The private sector costs of COVID-19 - Capital Economics
Global Economics

The private sector costs of COVID-19

Global Economics Focus
Written by Simon MacAdam
Cancel X

The immediate costs of the COVID crisis will be shouldered more by governments than the private sector. However, as fiscal support recedes in the coming years, a greater share of the costs will be borne by households and firms, and ultimately by their creditors. Our base case is that loan losses will not be big enough to destabilise banking systems, but some economies, particularly in the emerging world, are more vulnerable than others. And even in countries where banking systems are on a surer footing, losses are likely to be big enough to act as a brake on lending, as banks work to maintain or restore capital buffers.

  • The immediate costs of the COVID crisis will be shouldered more by governments than the private sector. However, as fiscal support recedes in the coming years, a greater share of the costs will be borne by households and firms, and ultimately by their creditors. Our base case is that loan losses will not be big enough to destabilise banking systems, but some economies, particularly in the emerging world, are more vulnerable than others. And even in countries where banking systems are on a surer footing, losses are likely to be big enough to act as a brake on lending, as banks work to maintain or restore capital buffers.
  • Considered in isolation, the fact that world GDP shrank by 10% in the first half of this year – compared to a contraction of just 0.5% in 2009 – suggests that by now we should have seen a tsunami of defaults dealing a tremendous blow to the financial system. But such has been the scale of support from governments and central banks, particularly in advanced economies, that the financial fallout has been remarkably limited. Households have higher savings now than they did before the virus, firms have bridged cash-flow shortfalls with cheap finance, financial conditions are pretty much as loose now as they were before the pandemic, and bank credit losses remain ultra-low.
  • So far, so good. But the situation will soon change. While central bank liquidity support to the financial sector will remain in place, fiscal transfers and guarantees to firms and households to prop up demand are set to be reined in. Some governments – mainly in EMs – came into this crisis with weak public finances and are therefore under pressure to curb borrowing. In other countries, the scale of fiscal support provided so far is causing jitters among politicians. Costs are mounting rapidly and there are concerns about how to wean the private sector off state support. Other economies – including many DMs – are not under any particular pressure to withdraw fiscal support, and in fact will provide support for viable firms and households as long as is necessary. But as economic recoveries gain traction, support is being and will continue to be tapered, which will mean that some firms in the likes of the aviation, hospitality, and office real estate sectors will be exposed for what they are in the “new normal” – unviable and hence insolvent.
  • As governments take a step back, a greater share of the costs of the crisis will be borne by the private sector. Ultimately, this means creditors will pick up the tab. There are five reasons why we suspect that credit losses will not be large enough to destabilise the banks. First, monetary policy and central bank liquidity provision should keep financial conditions loose for borrowers and lenders alike. Second, in general, borrowers were not overextended at the outset of the crisis. Third, many of the ingredients in the recipe for a housing market meltdown – the main threat to the banks – are not in place today. Fourth, banks have limited credit exposures to some of the worst-hit sectors of the economy. And fifth, regulatory forbearance over loss recognition should limit the scale of write-downs, at least in the near term.
  • Even if credit losses turn out to be large, fortunately the Basel III accord has paid dividends – banks came into this crisis well capitalised. Our reverse stress tests of over 500 of the world’s banks suggest that if loss rates rose by a factor of ten from their 2019 global level of 0.5% to 5%, core capital ratios would stay above not just regulatory minimum but also target levels at a majority of banks.
  • While all of this leads us to believe that the overall chances of a global banking crisis are low, there are all sorts of plausible downside risks which mean that we will be monitoring this threat closely. We are fully aware that much of our analysis is largely conducted at the aggregate level, beneath which may lurk vulnerabilities at individual systemically important institutions. And, even at the aggregate level, the risks vary between countries – we are particularly concerned about the banks in India and Turkey. Perhaps the bigger risk than the world’s banks reaching some cataclysm of a financial crisis is that they enter a period of Japanification, harbouring bad debts which eat into profits and restrict lending for years to come.

The private sector costs of COVID-19

This Focus forms part of a series exploring how the pandemic will change the global economy. You can find other publications in this series here.

This Focus considers whether banks are likely to withstand the costs of defaults in the non-financial private sector. In other words: can creditors absorb the costs of the coronavirus crisis without having to restrict lending or resort to recapitalisation?

