Back Matter
  • 1 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 2 https://isni.org/isni/0000000404811396, International Monetary Fund

Annex 1. Additional Figures

Annex Figure 1.1.
Annex Figure 1.1.

Euro Area Banks: Change of CET1 Capital Ratio under Different Assumptions

(Percentage points, end-2021 relative to end-2019)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

1 Sources: European Banking Authority; European Central Bank; FitchConnect; S&P Global Market Intelligence; and IMF staff calculations.Note: CET1 = common equity Tier 1; EBA = European Banking Authority.1 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).
Annex Figure 1.2.
Annex Figure 1.2.

Dispersion of Change in Asset Risk Weights (Baseline Scenario)

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; FitchConnect; S&P Global Market Intelligence; and IMF staff estimates.Note: The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. EBA = European Banking Authority.1 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).
Annex Figure 1.3.
Annex Figure 1.3.

Euro Area Banks: CET1 Capital Ratio (Baseline Scenario), Extended Coverage

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; FitchConnect; and authors’ calculations.Note: Data labels in the figure use International Organization for Standardization (ISO) country codes. CCB = capital conservation buffer; CESEE = Central, Eastern, and Southeastern European economies; CET1 = common equity Tier 1; EA = euro area.1 The Slovak Republic is not included in the EBA Transparency Exercise.
Annex Figure 1.4.
Annex Figure 1.4.

Solvency Stress Test—Dispersion of CET1 Capital Ratio (Adverse Scenario/Extended Coverage)

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; European Systemic Risk Board; FitchConnect; S&P Global Market Intelligence; and IMF staff estimates.Note: CCB = capital conservation buffer; CESEE = Central, Eastern, and Southeastern European economies; CET1 = common equity Tier 1; MDA = maximum distributable amount (weighted average). The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. The analysis covers all three channels affecting the capital adequacy ratio under stress—profitability (net interest income and provisions), nominal assets (net lending and write-offs after reserves), and risk exposure (changes in credit risk weights). The crisis-specific risk drivers of these channels are (1) write-offs due to the projected insolvency of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks); (2) the profitability impact of policy measures (lower provisions for guaranteed loans to solvent firms, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria); and (3) the increase in risk weights to the general increase of the default risk of mortgages and firms. In addition, there is a general change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for noncorporate exposures.1 Only larger banks covered by the EBA Transparency Exercise.2 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).
Annex Figure 1.5.
Annex Figure 1.5.

Euro Area Banks: CET1 Capital Ratio (Adverse Scenario), Extended Coverage

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; FitchConnect; and authors’ calculations.Note: Data labels in the figure use International Organization for Standardization (ISO) country codes. CCB = capital conservation buffer; CESEE = Central, Eastern, and Southeastern European economies; CET1 = common equity Tier 1; EA = euro area.1 The Slovak Republic is not included in the EBA Transparency Exercise.
Annex Figure 1.6.
Annex Figure 1.6.

Euro Area: Potential Capital Need and Number of Banks below Thresholds

(EUR billion/count)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Source: Authors’ calculations.Note: The thresholds of 4.5 and 9.1 percent represent the regulatory minimum for the CET1 capital ratio (assuming the current capital relief) and the average threshold for the maximum distributable amount (MDA) for euro area banks, respectively. CET1 = common equity Tier 1; EBA = European Banking Authority.
Annex Figure 1.7.
Annex Figure 1.7.

European Banks: Changes in Capital Ratio and GDP Growth, Extended Coverage

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Source: Authors’ calculations.Note: CESEE = Central, Eastern, and Southeastern European economies; CET1 = common equity Tier 1; EA = euro area. Policies within the scope of the stress test exercise, that is, demand- and supply-side measures to support lending (guarantees, debt moratoria, insolvency stays) and financial sector policies (capital relief and conservation).
Annex Figure 1.8.
Annex Figure 1.8.

Euro Area Banks (Extended Coverage): Solvency Stress Test—Baseline Scenario

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: EBA; ECB; ESRB; FitchConnect; S&P Market Intelligence; and IMF staff estimates.Note: CCB = capital conservation buffer; CET1 = common equity Tier 1; MDA = maximum distributable amount (weighted average). The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. The analysis covers all three channels affecting the capital adequacy ratio under stress—profitability (net interest income and provisions), nominal assets (net lending and write-offs after reserves), and risk exposure (changes in credit risk weights).1 Only larger banks covered by the EBA Transparency Exercise.2 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).3 Due to corporate write-offs and net lending—corporate write-offs are due to the rise of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks).4 Net profitability impact of policy measures (lower provisions for guaranteed loans to solvent corporates, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria (households and businesses)) and change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for non-corporate exposures.5 Increase of credit risk weights due to higher unexpected losses (derived from the increase of default risk implied by the projected increase of general provisions) and additional specific provisions for additional corporate loan losses.
Annex Figure 1.9.
Annex Figure 1.9.

Euro Area Banks (Extended Coverage): Solvency Stress Test—Adverse Scenario

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; European Systemic Risk Board; FitchConnect; S&P Market Intelligence; and IMF staff estimates.Note: CCB = capital conservation buffer; CET1 = common equity Tier 1; MDA = maximum distributable amount (weighted average). The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution.1 Only larger banks covered by the EBA Transparency Exercise.2 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).3 Due to corporate write-offs and net lending—corporate write-offs are due to the rise of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks).4 Net profitability impact of policy measures (lower provisions for guaranteed loans to solvent firms, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria (households and businesses)) and change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for noncorporate exposures.5 Increase of credit risk weights due to higher unexpected losses (derived from the increase of default risk implied by the projected increase of general provisions) and additional specific provisions for additional corporate loan losses.
Annex Figure 1.10.
Annex Figure 1.10.

