The COVID-19 pandemic and responses to it will have large and long-lasting effects on bank asset quality. Across the 88 banking systems S&P Global Ratings covers, we forecast their credit losses will be about $1.3 trillion in 2020--more than double their 2019 level of $0.6 trillion (see chart 1). In line with our economists' forecasts of a broad, strong economic recovery into 2021, we expect losses in that year will fall back to a more manageable $0.8 trillion--though this would still be more than one-third above the 2019 level. Indeed, we expect that 2019 marked the end of a multiyear period of benign credit losses for banks globally.

Chart 1

Here, 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.

Our forecasts are based on our analysts' estimates for nonperforming loans (NPLs) and related provisioning. Regulatory forbearance, jurisdiction-specific rules, or bank management decisions regarding slowing the pace of loss recognition, will affect the timing and recognition of credit losses in banks' financial reporting. In our view, and as the aftermath of the 2008-2009 financial crisis showed, delays in the recognition of credit losses by banks, or a lack of transparency in reporting such losses, could undermine investor confidence in banks and may delay the path to recovery for some countries.

S&P Global Ratings acknowledges a high degree of uncertainty about the evolution of the coronavirus pandemic. The consensus among health experts is that the pandemic may now be at, or near, its peak in some regions, but will remain a threat until a vaccine or effective treatment is widely available, which may not occur until the second half of 2021. We are using this assumption in assessing the economic and credit implications associated with the pandemic (see our research here: www.spglobal.com/ratings). As the situation evolves, we will update our assumptions and estimates accordingly.

Yearly Global Credit Losses To Increase By Over 75% In 2020-2021, With Large Regional Differences

We expect that on a global scale, bank credit cost ratios in 2020 will be around 160 basis points (bps), more than double their 2019 level of 78 bps (see chart 2). We estimate this ratio was around 100 bps to 120 bps in the aftermath of the 2008-2009 global financial crisis. That said, we note that current accounting rules require a more timely recognition of credit losses than during the financial crisis, the composition of global lending is more weighted toward developing-market economies (including China) that tend to have weaker asset quality, and that the financial crisis 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. Consistent with our expectations for an economic rebound in 2021, we forecast credit costs in 2021 will be around 95 bps--a level that is still about one-fifth higher than in 2019.

Chart 2

Our projections for credit losses vary widely among regions, in size and timing of their recognition (see charts 2 and 3). Of the $926 billion total increase in credit losses we forecast over this year and next (from the 2019 level), Asia-Pacific accounts for $518 billion, dominated by $398 billion of losses in China. In large part, this reflects the sheer size of the Chinese banking system in a global context. 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 relatively much more important in the supply of credit to its economy than the U.S., where borrowers can typically benefit from a deep, liquid, and mature bond market, a large nonbank financial institution sector, as well as access to funding via the banking system. North America's regions account for a further $240 billion of the increase, followed by $120 billion in Western Europe.

These differences can arise for myriad reasons. China was first hit by the outbreak, but this quickly came under control and the economy has improved since the peak in the first quarter of this year. The large loan exposures to state-owned enterprises (which tend to assume some levels of government support) buffer the potential loan quality fallout. Substantial monetary and targeted fiscal stimulus also helped stabilize confidence and contribute to bankers' having an improved view of recovery. Sectorwide, there are substantial loan provisions--about 1.8 times the level of official NPLs. However, this is much lower, at about 50%--once our estimate for forborne loans is included. Banks in China are required to comply with strict regulatory provision standards once these forborne loans migrate to weaker classifications. In addition, some listed Chinese banks must consider IFRS 9 provisioning requirements and apply the higher of the two requirements.

Chart 3

From a credit analysis perspective in the context of the pandemic, governments' actions to support their citizens (fiscal measures), regulatory actions to support borrowers (forbearance measures) and banks' disclosures (financial reporting norms) are all important elements in informing our assessments of bank asset quality. Another factor that influences differences in expected loss levels and their timing is the relative split of bank lending by sector among banking systems, with certain unsecured or corporate lending activities, for instance, likely to be more exposed to sharp spikes in provisioning than mortgage and other secured loans.

For each of these factors, the impact can differ among banks and banking systems. For example, social support programs in many Western European countries--in terms of employment protection and welfare benefits (in particular for unemployment or illness) tend to be a relatively more supportive factor for asset quality than in some other countries.

Major Banks Will Be Able To Absorb Credit Losses From Earnings, Though With Reduced Headroom For Further Upticks

We estimate that the top 200 rated banks represent about two-thirds of global bank lending. Pro rata, for 2020 we estimate that credit losses for these banks would absorb about 75% of their preprovision earnings. Under our base case, this ratio improves to about 40% in 2021. By way of comparison, in 2019 the ratio was just 30%. These figures illustrate how the sharp uptick in credit losses, combined with muted earnings from slower economic activity, leaves limited capacity to fully absorb further losses from current earnings.

