth, a still sluggish labor market, and modest credit growth this year--declines in asset quality should remain manageable, thanks to banks' conservative growth strategies that were implemented prior to the pandemic.

Central and Eastern Europe, the Middle East, and Africa: Declining cost of risk across much of the region

We forecast credit losses of around $120 billion for the two years to end-2023 for banking systems in Central and Eastern Europe and the Middle East and Africa.

In Gulf Cooperation Council (GCC) countries, the cost of risk has declined because banks hold good provision cushions to meet expected NPL increases. At Sept. 30, 2021, the average coverage ratio remained stable at 147.7%. We expect this ratio to reduce slightly in 2022. The NPL ratio continued to increase, reaching 3.7% at Sept. 30, 2021, compared with 3.1% in 2019. We expect NPLs to continue increasing, but not exceed 5% on average. We expect the cost of risk to normalize over the next couple of years and margins to benefit from the expected increase in interest rates.

For Turkish banks, we expect the cost of risk will rise to 320 bps on average in 2022 and 2023, from an already high 280 bps on average in 2019-2021. The real estate market's performance is a crucial factor in assessing the amount of losses banks will experience, given that they make large use of real estate collateral, and we expect that real estate prices in Turkey will stabilize in real terms.

Credit losses have started to decrease in South Africa, reaching an estimated 1.7% in 2021 from 2.1% in 2020--still higher than historical levels of under 1%. We expect the NPL ratio will increase to about 5% of systemwide loans and credit losses will gradually improve closer to 1% by 2023. Household leverage continues to pose a risk for banks because of eroded wealth and savings levels. Commercial real estate has shown growing signs of stress since 2018, but we expect this to ease marginally as economic activity resumes.

We expect the Russian banking sector to continue its gradual recovery, provided there are no new material economic or geopolitical shocks. These projections reflect frontloaded monetary policy normalization in 2021 coupled with faster credit growth, supported by economic recovery. In 2021, corporate credit growth accelerated in Russia, supporting new business generation for banks, and we expect this trend to continue in 2022. At the same time, retail loan growth--particularly in uncollateralized consumer loans--will continue, although the rate of growth will slow to about 15%-20%. This will continue to support banks' margins. We anticipate that the NPL ratio will likely improve to 7.5%-8.5% and credit losses will stabilize at about 0.6%-0.8% in 2022-2023 after an estimated 1% in 2021.

Estimations Of Expected Credit Losses In Banks' Financial Reporting

A key aspect of banks' financial reporting that is still relatively new--and broadly untested in a downturn--is the estimation of credit losses. Both International Financial Reporting Standards (IFRS, the financial reporting rules applicable across much of the world outside the U.S.) and U.S. Generally Accepted Accounting Principles (U.S. GAAP) require banks to take a more forward-looking approach when estimating credit losses on their loan portfolios. The new rules took effect in 2018 for banks reporting under IFRS and in 2020 for large banks reporting under U.S. GAAP.

The two sets of rules are markedly different in their approach. In broad terms:

• The U.S. GAAP model is based on an estimate of current expected credit losses, which requires banks to set reserves for expected lifetime losses on their entire loan portfolio.
• The IFRS model, expected credit loss (ECL, as set out in IFRS 9), requires a dual measurement approach under which a 12-month ECL allowance is established for performing loans and a lifetime ECL for underperforming and NPLs.

These rules mean that as banks increase their expectations for future credit losses, they may have to increase reserves markedly. And the difference between the two approaches is likely one reason why U.S. banks' credit losses in the early stages of the pandemic were markedly higher than some European banks', for example.

Related Research

Global:

• Global Banks Outlook 2022: Back On Course, Dec. 14, 2021
• Top 100 Banks: Capital Ratios Show Resilience To The Pandemic, Sept. 28, 2021

Asia-Pacific:

• Asia-Pacific Banks Outlook 2022: The Long And Winding Road (To COVID Recovery), Jan. 25, 2022
• China's Banks Face A Doubling In Real Estate NPLs, Dec. 15, 2021

North America:

• Industry Credit Outlook: U.S. Bank Outlook 2022: Higher Rates To Provide Boost In Uncertain Times, Jan. 12, 2022

Western Europe:

• The Top Trends Shaping European Bank Ratings In 2022, Jan. 31, 2022
• European Housing Market Inflation Is Here To Stay, Nov. 2, 2021

Latin America:

• Emerging Markets: Latin America Sector Round Up, Oct. 21, 2021
• Resilient Earnings and High Provision Coverage Position Latin American Banks Well To Face Tough Conditions, March 9, 2021