EU Banks (Extended Coverage): Solvency Stress Test—Baseline Scenario

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; European Systemic Risk Board; FitchConnect; S&P Market Intelligence; and IMF staff estimates.Note: CCB = capital conservation buffer; CET1 = common equity Tier 1; MDA = maximum distributable amount (weighted average). The grey shaded area of the boxplots shows the inter-quartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. The analysis covers all three channels affecting the capital adequacy ratio under stress—profitability (net interest income and provisions), nominal assets (net lending and write-offs after reserves), and risk exposure (changes in credit risk weights).1 Only larger banks covered by the EBA Transparency Exercise.2 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).3 Due to corporate write-offs and net lending—corporate write-offs are due to the rise of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks);4 Net profitability impact of policy measures (lower provisions for guaranteed loans to solvent corporates, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria (households and businesses)) and change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for non-corporate exposures;5 Increase of credit risk weights due to higher unexpected losses (derived from the increase of default risk implied by the projected increase of general provisions) and additional specific provisions for additional corporate loan losses.
Annex Figure 1.11.
Annex Figure 1.11.

EU Banks (Extended Coverage): Solvency Stress Test—Adverse Scenario

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; European Systemic Risk Board; FitchConnect; S&P Market Intelligence; and IMF staff estimates.Note: CCB = capital conservation buffer; CET1 = common equity Tier 1; MDA = maximum distributable amount (weighted average). The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. The analysis covers all three channels affecting the capital adequacy ratio under stress—profitability (net interest income and provisions), nominal assets (net lending and write-offs after reserves), and risk exposure (changes in credit risk weights).1 Only larger banks covered by the EBA Transparency Exercise.2 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).3 Due to corporate write-offs and net lending—corporate write-offs are due to the rise of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks);4 Net profitability impact of policy measures (lower provisions for guaranteed loans to solvent corporates, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria (households and businesses)) and change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for non-corporate exposures;3 Increase of credit risk weights due to higher unexpected losses (derived from the increase of default risk implied by the projected increase of general provisions) and additional specific provisions for additional corporate loan losses.
Annex Figure 1.12.
Annex Figure 1.12.

Non-EU CESEE Banks (Extended Coverage): Solvency Stress Test—Baseline Scenario

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; European Systemic Risk Board; FitchConnect; S&P Market Intelligence; and IMF staff estimates.Note: CCB = capital conservation buffer; CESEE = Central, Eastern, and Southeastern European economies; CET1 = common equity Tier 1; MDA = maximum distributable amount (weighted average). The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. The analysis covers all three channels affecting the capital adequacy ratio under stress—profitability (net interest income and provisions), nominal assets (net lending and write-offs after reserves), and risk exposure (changes in credit risk weights).1 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).2 Due to corporate write-offs and net lending—corporate write-offs are due to the rise of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks).3 Net profitability impact of policy measures (lower provisions for guaranteed loans to solvent corporates, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria (households and businesses)) and change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for non-corporate exposures.4 Increase of credit risk weights due to higher unexpected losses (derived from the increase of default risk implied by the projected increase of general provisions) and additional specific provisions for additional corporate loan losses.
Annex Figure 1.13.
Annex Figure 1.13.

Non-EU CESEE Banks (Extended Coverage): Solvency Stress Test—Adverse Scenario

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; European Systemic Risk Board; FitchConnect; S&P Market Intelligence; and IMF staff estimates.Note: CCB = capital conservation buffer; CESEE = Central, Eastern, and Southeastern European economies; CET1 = common equity Tier 1; MDA = maximum distributable amount (weighted average). The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. The analysis covers all three channels affecting the capital adequacy ratio under stress—profitability (net interest income and provisions), nominal assets (net lending and write-offs after reserves), and risk exposure (changes in credit risk weights).1 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).2 Due to corporate write-offs and net lending—corporate write-offs are due to the rise of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks).3 Net profitability impact of policy measures (lower provisions for guaranteed loans to solvent firms, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria (households and businesses)) and change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for noncorporate exposures.4 Increase of credit risk weights due to higher unexpected losses (derived from the increase of default risk implied by the projected increase of general provisions) and additional specific provisions for additional corporate loan losses.
Annex Figure 1.14.
Annex Figure 1.14.

All Banks (Extended Coverage): Solvency Stress Test—Baseline Scenario

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; European Systemic Risk Board; S&P Market Intelligence; and IMF staff estimates.Note: CCB = capital conservation buffer; CET1 = common equity Tier 1; MDA = maximum distributable amount (weighted average). The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. The analysis covers all three channels affecting the capital adequacy ratio under stress—profitability (net interest income and provisions), nominal assets (net lending and write-offs after reserves), and risk exposure (changes in credit risk weights).1 Only larger banks covered by the EBA Transparency Exercise.2 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).3 Due to corporate write-offs and net lending—corporate write-offs are due to the rise of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks);4 Net profitability impact of policy measures (lower provisions for guaranteed loans to solvent corporates, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria (households and businesses)) and change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for non-corporate exposures;5 Increase of credit risk weights due to higher unexpected losses (derived from the increase of default risk implied by the projected increase of general provisions) and additional specific provisions for additional corporate loan losses.
Annex Figure 1.15.
Annex Figure 1.15.