Should the COVID-19 pandemic prove to be worse or last longer than our current base case economic forecasts assume (see table 1), then a combination of higher credit losses and lower earnings will inevitably hit banks across the world. In particular, our economists note that the economic damage associated with the pandemic is nonlinear: If containment of the virus takes twice as long as expected, for example, the economic damage will be more than twice as bad as the length of the recession, with recovery longer and weaker (with more lost output). Moreover, even if the spread of the virus were to end tomorrow, the effects could linger, especially if social distancing becomes a new normal or business and consumer spending doesn't bounce back as people await the arrival of a vaccine.

Table 1

Although we've revised our GDP forecast for China in 2020 sharply downward from our earlier projections, it is nevertheless the only major region globally where we expect positive GDP growth in both 2020 and 2021. While we expect the U.S. economy will also rebound strongly in 2021, the contraction in 2020 is significant. Western Europe faces an even deeper contraction in 2020, with a strong though still partial rebound in 2021. The decline in output across all regions may well test rating boundaries for banks as asset quality weakens. In addition, the likely wind-down of government fiscal support and loan forbearance programs (such as short-term repayment moratoriums) may reveal new fragilities in asset quality.

Fiscal Measures Provide Essential And Immediate Support To The Economy, But Their Eventual Withdrawal Could Reveal New Fragilities In Asset Quality

The unprecedented level of fiscal support that many governments across the world have deployed in response to the pandemic-related slowdown has been a key factor in supporting their citizens and economies during lockdown periods (see chart 4). Time will tell whether the size and duration of such support has been effective enough. From a bank credit risk perspective, perhaps the greater danger at this time is the reduction of such support too early, resulting in a longer and deeper economic contraction, further impairing banks' asset quality and increasing credit losses. We do not see evidence of this yet, but even under our base case, the recovery will take time, with lower GDP growth (and higher unemployment) for a number of years.

Chart 4

Borrower Forbearance, Intended To Bridge Short-Term Liquidity Constraints, May End Up Masking Declining Asset Quality

Many banks across the world have introduced payment moratoriums or other forms of borrower forbearance as part of responses to the effects of the pandemic. Whether mandated by authorities or led by the banking industry, the aim of the forbearance is to help stabilize borrowers' creditworthiness--in other words, to prevent borrowers who are facing short-term liquidity shortfalls from becoming insolvent and defaulting on repayments.

Borrower forbearance has been effective in the past, such as after natural disasters. But the extent of the current, pandemic-related, forbearance activity is unprecedented in scale. Broader concerns about social stability and the public confidence effects of pandemic-related actions such as lockdowns may well have a played a role in the size of current forbearance programs. Regulators and standard-setters have demonstrated a relatively transparent and flexible approach in the application of accounting and regulatory rules in relation to forbearance activity, encouraging banks to consider the effects of the expected relatively rapid rebound in economies rather than just the near-term economic downturn.

Indeed, forbearance activities are generally intended to be short term--a three- or six-month repayment holiday has been a common forbearance feature in many countries--in the expectation of a relatively rapid economic recovery. As such, forbearance is not necessarily an automatic indication of a decline in asset quality, for now. Even so, we note there have been extensions to original payment holidays in a number of countries, as the end of lockdown periods and the starting path to recovery comes later than initially anticipated. That is why we are mindful of the risk that protracted borrower forbearance activity ends up masking declines in underlying asset quality, though we are not in that scenario now, in our view.

Estimations Of Expected Credit Losses In Banks' Financial Reporting

A key aspect of bank financial reporting that is 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 of 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 this year 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 (CECL), 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 first quarter have been markedly higher than some European banks', for example. While that may weigh on U.S. banks' earnings and capital initially, it should mean their provisions will be lower in future periods than otherwise.

For banks reporting under IFRS, a crucial question is whether loans under forbearance measures are experiencing a significant increase in credit risk since origination, because that is the trigger for the loan moving from a 12-month ECL provisioning requirement to a lifetime ECL. We think this is unlikely in the near term, since we understand that interest will continue to accrue during forbearance activity such as loan repayment holidays, which means that the present value of contractual cash flow on the loan is unlikely to fall below the original contractual amounts. Over a longer time frame, that assessment would likely become more difficult to justify, potentially resulting in a surge in new credit costs.

What's On The Horizon For Bank Credit Losses, By Region

Below, we set out a synopsis of credit loss forecasts for regional banking systems across the world over the next two years to end-2021.