The report comprises six sections. We will start by taking stock of what governments and central banks have achieved so far in funnelling cash to firms and households and limiting losses for creditors. We’ll then set out why things are going to become more challenging in the months and years ahead. In the third section, we’ll consider five factors that should limit the financial fallout for lenders, before assessing what losses the banks are able to withstand in the fourth section. In part five, we’ll draw out cross country differences with the aid of a heat map of risk factors for banks in major economies. Finally, we will make some conclusive remarks in part six.


1. Policymakers into the breach

The evidence is clear: speedy action by governments and central banks has supported asset prices and helped see off a funding crisis that could have wiped out vast swathes of otherwise solvent businesses and plunged households into bankruptcy.

On the household side, incomes have been supported by a mixture of wage subsidies for furloughed workers, more generous unemployment benefits for the laid off, and stimulus cheques in some cases. Meanwhile, mortgage payment relief has been adopted in several countries, which has buoyed discretionary incomes still further. Rather than incomes falling in line with spending, households have been able to accumulate savings. (See Chart 1.)

Chart 1: Household Savings as % of Disposable Income

Sources: Refinitiv, Capital Economics

As for firms, in addition to drawing down credit lines, they have benefitted from government-backed loan schemes. The flow of credit has been improved by central banks’ liquidity operations, which ensured that wholesale funding markets remained functional. And to varying degrees, central banks’ targeted lending facilities have made a difference too. (See here for details.) Uptake of the Fed’s “Section 13(3)” facilities – such as the Main Street Lending Program – has been low (see here), but the ECB’s latest round of TLTROs has gone down well in the euro-zone.

All in all, in stark contrast to the credit crunch of 2008-09, businesses have been awash with credit this time around. (See Chart 2.) Now that economies have opened up and revenues have partly recovered, emergency liquidity needs have diminished, and debt has begun to stabilise in a couple of cases.

Chart 2: Level of Bank Loans to Firms (Jan. 20 = 100)

Sources: Refinitiv, ECB, CE. Break-adjusted data in US/UK/Aus/E-Z.

Companies have been making extensive use of the bond market too. Monetary expansion and explicit central bank commitments to purchase corporate bonds restored calm to credit markets in relatively short order. (See Chart 3.) At about $1.3tn, almost as many bonds were issued between March and August in DMs as in the previous three years combined.

Chart 3: Option-Adjusted Spreads of Corporate Bonds Over Sovereign Benchmarks (bp)

Sources: Refinitiv, Capital Economics

Moreover, various pockets of risk in corporate debt that we flagged back in March (see here) have not crystallised. One of those was the “BBB bulge” in the corporate bond market. This describes the trend increase in the share of outstanding bonds rated just one notch above junk status. It raised the prospect of widespread credit rating downgrades pushing up borrowing costs for downgraded firms and increasing rollover risk for existing sub-investment grade issuers. Given that banks hold corporate bonds, this was one route through which they could incur losses from defaults.

As it happens, liquidity operations and stimulus efforts have been sufficient to prevent a spike in bond issuer downgrades to junk status. The share of the bond market accounted for by the debt of these so-called “fallen angels” is around its average level this century in the euro-zone and remains low by past standards in the US. (See Chart 4.)

Chart 4: Fallen Angels’ Share of Corporate Bond Markets (%, Outstanding Face Value)

Sources: ICE BoA ML, Bloomberg, Capital Economics

Another area of vulnerability that we flagged was the leveraged loan market, to which global banks have an almost $2tn exposure. But, again, there has been a decent turnaround in fortunes. While a third of the market has been downgraded, loan prices are back to just 4% below pre-virus levels (see Chart 5), and default rates have stayed low (aside from oil & gas).

Chart 5: S&P Leveraged Loan Price Indices

Sources: S&P, Bloomberg, Capital Economics

In terms of credit losses from the write-down of bank loans, the financial fallout has so far been minimal. Even allowing for the time lags involved, given the hit to GDP early in the second quarter, one might have expected the charge-off rate on US bank loans (annualised write-downs net of recoveries as a share of outstanding loans) to have gone through the roof. (See Chart 6.) In actual fact, it barely moved – the loan loss rate ticked up to 0.57% from 0.53% in Q1, compared to an average rate of 0.51% in 2019.

Chart 6: US Loan Charge-Off Rate & GDP

Sources: Refinitiv, FDIC, Capital Economics

In the euro-zone, the charge-off rate on loans to non-financial corporations fell to a 12-year low in the second quarter. (See Chart 7.) In the UK, it fell to its lowest level since the series began in 1993, at 0.1%.