All Banks (Extended Coverage): Solvency Stress Test—Adverse Scenario

(Percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; European Systemic Risk Board; FitchConnect; S&P Market Intelligence; and IMF staff estimates.Note: CCB = capital conservation buffer; CET1 = common equity Tier 1; MDA = maximum distributable amount (weighted average). The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. The analysis covers all three channels affecting the capital adequacy ratio under stress—profitability (net interest income and provisions), nominal assets (net lending and write-offs after reserves), and risk exposure (changes in credit risk weights).1 Only larger banks covered by the EBA Transparency Exercise.2 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).3 Due to corporate write-offs and net lending—corporate write-offs are due to the rise of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks).4 Net profitability impact of policy measures (lower provisions for guaranteed loans to solvent firms, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria (households and businesses)) and change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for noncorporate exposures.5 Increase of credit risk weights due to higher unexpected losses (derived from the increase of default risk implied by the projected increase of general provisions) and additional specific provisions for additional corporate loan losses.

Annex 2. Methodology

Our paper examines how the implications of the current crisis for bank-specific and general macroeconomic conditions impact banks’ capitalization over time. And we do so by considering a wide range of COVID-19-related policy measures that support credit supply (that is, financial sector policies, such as greater supervisory and regulatory flexibility, providing banks with more headroom to lend through capital easing) and credit demand (for example, debt moratoria, credit guarantees) (Figure 16). These “bank-facing” measures work in tandem with other (fiscal) measures, such as grants, wage subsidies, commercial rate reductions, and tax deferrals, which indirectly affect banks by mitigating potential liquidity and/or solvency risks of borrowers, especially highly affected firms and households in areas where lockdowns have a higher impact on employment.

We project the crisis impact on bank’s solvency by considering three channels: profitability, the amount and types of its assets, and their associated riskiness. We take the Common Equity Tier 1 (CET1) capital ratio at the end of 2019 as starting point and estimate how it changes over time due to retained earnings from projected profits (net of taxes and dividend payouts), after adjusting for the net change in assets and their associated riskiness under baseline and adverse conditions (Figure 14). More specifically, we assess (1) the impact of projected GDP growth and unemployment under the current WEO projections on the return on assets (ROA) and its components (that is, net interest income, fee/commission income, operating expenses, and loan loss provisioning) (“profitability channel”); (2) the impact of write-offs (or write-offs) for actual losses (in excess of available provisions), lowering the book value of the stock of assets (net of estimated new lending) (“asset channel”); and (3) the impact of higher credit risk-weights to account for rising default risk (“risk channel”).

For each component, we account for the specific impact of crisis-related policy measures and shocks:

  • For the profitability channel, we consider lower provisions for guaranteed loans to solvent firms, some loss forbearance on eligible loans under debt moratoria, and the decline in accrued interest income due to duration of debt moratoria.

  • For the asset channel, we account for the potential surge of corporate defaults (as single factor shock) in the form of additional write-offs due to the projected insolvency of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks), in addition to impairment charges for noncorporate exposures.

  • For the risk channel, we recognize the loss mitigating impact of public sector guarantees, which should reduce the marginal credit risk weight of new corporate loans to solvent borrowers (up to the availability of such guarantees).

In the following section, we discuss the technical specification of each component, including how the impact of policy measures and the single-factor corporate shock are incorporated.

Profitability Channel1

Specification and Estimation

We measure bank profitability based on the reported ROA. ROA contributes to organic capital growth during each reporting period after adjusting for changes in assets (due to credit losses and valuation changes) and net of dividends and taxes over a two-year time horizon, where t ∈ {2020, 2021}.

In the simplest form, the impact of bank profits on capital can be shown using the capital-to-asset ratio as a simplified representation of bank leverage,

Ki,tAi,t=Ki,t1Ai,t1(1+ROAi,t|ΔBj,tΔCi,k,t)(1τ)(1dt),(A2.1)

in which ROAit is the return on total assets A of bank i at time t, K is total capital, ΔΒ and ΔC are 1 – (1xj) and (1xk) vectors of one-period changes in bank-specific variables and macro-financial factors (which apply to all banks within a specific country), respectively, τ is the time-invariant tax rate, and dt is the dividend payout ratio.2 Thus, we re-write equation (A2.1) above in risk-weighted terms to derive the capital-generating impact of profitability. We use CET1 as the most junior (and most risk-sensitive) form of capital to define the contribution of retained earnings to the change of the capital adequacy ratio (CAR) as

CET1i,tAi,tRWi,t=CET1i,t1Ai,t1RWi,t1+(ROAi,t|ΔBj,tΔCi,k,t1RWi,t)(1τ)(1dt),(A2.2)

in which RWi,t is the average risk weight of total assets A.

Policy Impact

Since policy measures affect various elements of the profit and loss statement, we also determine the components of ROA (in percent of total assets, Ai,t) so that

ROAi,t=NIIi,t+NoninterestIncomei,tLLPi,tOtheri,t*(A2.3)

and Other*i,t comprises operating expenses and write-offs of NPLs, where NIIi,t and LLPi,t denote the net interest income and loan loss provisioning expenses, respectively.