Asia-Pacific: Asset quality metrics won't fully recover until 2023

We forecast credit losses of $1.266 trillion over the two years to end-2021 for Asia-Pacific banking systems. Given its size, China accounts for more than one-half of the region's loans and more than three-quarters of these losses. For China, we assume that asset quality pressures are frontloaded by maintaining a reasonable level of loan loss provisions to nonperforming asset (NPA) coverage. India has the second-highest NPA level, albeit ranking fifth in system loans. India's system loans contribute to less than 5% of loans in the Asia-Pacific.

We expect that credit metrics for the region's banking systems as a whole may not recover to 2019 levels until 2023. For India and Indonesia, the path to recovery from the pandemic may be more painful than in some other Asia-Pacific banking jurisdictions. This is because India is seeing a deep recession and had high NPLs leading into the pandemic, and Indonesian corporates are burdened by its slow economy, weak commodity prices, and high foreign currency debt; the effects of these factors may spill over on banks. The ratings outlook bias for Asia-Pacific financial institutions is tilted toward the negative, reflecting the effects from the pandemic, oil price shocks, and market volatility.

Nevertheless, asset quality for banks in the region (excluding India), profitability (excluding Japan), and capitalization are in generally good shape, even as the pathway for the virus remains uncertain. Still, we anticipate that bank capital buffers will be significantly tested over the next six to 18 months.

North America: Robust uptick in loan provisions driven by regulatory requirements

We forecast credit losses of $366 billion over the two years to end-2021 for the U.S. and Canada. This compares with losses of just $63 billion in 2019. For the U.S., our base case estimate is that loan loss rates will be about half the roughly 6% level the U.S. Federal Reserve has projected in its nine-quarter severely adverse scenario (which is unrelated to the pandemic) in its last two annual stress tests. We expect banks to provision for those losses faster than banks in most other regions, and particularly in 2020, because of this year's implementation of the CECL accounting methodology.

In fact, in the first quarter of 2020, U.S. FDIC-insured banks reported a robust $53 billion in provisions for credit losses, and some banks have indicated that provisions could be just as high in the quarter ended in June. As a result, the ratio of allowances for credit losses to loans and leases rose to 1.80% at March 31, 2020, from 1.18% at the end of 2019.

We underscore the fluidity of the current situation and the difficulty in forecasting credit losses. Whether our estimate proves too conservative or lenient will depend heavily on the performance of the economy and the continued effectiveness of support measures the government has provided individuals and businesses. Our estimate factors in the U-shape, gradual recovery our economists have set as their base case as well as the assumed benefits of government support measures.

As part of the stress test results released in June 2020, the Fed also provided a sensitivity analysis where it estimated potential losses related to the pandemic under three scenarios for the 33 large banks that were part of the test: 8.2% under a V-shaped recession, 10.3% under a U-shaped one, and 9.9% under a W-shaped recession. Those figures are all higher than the 6.8% loss rate during the global financial crisis. Importantly, however, the Fed's analysis did not factor in any benefit from government support measures.

Western Europe: Forbearance and fiscal support should prevent an even larger rise in credit losses

We forecast credit losses of $228 billion over the two years to end-2021 for Western European banking systems, compared with actual losses of just $54 billion in 2019. We don't discount the possibility that credit losses will remain elevated beyond our forecast horizon to end-2021.

We expect that losses up to end-2021 will be spread broadly evenly across this year and next. This is in contrast to our forecasts for North America and Asia-Pacific, where we expect about two-thirds of credit losses to arise in 2020. The difference primarily reflects the relatively less volatile nature of Western European economies, with broadly more generous social welfare support for households--enhanced further in the near term as a response to the pandemic-related economic slowdown--and greater employment protections. These factors also play a role in ameliorating the extent of higher credit losses overall. Similarly, banks' forbearance activities, such as offering borrowers (primarily households, but also in some cases, small and midsize enterprises) short-term payment holidays, play a role in preventing households' short-term COVID-19-related liquidity problems from ultimately turning into insolvency issues.

Still, as our estimates indicate, we don't expect that the region is immune from a downturn this year, and it is likely that the creditworthiness of some borrowers will not recover. These borrowers may include, for example, certain employees and companies in industries most severely affected by the pandemic, such as commercial aerospace, autos, and nonessential retail. The ultimate size of credit losses depends on the speed and magnitude of the rebound, and we expect this will become more evident as payment holiday schemes wind down later this year. All in all, bank provisioning will likely peak in the second or third quarter, but could persist at an elevated level well beyond this.