Chart 7: Net Charge-Off Rates on Bank Loans to Non-Financial Corporations (%, SA, Annualised)

Sources: Refinitiv, FDIC, Capital Economics. US non-financial corp. loans = “Commercial & Industrial” + commercial real estate.

Having taken stock of where we’ve got to in terms of the private sector costs of the coronavirus crisis, we’ll now consider why they are soon set to rise.


2. Passing the buck

The principal reason why the private sector will soon have to shoulder more of the financial burden of the crisis is that the scale of government support to firms and households is set to be reined in, which will make some existing debt difficult to honour.

Several emerging economies with precarious fiscal positions, like Brazil and South Africa, have already set out plans for austerity. Several Gulf countries have done so too. Other EMs with weak public finances may well join them in the coming months, or at least draw up plans to undertake some degree of active fiscal restraint. These include the likes of India and various Latin American economies.

Even in EMs with healthy public finances – such as most of CEE and emerging Asia – as well as in advanced economies, governments will not go on spending like they have done in 2020. Partly, this just reflects the fact that economies are recovering, and automatic stabilisers are going into reverse. If this were the whole story, then falling government assistance would not be a problem for the economy.

The issue in some countries is that government support is likely to taper before private demand has fully recovered. Indeed, there are growing concerns about ballooning government debt and private sector dependency on the state, as well as a general desire to try to steer fiscal policy back to something like normal. This means less generous income support schemes for households, expiring mortgage relief, and less generous debt relief and loan guarantees for firms and banks, rather than programmes of extensive spending cuts and tax rises. Even so, in terms of the effects on the economy, more firms and households will start slipping through the less generous fiscal safety net. This is reflected in our forecasts for unemployment rates to rise or stay elevated in the coming years. (See Chart 8.) Defaults will rise as a result.

Chart 8: Unemployment Rates (%)

Sources: Refinitiv, Capital Economics

Even with governments that do not find themselves under much pressure to withdraw fiscal support will do so as their economies make headway on the path to recovery. As the threat from the virus fades, and fiscal lifelines are reined in, the day of reckoning will come for firms and households whose business and jobs will be unviable in the new normal.

Fragile labour markets, subdued wage growth, and smaller government transfers mean that households’ disposable incomes will take a hit. Weaker income growth will undermine households’ capacity to service their debt, including mortgages. So, while households, and the firms they work for, will take the initial brunt of the retreat of government, some of these costs will be passed onto the financial sector.

Defaults on residential mortgages are potentially a big deal for banks because they are typically one of the largest credit exposures on the balance sheet. While house prices have held up well in recent months (see Chart 9), weaker labour markets often go hand-in-hand with weaker housing markets. Consequently, there is a chance of house-price corrections down the line. Lower house prices mean lower collateral values with which banks can recover unpaid loans, increasing the scope for credit losses.

Chart 9: Residential Property Prices
(% change from Q1 average level, SA)

Sources: Refinitiv, CEIC, Capital Economics

Banks are not just facing higher defaults on residential mortgages – their portfolio of commercial mortgages is likely to struggle too. The apparent success of the COVID-induced remote-working experiment leads us to believe that half of office-based employees will continue to work from home at least once a week from next year, up from 10-20% prior to the pandemic in advanced economies. (See Chart 10 for the US.) The fall in demand for office space should see vacancies rise markedly in the next few years and remain high by the end of the decade. As government support fades, office rental income is likely to fall markedly in the near term. And we doubt that rents will have recovered to 2019 levels even by the end of the decade. (See here.) A long period of subdued rents and property values means that banks face a high chance of write-downs on commercial mortgages.

Chart 10: Share of US Office-Based Workers Working At Home (%)

Sources: US Census, Capital Economics

The office real estate sector is not the only area of the economy to be dealt a structural blow to demand. Aviation, hospitality, and the tourism industries in general will find business life – and paying back debt – especially hard in the coming years, due to lingering restrictions on international travel, social distancing measures, and caution on the part of would-be holidaymakers. As government support recedes, and revenues in these industries fail to recover to anywhere near pre-virus levels, bankruptcy and forbearance will be the two options available, both of which impose costs on banks.

So, the upshot is that governments are passing the buck to the private sector. And the banks know it. Surveys show that a growing number of commercial banks are starting to tighten credit standards on loans to firms and households. And the recent surge in loan loss provisions suggests that banks are bracing themselves for a wave of defaults. (See Chart 11.)