If policy measures operate as intended, these sub-components are adjusted as follows:

  • Debt moratoria. Different countries have allowed for repayment relief on household loans and/or corporate loans together with greater regulatory flexibility in the classification and provisioning rules. The negative impact of moratoria on net interest income, NIIi,t, is modeled as

    NII˜i,t=NII˜i,t(1m12×θtM×{Si,tHSi,tC),(A2.4)

    in which m12 is the share of accrued but non-paid interest income (in which m denotes the duration in months), multiplied by the projected average usage rate of moratoria across all eligible loans in a given country,3 θtM, and the bank-specific share of household loans, sH, and/or corporate loans, sC, in total loans. We also assume that suspending the automatic classification of impaired loans under IFRS-9 in favor of a case-by-case assessment entails some sluggishness in how provisions adjust to deteriorating credit quality. Thus, we update the estimate provisioning expenses, LLP^i,t, to

    LLP˜i,t=LLP^i,t(LLPtIFRS9ωt1)(ΔL^i,tC+ΔL^i,tH)(n12×θt×{Si,tHSi,tC),

    in which ΔL^i,tC+ΔL^i,tH denote the change in the total amount of corporate and household loans (consumer and mortgage lending), ωt-1 is the historical, bank-specific coverage ratio (at end-2019), and ΔLLPtIFRS9 is the implied increase of provisions based on the expected migration of impaired loans from “Stage 1” to “Stage 2” as well as “Stage 2” to “Stage 3” according to the automatic loan classification under IFRS-9 (EBA 2020e; Annex 4). Given that this approach assumes a certain degree of under-provisioning of moratoria loans (as banks can no longer use non-payment as a signal of deteriorating borrower quality), the capital add-back of provisioning expenses under the transitional arrangement for the implementation of IFRS-9 is not considered.

  • Public sector guarantees. In addition, public sector guarantees for corporate loans reduce the amount of provisions proportionate to the share of expected losses covered by the government. The available amount of guarantees is allocated proportionate to each bank’s share of corporate lending within a given country, subject to the larger of (1) the estimated loan growth, ΔL^i,tC (see equation (A2.23) below), by the share of solvent firms, μi,tC, and (2) the projected usage rate of guarantees, θtG, relative to the total stock of corporate loans, Li,tC. For most countries, the size of the envelope for guarantee programs exceeds the amount of likely corporate credit growth over the time period covered by the stress test. Governments cover losses up to φ percent of guaranteed loans uniformly across all banks within a country but loss coverage differs across countries. Thus, we can further refine equation (A2.5) above to

    LLP˜i,t=LLP^i,t(LLPtIFRS9ωt1)(ΔLi,tC+ΔLi,tH)(m12×θt×{Si,tHSi,tC)(φωt1)max(μi,tCΔLi,tC,θtGLi,tC),(A2.6)

    and re-state equation (A2.3) as

    ROA˜i,t=NII˜i.t+NoninterestIncomei,tLLP˜i,tOtheri,t*(A2.7)
  • Dividend payouts. In addition, regulators have encouraged banks to suspend distribution of dividends and share buybacks to conserve capital but have recently began to lift these restrictions. For instance, on December 2020, the ECB, after an assessment of the macroeconomic outlook, financial stability, and reliability of banks’ capital planning, has issued guidance that would allow banks to resume dividend payouts within strict limits (until end-September 2021). We incorporate the conditionality of remaining restrictions into the specification of equation (A2.2) above and set the dividend payout ratio d2020 = 0, and, for 2021, d2021=min(0.2×Σt2020ROAi,t1,0.015×CET1i,t).

Estimation

We estimate ROA and its sub-components using annual time series data from the statutory annual flings of 3,421 banks in 41 European countries (2008–19) on a consolidated reporting basis. The bank-specific data from FitchConnect are combined with macroeconomic data for each country from the World Economic Outlook (WEO) database. ROA is most sensitive to real GDP growth, yt, the unemployment rate, URt, and the lagged NPL ratio, NPLi,t-1 (Jobst and Weber 2016; Elekdag, Malik, and Mitra 2020). The NPL ratio captures the legacy impact of accumulated impaired loans on banks’ profit generating capacity during different points in the economic cycle. Thus, we can estimate the panel regression as

ROAi,t=α0ROA+β1ROAyt+β2ROAURt+β3ROANPLi,t1+Σj=15γjROABj,i,t1+BankFE+YearFE+εi,tROA(A2.8)

where B is a (1xj)-vector of bank-specific variables: total assets, leverage, the share of loans and deposits relative to total assets, and operational efficiency. Bank and year fixed effects are included, and the standard errors are clustered by country*year.4

Annex Table 2.1 shows the estimation results for five different samples of banks corresponding to the following (also summarized in Table 2 in the main text): (1) “SSM,” that is, the largest euro area banks that fall within the perimeter of the Single Supervisory Mechanism (SSM) and are directly supervised by the ECB, (2) euro area banks without SSM banks, (3) banks in non-euro area EU countries, (4) banks in central and eastern European (CESEE) countries that are not EU member states, and (5) all banks included in the EBA Transparency Exercise, which is used for projecting bank profitability in non-EU advanced economies (Annex Figure 2.1). For example, for the largest euro area banks, the coefficients for Model 1 are applied, that is, β1 = 0.27, β2 = -0.21, and β3 = -0.02.

Annex Table 2.1.

European Banks: ROA Satellite Models—Estimation Results

article image
Source: IMF staff estimates.Note: CESEE = Central, Eastern, and Southeastern European economies; EA = euro area; EBA = European Banking Authority; SSM = Single Supervisory Mechanism. Bank and year fixed effects included; standard errors clustered by country*year (in parentheses). ***p < 0.01, **p < 0.05, *p < 0.1.
Annex Figure 2.1.
Annex Figure 2.1.