We see bank transparency about asset quality metrics as a key element for investor confidence in banks. We expect that IFRS 9's reporting requirements will be helpful in this regard, though a lack of comparability across banks will likely remain less than ideal. That said, the European Banking Authority has rapidly mandated standardized disclosure templates for the extent of loans subject to pandemic-related forbearance and new loans granted that carry a government guarantee. These disclosure templates, which must be published twice a year by banks, take effect from end-June and are intended to end in December 2021.

Latin America: Despite higher credit losses, the region's larger banks will remain profitable

We forecast credit losses of $131 billion over the two years to end-2021 for banking systems in Latin America, with the bulk of these from the region's largest economies, Brazil ($75 billion) and Mexico ($18 billion). For the region as a whole, we expect that credit metrics may not recover to 2019 levels until 2022.

Brazilian banks have focused their growth strategies in the past two years on granting payroll-deductible loans to government employees, mortgages with conservative loan-to-value ratios, and corporate loans with stronger guarantees. As they were recovering from the previous crises (2015-2016), banks have maintained sound provisioning coverage. However, a number of corporates entered the pandemic in fragile shape, and we expect unemployment will rise. As such, we expect credit losses will substantially increase in 2020, with some moderation in 2021 as economic conditions recover. On the other hand, our estimate of losses for Mexican banks consider their relatively low credit growth prior to the crises, low leverage in the economy (as measured by credit to GDP), and relatively high provisioning coverage. For both countries, it is important to note that credit losses in 2020 measured in U.S. dollars are lower than in local currency due to the effect of currency depreciation.

Because of payment moratoriums offered by regulators, we expect NPLs to take longer to surface on banks' balance sheets, but loan loss provisions will be recognized sooner. We expect asset quality to be sharply affected as a result of the economic downturn, in particular for more sensitive economic sectors, and especially for small and midsize enterprises and self-employed workers. These groups have limited financial flexibility to cope with a sudden stop in cash flow, considering the high level of informality in the region. However, given the short-term nature and relatively low proportion of mortgage loans in the region, asset quality deterioration tends to surface faster--and recovery occur sooner--as banks charge off the loans and clean up their balance sheets. Given high margins, we expect large banks to remain profitable during the next two years, though with a significant hit to revenues, while smaller banks may incur losses.

Banks in the region are used to operating under volatile economic and political conditions, with healthy earnings and regulatory capital buffers. Still, much depends on the duration and intensity of the pandemic's effects on the region's economies.

Central And Eastern Europe, Middle East And Africa: Navigating through COVID-19 and the oil price shock

We forecast credit losses of $142 billion over the two years to end-2021 for banking systems in Central and Eastern Europe (CEE) and the Middle East and Africa (MENA). Of this total, $40 billion relates to the Russian banking system, $27 billion to Turkey, and $23 billion to Gulf Cooperation Council (GCC) countries.

In addition to the asset quality deterioration that we expect for banks in developed markets, some banks in emerging markets, including in CEE and MENA, are exposed to risks such as:

• Heavy reliance on external funding amid changing investor sentiment;
• Concentration of their economies on specific sectors (such as the hospitality sector or industrial or service exports to developed countries) or commodities (such as oil or gas); and
• Lack of government capacity to provide extraordinary support, weaker governance and efficiency of government institutions, and a higher likelihood of political and social tensions.

In Russia as well as in GCC countries, the oil price shock will compound the impact of the COVID-19 pandemic on banks' asset quality and profitability, despite measures implemented by governments to contain the coronavirus. For certain emerging-market banks, such as in Turkey, currency fluctuations will likely contribute to the deterioration in asset quality and capital, due to the still high level of foreign currency loans, in addition to the expected pressure on funding costs as banks continue to roll over external debt.

GCC governments have announced several measures to help corporates and retailers navigate the challenging environment. Some governments have opted for reduced taxes and levies. Other have asked banks to extend additional subsidized loans to affected clients to maintain employment and avoid production capacity destruction. To our knowledge, no regional government has announced wide measures that would reduce credit risk on banks' balance sheets or inject additional capital. As a result, we think that risks will continue to build and ultimately weigh on banks' financial profiles if the crisis worsens. In this environment, we see banks in the UAE, Oman, and Bahrain as the most exposed.

We think that Qatari banks' asset quality indicators will deteriorate, but the large state footprint in the economy will act as a backstop. In Saudi Arabia, we also think that asset quality indicators will deteriorate, especially given that current conditions come at a time when the private sector was already under significant pressure. However, we note that most bank growth in recent years was spurred by mortgages to Saudi nationals, who are predominantly working for the government. As the pilgrimage season is being restricted, we think the shock to the private sector will be stronger and the effect on asset quality indicators will be more visible.