Chart 11: Bank Loan Loss Provisions (€/$ Billions)

Sources: Refinitiv, Company Reports, Capital Economics

There is a lot of uncertainty about how well households and firms will cope servicing their debts when policymakers step back. But there are five reasons why we are cautiously optimistic that losses will not get high enough to endanger global financial stability.


3. Five mitigating factors

  1. Central banks to offer lasting support

While central banks have pretty much reached the limits of what they can do to further assist the recovery, what they have already achieved will have lasting positive effects. The loosening of financial conditions that they have orchestrated is still feeding through to lower interest rates on loans to firms and households. This means lower rollover risk, and ultimately lower default risk. Going back further in time, what central banks have achieved in conjunction with prudential regulators over the past ten years is to shore up bank lending standards. So, when credit stress does build in the coming months and years, debtors ought to be in a better position to cope than in decades gone by.

What’s more, while there is little more central banks can do to speed up the recovery, they can help to prevent it from slowing down when governments scale back support. When fiscal policy does become less accommodative, central banks have given every indication that they will keep yields low and spreads tight. And while lending facilities have had mixed take up so far, they will be there to be tapped into if businesses once again run into cash flow problems. Admittedly, some – such as the Fed’s 13(3) facilities – are scheduled to expire soon. But central banks have discretion over whether they extend or restart the schemes. The support will be there if needed.

2. Borrowers generally not overextended

Not only should stricter credit assessments during the past decade have reduced credit risk, but in terms of the quantity of debt, borrowers’ balance sheets were not in too bad a state coming into the crisis.

Household leverage has risen in some DMs that did not suffer big house price declines in the wake of the financial crisis. But, in major DMs, households took to strengthening their balance sheets after housing crashes a decade ago. Household indebtedness has risen in most EMs in recent years, but it is only high in the parts of Asia that have suffered the least economic disruption from the virus. (See Chart 12.)

Chart 12: Household Debt as a Percentage of GDP

Sources: Refinitiv, BIS, Capital Economics

As mentioned earlier, in previous work we showed that business finances on the eve of the pandemic were broadly in good shape, in aggregate, but that pockets of weakness do exist. One such pocket is leveraged finance. This is where the main risks lie in terms of overleveraged corporate borrowers who may struggle to honour debts in the years ahead. An optimistic, and somewhat paradoxical, argument that is sometimes made in defence of leveraged loans is that their credit standards have fallen so much over the past decade that it is practically difficult for debtors to default on such loose terms, and therefore for the asset holders – including banks – to incur losses. We are not convinced by this, but it is something to bear in mind as another potential mitigant against large bank credit losses.

3. Housing market crashes unlikely

As we argued above, fragile labour markets, expiring mortgage relief schemes, and less generous fiscal transfers to households pose a risk to banks’ portfolio of residential mortgages. But while household income growth will be subdued in the coming years, few of the other ingredients of a housing market meltdown are in place.

Big house price declines usually follow periods of higher interest rates, consumer credit booms, rising housing inventory, and rapid price rises relative to incomes, rents, and general consumer prices. But we have seen few of these tell-tale signs in recent years.

4. Limited exposures to worst-hit sectors

Among banks’ business credit exposures, only a small portion relates to the sectors of the economy worst hit by lockdowns and social distancing, such as leisure and hospitality. While data on the industry breakdown of banks’ loan books are limited in the US, figures from the euro-zone, Japan and UK show that less than 10% of loans to non-financial businesses go to companies in recreation, social, accommodation, and food services. (See Chart 13.) We do not have enough detail to know the size of banks’ exposure to airlines. But given that air transport services rarely make up more than 0.5% of GDP, credit exposures to airlines will probably be only a small part of the 10% of bank loans that go to the transport, storage, and communication sector.

Chart 13: Industry Breakdown of MFIs’ Non-Financial Business Loan Portfolio (% of Total)

Sources: Refinitiv, ECB, Capital Economics

Banks’ biggest credit exposure to non-financial businesses is loans to real estate service companies, including commercial landlords who use rental income to service mortgages on office space. Unlike with residential real estate, aggregate banking sector exposures to commercial real estate are small, but they are notably higher in the US, and especially in South Korea. (See Chart 14.)

Chart 14: Commercial Real Estate Share of Loans (%)

Sources: Refinitiv, IMF, Capital Economics

Banks also have credit exposure to commercial real estate via MBS holdings. This is a bigger deal in the US than elsewhere. But even in the US, at just over $400bn at the beginning of the year, commercial mortgage-backed securities made up 8% of banks’ $5 trillion portfolio of securities. (See Chart 15.) And almost 90% of US CMBS are government-backed.