Sample Coverage

(Asofend-2019)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Source: Authors.Note: EBA = European Banking Authority; ECB = European Central Bank.1 Bank data from EBA, ECB, and FitchConnect.2 Bank data from S&P Market Intelligence.

The NPL ratio and unemployment rate are highly correlated, with the estimated effect being more persistent for the former. Thus, in some samples, the impact of the NPL ratio is statistically small once the unemployment rate is included in the regression. Monetary conditions, such as the short-term interest rates and the slope of the country-specific yield curve, seem to have little explanatory power for predicting ROA beyond their impact on growth and unemployment.

For each bank, the estimated model coefficients above are used to estimate profitability, ROA^i,t, of each bank at the end of period t ∈ {2020; 2021}. Assuming that bank-specific factors remain unchanged over the two-year stress test horizon, the estimated ROA at the end of 2020 would be calculated as

ROA^i,2020|Bi,2019)=ROAi̇,2019+β1ROAΔy2020+β2ROAΔUR2020+β3ROAΔNPLi,2019(A2.9)

For the subsequent period, the NPL ratio is no longer observable and can be derived separately via a satellite model (whose coefficients have been estimated similar to the specification in the equation above; Annex Table 2.2) so that5

Annex Table 2.2.

European Banks: NPL Ratio Model—Estimation Results

article image
Source: IMF staff estimates.Note: CESEE = Central, Eastern, and Southeastern European economies; EA = euro area; EBA = European Banking Authority; NPL = nonperforming loans. Bank and year fixed effects included; standard errors clustered by country*year (in parentheses). ***p < 0.01, **p < 0.05, *p < 0.1.
NPL^i,2020|Bi,2019=NPLi,2019+β1NPLΔy2020+β2NPLΔUR2020+β3NPLΔNPLi,2019,(A2.10)

and

ROA^i,2021|Bi,2020=ROA^i,2020+β1ROAΔy2021+β2ROAΔUR2021+β3ROANPL^i̇,2020|Bi,2019.A2.11

The main additive components of ROA—net interest income, non-interest income, and loan loss provisions (as specified in equation (A2.3) above), all expressed as a share of total assets—are estimated as well. For instance, the expected level of provisioning expenses can be specified as

LLPi,t=α0LLP+β1LLPyt+β2LLPURt+β3LLPNPLi,t1+Σj=15γjLLPBj,i,t1+BankFE+YearFE+εi,tLLP,(A2.12)

where B is a (1xj) -vector of the same bank-specific variables as in equation (A2.8) above. Annex Table 2.3 shows the estimated parameter coefficients for all banks covered by the EBA Transparency Exercise, which has been applied to all sample banks. Thus, the new flow of provisions at the end of the first period of the stress test is estimated as

Annex Table 2.3.

Satellite Models for Components of Return on Assets—Estimation Results1

article image
Source: IMF staff estimates.Note: ECB = European Central Bank; SSM = Single Supervisory Mechanism. Bank and year fixed effects included; standard errors clustered by country*year (in parentheses). ***p < 0.01, **p < 0.05, *p < 0.1.

Only euro area banks directly supervised by the ECB (SSM).

LLP^i,2020=β1LLPΔy2020+β2LLPΔUR2020+γ1LLPloanstoasseti,2019,(A2.13)

and analogously for interest income, NIIi,2020, and non-interest income, NoninterestIncomei,2020. The estimation of LLP^i,2021, for the next period (end-2021) follows the same specification, with relevant bank-specific explanatory variables, such as loans_to_asseti,2020 being determined endogenously based on each bank’s interim (end-2020) balance sheet.

Thus, the estimated ROA with effective policy measures in 2020 and 2021 would be defined as

ROA˜i,2020=ROA^i,2020|Bi,2019+(NII˜i,2020NII^i,2020)(LLP˜i,2020LLP^i,2020)(A2.14)

and

ROA˜i,2021=ROA^i,2021|Bi,2020+(NII˜i,2021NII^i,2021)(LLP˜i,2021LLP^i,2021)(A2.15)

respectively.

Impact of Corporate Shock

The estimated profitability also needs to be adjusted by the impact of the corporate shock on loan loss provisions. Given the outsized impact of the pandemic on the “highly-affected” sectors, the analysis distinguishes between banks’ corporate exposures to “highly-affected” and other sectors using granular data from the 2020 EBA Transparency Exercise (Box 1). For banks in non-EU/EEA countries, which fall outside the scope of the EBA Transparency Exercise, banks’ sectoral exposure is assumed to be the same as the average sectoral exposure in one or the average of a few neighboring countries whose banks are included in the EBA Transparency Exercise.6 For each sector and each country in the sample, we determine the average share of firms (weighed by outstanding debt) that experience a deterioration of their “financial status” in terms of liquidity and solvency, resulting in four categories in the corporate matrix described in Box Figure 1.1 in the main text— solvent-liquid (green), insolvent-liquid (pink), insolvent-illiquid (red), and solvent-illiquid (yellow).

Provisions increase relative to corporate loans for borrowers that have become either insolvent-liquid or solvent-illiquid. In addition, the incidence of insolvent-illiquid borrowers determines the scale and timing of write-offs of corporate loans, which releases provisions up to the average coverage ratio. Thus, we can amend the specification of loan loss provisions without considering policy measures as

LLP^i,t=(1+pliquid,insolventC+pilliquid,solventCPilliquid,insolventCωt1υt)Li,tCLi,t(β1LLPΔyt+β2LLPΔURt+γ1LLPloanstoasseti,t1),(A2.16)

and with policies as

LLP^i,t=(1+pliquid,insolventC[withpolicy]+pilliquid,solventC[withpolicy]Pilliquid,insolventC[withpolicy]ωt1υt)Li,tCLi,t(β1LLPΔyt+β2LLPΔURt+γ1LLPLoanstoasseti,t1),(LLPtIFRS9ωt1)(ΔLi,tC+ΔLi,tH)(m12×θt×{SitHSitC)(φωt1)max(μi,tCΔLi,tCθtGLi,tC),(A2.17)

in which vt=min(9m;1) indicates the share of write-offs that occur in this period (and the remainder, 1 – νt, in the subsequent period) for the duration of deferred bankruptcies of m months [since end-March 2020]) within the two-year stress test time horizon (see the specification of the corporate shock impact on the asset channel below). Note that corporate shock has been defined in aggregate to avoid complex notation. Consistent with the description in Box 1, the three relevant categories in the corporate matrix are in fact implemented on a sector-by-sector basis so that

Annex Table 2.4.

Satellite Models for Loan Growth—Estimation Results1

article image
Source: IMF staff estimates.Note: CET1 = common equity Tier 1; ECB = European Central Bank; FE = fixed effects; NPL = nonperforming loan; SSM = Single Supervisory Mechanism. Bank and year fixed effects included; standard errors clustered by country*year (in parentheses). Standard errors clustered by country*year (in parentheses). ***p < 0.01, **p < 0.05, *p < 0.1.

Only euro area banks directly supervised by the ECB (SSM) with values reported on a consolidated basis.

Model 2 is estimated using the Arellano Bond GMM dynamic panel model, for which an R-squared is not available (since it would not be the correct measure of model ft).

ρ/iquid,insolventCLi,tC=Πs=0Sρliquid,insolventCsLi,tC.s(A2.18)
ρ/iquid,solventCLi,tC=Πs=0Sρliquid,solventCsLi,tC,s(A2.19)
ρ/iquid,insolventCLi,tC=Πs=0Sρliquid,insolventCsLi,tC,s(A2.20)

where sS denotes the total number of sectors in the economy of each sample country.

Given that the write-offs of corporate loans will also reduce total assets (see equation (A2.25) below),7 accounting for the asset channel impact of the corporate shock, estimated ROA with effective policy measures in 2020 and 2021 would be defined as

ROA˜i,2020=ROA^i,2020|Bi,2019×Ai,2019Ai,2020+(NII˜i,2020NII˜i,2020)(LLP˜i,2020LLP˜i,2020)(A2.21)

and

ROA˜i,2021=ROA^i,2021|Bi,2020×A˜i,2019A˜i,2020+(NII˜i,2021NII˜i,2021)(LLP˜i,2021LLP˜i,2021)(A2.22)

respectively.

Asset Channel

Specification and Estimation

We assume that total assets of banks change in response to net lending and write-offs of impaired exposures after considering available provisions and the recovery value of collateral. We estimate lending growth based on the credit supply equation for the total stock of loans Li,t as

ΔLi,tLi,t1=α0L+β1LCETli,t1+β2LNPL1i,t1+BankFE+εi,tL(A2.23)

which accounts for the initial capital position (lagged CET1 capital ratio), the lagged NPL ratio, and previous year’s loan growth at the bank level while demand conditions are absorbed by the country-year fixed effect. The predicted loan growth is only applied to banks whose CET1 capital ratio clears the MDA threshold of 9.1 percent as the average across all euro area banks (and 10.6 percent for non-euro area European banks). Thus, we can project each bank’s change of total assets as

A^i,t=(1δ)Li,t1+ΔLi,t^ifCET1i,t0.091,(A2.24)

in which the estimated loan growth is uniform across all major asset classes so that ΔLi,t^Li,t1=ΔLi,tC^Li,tC=ΔLi,tH^Li,tH (with ΔLi,tCandΔLi,tH denoting new corporate and mortgage lending [see equation (A2.5) above]), and δ is the average amortization rate of total outstanding loans, which is derived as reciprocal of the weighted average of the uniform maturities of corporate and mortgage loans MC and MH in all countries.

Impact of Corporate Shock

In addition, we explicitly model the write-offs of corporate exposures due to the significant impact of the crisis on default risk across sectors and the mitigating impact of supportive policy measures (Box 1). Thus, equation (A2.24) above can be augmented so that

A˜=(1δ)Li,t1Li,t1Cρilliquid,insolventCLGDC(1ωi)+ΔLi,t^,(A2.25)

in which ρilliquid,insolventC denotes the write-of rate, which is defined as the expected increase of the debt-weighed share of illiquid and insolvent corporate borrowers in each sector relative to the pre-crisis situation (Box 1), ω reflects the average provisioning coverage ratio, and LGDC is the average country-specific loss given default for corporate loans (after accounting for the recovery value of available collateral), which was obtained as quarterly risk parameters from the EBA Risk Dashboard (EBA 2021a).