Chart 15: US Bank Securities Portfolio* ($bn, Q1 20)

Sources: Refinitiv, Federal Reserve, FDIC, Capital Economics. *Estimate of corporate bond holdings = total holdings of corporate and foreign bonds less private mortgage pass-through securities, CMOs, other structured MBS, and holdings of foreign AfS/HtM debt securities. Sum of AfS, HtM & HfT securities = $5,005bn.

Finally, in terms of exposures to highly leveraged companies, banks tend to hold the less risky debt. For one thing, half of their holdings of leveraged loans is in the form of credit facilities, which take priority among debtor liabilities and offer more flexible repayment than term loans. What’s more, in the US – where leveraged loans are most prevalent – banks seem to own term loans with lower default risk than those held by CLO managers and investment funds. (See Chart 16.) And, as for banks’ holdings of CLO securities, they typically own the AAA-rated tranches, meaning that they are first in line for repayment. (For more detail, see here and here.)

Chart 16: US Leveraged Term Loan Default Rate (%)

Sources: Federal Reserve, Capital Economics

5. Regulatory relief has created breathing space

In addition to the de-activation of counter-cyclical capital buffers and some surcharges for systemically important banks, many national banking supervisors are following new Basel guidance that relaxes regulatory standards. This should limit and smooth out the recognition of loan losses.

For one thing, provisions charged to cover expected credit losses on non-performing loans can now count towards CET1 regulatory capital this year and next. This means that the calculation of banks’ CET1 capital in 2020 and 2021 will mainly reflect accounting practices for incurred losses only.

Moreover, banks no longer need to provision for losses on loans that are subject to payment holidays, deferrals, or other short-term modifications to loan terms. This relief applies provided that borrowers had not missed repayments prior to COVID-19.

The relaxation of standards in identifying problem loans also acts to mitigate the negative effect of risk weight inflation on capital ratios. If previously sound firms suffer rating downgrades or go into arrears due to cash-flow disruption from COVID-19, then risk weights no longer have to rise on these assets. This keeps a bank’s risk-weighted assets lower and hence its capital ratio higher than it would otherwise be.

By moving the regulatory goal posts, supervisors have helped to limit losses and free up resources for banks – at least in the short term – that would have been used to cover higher provisions for loan losses and to conserve capital.

4. The banks can (mostly) take a big hit

Even if losses do turn out to be large, banks have come into this crisis well prepared to withstand a bout of defaults. As we explain in Box 1, regulatory capital is what is used to absorb bank credit losses. Owing in part to the Basel III reforms undertaken in response to the global financial crisis, banks have amassed a lot of regulatory capital relative to the size of their balance sheets during the past decade.

Chart 17 provides a snapshot of the regulatory capital ratios of the 558 banks from around the world with more than $1 billion of regulatory capital on the eve of the coronavirus outbreak. For each region, we have shown the range of capital ratios excluding extreme outliers, as well as regional aggregates.

Chart 17: Regulatory Capital as a Percentage of Risk-Weighted Assets (End-2019)

Sources: Refinitiv, CE.
Banks with at least $1bn of regulatory capital. Number of banks in parentheses. Dashed line thresholds do not account for counter-cyclical capital or Pillar II buffers, or SIB surcharges.

Box 1: How do losses affect bank capital ratios?

To recap, after the financial crisis, the Basel III accord introduced a reform agenda that aimed to beef up banks’ capacity to absorb losses. The regulations set out how much capital banks are required to hold given the size of their risk-weighted assets (RWA).

Banks do not need to hold capital to absorb losses on the likes of US Treasuries, because such securities are deemed safe and carry a zero risk weight. But loan assets of varying credit risk do carry regulatory requirements for banks to provision for losses.

If credit losses cause ratios of capital to RWA to fall below regulatory minima, constraints are imposed on operations and supervisors step in to implement a recapitalisation plan. If capital ratios do not fall as far as the regulatory minimum level, but fall below “target” levels – i.e. into capital buffers – banks are obliged to draw up capital conservation plans and are subjected to limits on pay-outs to shareholders.

Credit losses affect both the numerator and the denominator of banks’ capital ratios. Let’s start with numerator – regulatory capital. There are numerous instruments that count towards regulatory capital. But the big picture is that CET1 capital constitutes almost 80% of global regulatory capital, and this is by and large made up of two elements: accumulated retained earnings and common share capital (common equity issued). (See Chart 18.)