The write-offs are also added to the change in Δ NPLi,2020 for estimating the bank profitability in equation (A2.9) above, so that the satellite model-derived NPL ratio at the end of the first period in equation (A2.10) now reads as

NPL¯|Bi,2019=NPLi,2019+β1NPLΔy2020+β2NPLΔUR2020+β3NPLΔNPLi,2019++Li,t1CPilliquid,insolventCLGDC(1ωi)+ΔLi,t^,(A2.26)

Consequently, equation (A2.11) for estimating the ROA at the end of 2021 becomes

ROA˜i,2021=ROA^i,2020+β1ROAΔy2021+β2ROAΔUR2021+β3ROANPL^i,2020|Bi,2019.(A2.27)

Impact of Policy Measures

With policy measures in place, public sector guarantees for corporate loans influence the specification of loan growth as well as the scale and timing of corporate default (and associated write-offs). More specifically, we revise equation (A2.25) above to

A˜i,t=(1δ)Li,t1Li,t1Cρilliquid,insolventC[withpolicy]LGDC(1ωi)(1+φθtG)+ΔLi,t^+ΔLi,tC^+max(μi,tCΔLi,tC^,θtGLi,tC),(A2.28)

in which ρilliquid,insolventC[withpolicy] recognizes the mitigating impact of borrower support on the default risk of illiquid and insolvent corporate borrowers on the availability of borrower support, and φ is the uniform loss coverage provided by the public sector proportionate to the relative share of corporate lending of each bank within a particular country. The availability of guarantees also influences new corporate lending to the extent that corporate borrowers are eligible to receive them (that is, they are solvent); thus, credit demand from firms is the larger of (1) the share of solvent firms, μi,tC, relative to estimated credit demand and (2) the projected usage rate of guarantees, θtG relative to the total stock of corporate loans, Li,tC (which is also taken into account the estimated provisions in equation (A2.6) above).

Since deferred bankruptcy proceedings (Figure 31) delay the realization of losses from defaults (on the assumption of some under-provisioning), we split the write-offs between 2020 and 2021 according to the duration of insolvency stays. Thus, the estimated size of each bank’s balance sheet at t = 2020 becomes

A˜i,t=(1δ)Li,t1Li,t1Cρilliquid,insolventC[withpolicy]LGDC(1ωi)(1+φθtG)vt+ΔLi,t^+ΔLi,tC^+max(μi,tCΔLi,tC^,θtGLi,tC),(A2.29)

where Vt=min(9m;1) indicates the share of write-offs that occur in this period (and the remainder, 1 – νt , in the subsequent period) for the duration of deferred bankruptcies of m months (since end-March 2020) within the two-year stress test time horizon.

The delayed write-offs in t = 2021 are added to the change in ΔNPLi,2020 for estimating the bank profitability with effective policy measures in equation (A2.11) above, so that after considering the policy impact on interest income and loan loss provisioning, equation (A2.15) becomes

ROA˜i,2021=ROA^i,2020+β1ROAΔy2021+β2ROAΔUR2021+β3ROANPL˜i,2020|Bi,2019+(NII˜i,2021NII˜i,2021)(LLP˜i,2021LLP˜i,2021),(A2.30)

where the NPL ratio at the end of the first period in equation (A2.26) now reads as

NPL˜i,2020|Bi,2019=NPLi,2019+β1NPLΔy2020+β2NPLΔUR2020+β3NPLΔNPLi,2019+Li,t1Cρilliquid,insolventC[withpolicy]LGDC(1ωi)(1+φθtG)vt+ΔLi,t^,(A2.31)

Risk Channel

Specification and Estimation

The riskiness of banks’ assets is defined by risk weights (RWs), which determine the required capital for an exposure relative to the prudential minimum consistent with the current Basel regulatory framework. For a bank operating exactly at the minimum CAR, a risk weight of 100 percent implies that 8 percent of the nominal amount of existing exposures are covered by total capital on average. Risk weights are calibrated to the amount of unexpected losses of an exposure (that is, losses that exceed the expected level with a certain degree of statistical confidence).

The modeling of the “risk channel” is focused on the sensitivity of credit risk weights, which define the amount of unexpected losses from a credit-sensitive assets. Higher risk weights increase the denominator of the capital ratio (while impairment charges and higher provisioning expenses decrease the numerator).

In our analysis, we evaluate how of a bank’s credit risk weights would change during times of stress by shocking the default risk of its corporate and mortgage loans. We specify this shock consistent with the expected loan loss coverage using a satellite model for provisioning (including the impact of the corporate sector shock, see below). The shocked default risk is then used to update the credit risk weights for corporate and mortgage loans, which are then combined to derive the stressed overall credit risk weight (based on the share of the risk-weighted amounts of corporate and mortgage loans held by each bank).

The capital requirement (K) limits a bank’s leverage based on the riskiness of their exposure to ensure that available capital is enough to cover unexpected losses above the level of provisions. We assume that all sample banks apply the advanced internal ratings-based (IRB) approach for credit risk (BCBS 2017) to determine expected losses of a bank’s credit-sensitive exposures based on the one-year PD, the LGD, and the exposure-at-default (EAD). The capital requirement (K) —based on the underlying specification of an asymptotic single risk factor model—is defined as:

K=(LGD×Φ(+11RΦ1(PD)R1R×Φ1(a))stresseddefaultratePD×LGD)×1+(M2.5)b11.5bfullmaturityadjustment(A2.32)

in which PD is the default rate over one year, with statistical confidence (single-sided) a = 0.999 (that is, 99.9 percent) assuming standard normal cumulative distribution function Φ(⋅), effective maturity M = ∑tt × CFt/tCFt,8 in which CF denotes the cash flows (principal, interest payments, and fees) contractually payable by the borrower in period t, maturity adjustment b = (0.11852 – 0.05478 × ln(PD))2, and correlation factor

R=0.12×1exp(150×PD)1exp(50)+0.24×(11exp(50×PD)1exp(50)).(A2.33)