Chart 18: Breakdown of Banks’ Common Equity
Tier One Capital (% of Total)

Sources: Refinitiv, CE. Banks with at least $1bn of regulatory capital.

When banks incur credit losses that are large enough to erode capital buffers, they may turn to issuing additional share capital. So, the share capital portion of regulatory capital often rises when losses surge.

However, retained earnings take a hit. Retained earnings in bank capital on the balance sheet are the accumulation of profits not paid out to shareholders in each financial period. These in turn are determined by three components: loan loss provisions; dividends;

and underlying profits (pre-provision net income). When a loan ends up performing – or is expected to perform – worse than anticipated at the time it was originated, specific loss provisions are expensed to cover the costs of the write-down on the balance sheet. It is not a given that write-down charges are big enough to cause retained profits to fall and therefore enough to eat into the accumulated stock of retained earnings in bank capital. If underlying profitability is high, operational expenses are paired back, and shareholder pay-outs are slashed, capital may not be required to absorb any losses at all.

Even if regulatory capital does not fall in absolute terms, it could fall relative to RWA, causing the capital ratio to fall towards regulatory thresholds. This takes us to what happens to the denominator of the capital ratio when banks incur credit losses.

Loan losses are write-offs, or write-downs, of the book value of loans to their fair value on the balance sheet – in order words, an accounting reality check. So, mechanically, credit losses reduce the value of RWA. What’s more, in periods of credit stress, banks will often shift the composition of their asset holdings towards safer assets – that is, to assets with lower regulatory risk weights. This also acts to reduce RWA.

But credit losses also have an effect on RWA moving in the opposite direction. To varying degrees, depending on the type of financial asset, write-downs cause risk weights on impaired assets to rise, thereby boosting RWA. Even for performing assets, credit rating downgrades of the debtors in question can lead to risk weight inflation. For example, under the standardised approach to credit risk assessments, if a debtor firm is downgraded from A to BBB, the risk weight on all bank credit exposures to that entity can jump up from 50% to 100%.

The bottom line is that there are lots of moving parts to the effect of credit losses on capital ratios. Table 1 summarises the likely changes in the main elements.

Table 1: How Loan Losses & Credit Stress Affect Banks’ CET1 Capital Ratios (Green = Raise; Red = Reduce)


Source: Capital Economics

It shows that capital ratios at the overwhelming majority of the world’s banks were well above regulatory minimums and targets, as shown by the red and orange lines. Not only this, but capital ratios were typically well above an even higher, de facto threshold of 12.5% of risk-weighted assets – depicted by the black, dashed line. This is a threshold beneath which banks prefer not to slide too far, for fear of dipping below regulatory target levels and thus being obliged to draw up capital conservation plans, potentially frightening investors.

The one country that stands out with a lot of banks close to, or already some way into, their capital buffers is Japan. However, these are all small, regional banks, as depicted by the dark grey dots in Chart 19. None of them account for more than 1% of the risk-weighted assets of the whole Japanese banking sector, and collectively they make up just 15% of RWA. While some of these may well default in the coming years, we don’t believe this would pose a systemic risk to the Japanese banking system. (See here.)

Chart 19: Japanese Banks’ Regulatory Capital as a Percentage of Risk-Weighted Assets (End-2019)

Sources: Refinitiv, CE. Banks with at least $1bn of regulatory capital.

The current level of capital isn’t everything. Profitability matters too. While European banks are the most capitalised in the world, they are among the least profitable, which means that it wouldn’t take much for loss provisions to push profits into negative territory and begin eating into capital. The return on assets of German banks was already negative before COVID, though this was entirely due to Deutsche Bank. Other European, Indian, and Japanese banks also suffer from low profitability. (See Chart 20.)

Chart 20: Return on Assets (%, 2018-2019 average)

Sources: Refinitiv, CE. Banks with at least $1bn of regulatory capital.

As we explain in Box 1 and summarise in Table 1, there are many ways in which credit losses affect capital ratios, some of which move in opposite directions. Consequently, to work out what loan loss rate banks could cope with before testing their solvency, we have boiled things down for simplicity (or as much simplicity as is possible for such a technical exercise!).

We have focussed on the direct impact of loan losses on the accumulated retained earnings portion of common equity tier one (CET1, or “core”) capital, via higher loan loss provisions and lower dividends. (The second and third rows in Table 1.) So, we have assumed that common share capital as well as other types of CET1 capital – such as goodwill, pension liabilities, and exchange rate translation adjustments of the balance sheets of foreign subsidiaries – are unchanged. Given that, in reality, banks may try to raise external capital to bolster capital ratios (fifth row in Table 1), our test basically seeks to determine how big loan losses could get before banks are forced to issue additional common equity.

Note, this is not a “stress test” like those carried out by central banks. A normal stress test models the effect of an economic shock (in terms of GDP, unemployment, house prices etc.) on loss rates, capital components and risk-weighted assets to estimate resulting capital ratios. In contrast, we have effectively taken the lower limit capital ratio and backed out the “stress” (measured by loan losses) that would be required to get there from end-2019 levels.

In this exercise, banks’ capacity to absorb loan losses before their CET1 capital ratios fall below the regulatory minimum of 4.5% of RWA (excluding idiosyncratic buffers, such as SIB surcharges) can be broken down into four lines of defence:

  1. Income flow of retained earnings;
  2. Dividend payments;
  3. Surplus CET1 capital above the regulatory target ratio of 7.0% of RWA;
  4. CET1 capital conservation buffer between 4.5% minimum ratio and 7.0% target.

Chart 21 summarises what loan loss rates are needed, net of recoveries, to exhaust banks’ lines of defence. You could call this banks’ “loss absorption capacity”. For instance, globally speaking, banks would collectively need to endure an annual loan loss rate of 9.4% – the magenta diamond in the chart – before the aggregate core capital ratio would fall to the regulatory minimum level of 4.5%. Banks’ capacity to absorb loan losses is smaller in DMs outside the US than in EMs. But considering that loan loss rates of reporting banks have typically been below 1.5% for the past seven years (see Chart 22), it is encouraging to see that almost all banks would have to experience far higher loss rates for their solvency to be brought into question.

Chart 21: Banks’ Loan Loss Absorption Capacity (Loss Rate to Reduce CET1 Ratio to Regulatory Min., %)

Sources: Refinitiv, CE. Banks with at least $1bn of regulatory capital. Number of banks in parentheses. US aggregate skewed above the interquartile range by a few highly capitalised investment banks.

Chart 22: Loan Loss Rates of Reporting Global Banks* (Net Loan Write-Downs as a % of Gross Loans)

Sources: Refinitiv, CE. Banks with at least $1bn of regulatory capital. *Sample reduced from 558 to 241 banks that have consistently reported credit losses since 2012.

However, the risks almost always lie in the tails of the distribution, not in the hump. Chart 23 zooms in on the bottom 10% – the 10% of banks in each region with the lowest capacity to absorb credit losses. Even among these weakest links, banks in most regions would be able to cope with loan loss rates of over 5% before CET1 capital ratios would approach regulatory minima.

Chart 23: Loss Absorption Capacity Vs. Share of RWA (Banks with 10% Lowest Loss Absorption Capacity)

Sources: Refinitiv, CE. Banks with at least $1bn of regulatory capital.

The two exceptions are banks in Other Asia, all of which are Indian, and Other DMs, all except one of which are Japanese. As discussed above, there are some banks in Japan with low capital ratios and profitability, but these are small. By contrast, the weakest links in India make up a disproportionate share of the country’s banking sector.

We have long flagged India’s banks as a global weak spot. (See here.) Most came into this crisis with high shares of non-performing loans and low regulatory capital ratios by international standards. (See Chart 24.) Missed payments by debtors, high loss provisions, and high costs for managing a ballooning portfolio of impaired assets meant that almost half of India’s banks were loss-making in 2019.

Chart 24: Regulatory Capital & Non-Performing Loan Ratios of Indian Banks* (%, 2019)

Sources: Refinitiv, CE. Banks with at least $1bn of regulatory capital.

The upshot of this section is that, broadly speaking outside India, banks seem to have enough headroom to deal with high loss rates on loans. While most banks would take pre-emptive action to prevent capital ratios from falling so far, it is possible under the scenario we have considered for about half of them to swallow write-downs worth 10% of their loan portfolios before breaching regulatory limits. If loss rates rose by a factor of ten from their 2019 global level of 0.5% to 5%, CET1 capital ratios would stay above not just minimum levels, but also above regulatory target levels at a majority of banks.


5. Assessing country-level risks

So, to wrap up what has been discussed so far:

  • policymakers have so far limited and delayed the financial fallout from COVID-19 for the private sector, not prevented it altogether;
  • as governments step back, banks will pick up a greater share of the costs of the crisis;
  • there are five factors that will mitigate the scale of loan losses, thereby reducing the risk of a systemic banking crisis;
  • banks generally came into the crisis highly capitalised and, in most cases, are able to withstand a big jump in loan losses.

These are the general conclusions, but the picture varies across countries. The heat map in Table 2 summarises various risk factors facing banking sectors in 20 major economies. From green to red, the four colours depict how big a threat we consider each factor to be in contributing to higher loan losses

and to lower bank capital ratios: low, medium, high, and very high risks. In the bottom row, we illustrate our overall judgement to give a sense of where we consider banking sector problems are biggest. The upshot is that the risks that bank losses will be so large as to weigh heavily on regulatory capital ratios appear low in about half of major economies.

Among advanced economies, a highly indebted non-financial private sector and an expensive housing market means that risks of substantial loan losses are elevated in Canada. Highly indebted households and a big economic hit warrant some concern for banks in Australia. Bloated borrower balance sheets in Switzerland, a legacy of high NPLs, a considerable reliance on tourism and constrained fiscal policy in Italy, as well as low bank profitability in both countries mean that there are moderate risks there too. Spain is the main concern in DMs. Not only is fiscal support there likely to be more limited than elsewhere in the coming years, but the country – and inescapably its banking sector – has a high reliance on hospitality and especially on foreign tourism.

Among emerging economies, we are concerned by Turkey’s banks, particularly on the grounds that they have a high reliance on foreign debt that they will struggle to roll over. Overextended property developers and some poorly capitalised regional banks in China mean that risks there cannot be dismissed. But we suspect that Chinese policymakers would do anything necessary to save problem banks. A weak recovery, austerity, and bank exposures to beleaguered state-owned enterprises mean that there are noteworthy risks in South Africa too.

Table 2: Heat Map of Banking Sector Risk Factors in 20 Major Economies

Source: Capital Economics

Our main worry, by far, is India’s banking sector. Weak bank balance sheets, low profitability, and a limited fiscal response to the crisis all point to India’s banks struggling a lot in the years ahead. As loan moratoriums expire, and defaults inevitably rise, talks of government recapitalisations will not be far behind. The Indian banking system is probably entering a period of Japanification, which will hold back the economic recovery for many years to come.


6. Conclusion

The bottom line of all this so far is that, apart from in India, things don’t look too bad. But it would be wrong to come away from reading this report with a sense of complacency about the resilience of the world’s banks and associated financial stability risks.

For one thing, limited data mean we are restricted to conducting high-level analysis, so there could be systemically important vulnerabilities that we can’t detect. Even if banks have low direct exposures to high-risk credit, they may have indirect exposures via financial links to more exposed third parties.

What’s more, while it has been said over and over again this year, it’s no less true today – there is a lot of uncertainty about the economic outlook. Many of our benign conclusions hinge on our generally cautiously optimistic forecasts, which may (heaven forbid) be wrong. If there was one area that we would be most concerned about getting wrong from a banking perspective, it would be the labour – and ultimately the housing – market. If we end up with big declines in house prices and jumps in mortgage defaults, the risks to banks will rise substantially.

Another problem is that policy action may be storing up problems for further ahead. A relaxation of regulatory standards means that loss provisions may be understated in the near term. And banks may reasonably misjudge the severity of the problems afflicting firms that were deemed solvent before the crisis. It will take time for many firms’ economic viability to be deciphered. The ultimate costs for the banks may turn out to be only moderate in 2020-22, but larger over a more drawn-out period. In other words, while an episode of acute credit stress may be avoided, many banks may be headed towards what Japan experienced in the 1990s and at the start of this century – with underperforming loans gnawing away at profits over a protracted period, rather than gobbling up capital in a knock-out blow. This would result in weaker lending and weaker economic recoveries throughout the 2020s.

Finally, while losses on security portfolios are not a serious threat to most banks, they are a bigger deal for insurance companies, pension funds and other asset managers. We’ll do more work on the risks to the non-bank financial sector in future research. But our general take is that losses on bonds, MBS, syndicated loans, and CLOs at such institutions mainly hit the wealth of households, rather than cause problems for the funds themselves that could have systemic consequences. One of the main ways that non-financial sector defaults could cause problems for funds is if they triggered fire sales and fund flight, draining funds’ cash buffers. But in a world where central banks are primed to do whatever it takes to support security markets, this probably isn’t much of a threat.


Simon MacAdam, Senior Global Economist, simon.macadam@capitaleconomics.com