This specification is derived from a single risk factor model, which assumes that the obligor’s asset value follows a lognormal distribution, so we can write

dA=μAdt+σAdx(A2.34)

in which x is a random stochastic process (“white noise”), so that at any time t in the future

lnA(t)=lnA(0)+μtσ22t+σtX(t),(A2.35)

in which X~Φ (0,1) is a standard normal random variable. It is assumed that

X(t)=RY(t)+1RZ(t),(A2.36)

in which Y is a single, global risk factor, Z is an obligor-specific, idiosyncratic risk factor, and R is the correlation of the obligor to the global risk factor, where Y,Z~Φ(0,1) are independent of each other. Thus, the stressed default rate at statistical confidence of 99.9 percent, a = 0.999 , in the IRB formula of the Basel III framework in equation (A2.32), with the asset process in equation (A2.34), is derived from

Pr(X(t)<Φ1(PD)|Y(t)=Φ1(a))=Pr(+RY(t)1RZ(t)<Φ1(PD)|Y(t)=Φ1(a))=Pr(Z(t)<Φ1(PD)RΦ1(a)1R)=Φ(Φ1(PD)RΦ1(a)1R)(A2.37)

Since the degree of capital adequacy depends on the riskiness of exposures relative to available capital, K , the risk-weighted amount of capital implies a leverage ratio of 12.5 at the minimum capital requirement, CAR = 8.0 percent and EAD at unity. Given that RW=1CAR×K, we can derive the implicit risk-weight from equation (A2.32) above as

RW=(LGD×Φ(11RΦ1(PD)+R1R×Φ1(a))PD×LGD)×1+(M2.5)b11.5b×12.5,(A2.38)

if each component of the banks’ exposures contributes uniformly to the minimum CAR of 8.0 percent.

Thus, for each bank, we can derive the one-year default risk, PDi,t1C and PDi,t1H, implied by observed risk weights for corporate and mortgage loans, RWi,t1CandRWi,t1H

RWi,t1{C;H}=(LGDt{C;H}×Φ(Φ1(PDi,t1{C;H})+RΦ1(a)1R)PDi,t1{C;H}LGDt{C;H})0.08M{C;H}(A2.39)

assuming that corporate and mortgage loans have average maturities of MC = 6.3 [years] and MH = 22.0 [years], respectively, and loss-given default (LGD) rates, LGDiCandLGDiH, which vary across countries.9 Corporate and mortgage loans are also assigned a bank-specific and time-varying maturity adjustment, bi,t1Candbi,t1H, as well as correlation factors, Ri,t1CandRi,t1H, which are calculated using equations (A2.32) and (A2.33) above. The LGDs were derived as the lower of the net recovery rate (as reported in EBA’s benchmarking exercise [EBA 2020c]) and the LGDs for corporate and retail exposures published in the risk parameter statistics of EBA’s Risk Dashboard (EBA 2020a).10 Both PDi,t1CandPDi,t1H were cross-validated with general PDs reported by EBA for corporate and retail exposures and increased by up to 50 percent of their estimated value in cases where EBA-reported values were significantly higher.

We can then determine the change of PDi,t1{C;H} in each period under stress consistent with the change in loan loss provisions after accounting for the change in non-performing loans (using the respective satellite models that capture change in macroeconomic and bank-specific conditions; see equations (A2.10), (A2.13), and (A2.16) above) so that

PD^i,t{C;H}=PDi,t1{C;H}×(1+max(LLP^i,t{C;H}ΔNPLi,t{C;H},0)×LLPi,t1{C;H}NPLi,t1{C;H}),(A2.40)

in which the pre-stress provisions for corporate and mortgage loans, LLPi,t1C and LLPi,t1H, are derived from LLP = (1.662 + 0.00092(RW × 100)2 – 0.06 (RW × 100)) × LGD (Jobst and Weber 2016).

We can now plug the updated PDs above back in the IRB formula to derive the “shocked” RWs for both corporate loans and mortgages, respectively.

{RW^i,t{C;H}RW˜i,t{C;H}}=(LGDt{C;H}×Φ(Φ1({PD^i,t{C;H}PD˜i,t1{C;H}})+RΦ1(a)1R){PD^i,t{C;H}PD˜i,t1{C;H}}LGDt{C;H})0.08M{C;H}(A2.41)

Impact of Corporate Shock

We account for the write-offs of each bank’s corporate exposures in equation (A2.25) of the asset channel by revising equation (A2.40) above to

PD^i,t{C;H}=PDi,t1{C;H}×(1+max(LLP^i,t{C;H}ΔNPL^i,t{C;H}+Li,t1Cρilliquid,insolventCLGDC,0)×LLPi,t1{C;H}NPLi,t1{C;H}),A2.42

Impact of Policy Measures

Accounting for the impact of policy measures on the change in credit risk weights requires replacing the estimated loan loss provisions in equation (A2.42) with the specification in equation (A2.17) of the profitability channel after accounting for the corporate sector shock from equation (A2.29) of the asset channel so that

PD˜i,t{C;H}=PDi,t1{C;H}×(1+max(LLP˜i,t{C;H}ΔNPL˜i,t{C;H}+Li,t1Cρilliquid,insolventCLGDC(1ωi)(1φθtG)vt,0)×LLPi,t1{C;H}NPLi,t1{C;H}),A2.43

in t = 2020, and

PD˜i,t{C;H}=PDi,t1{C;H}×(1+max(LLP˜i,t{C;H}ΔNPL˜i,t{C;H}+Li,tCρilliquid,insolventCLGDC(1ωi)(1φθtG)vt,0)×LLPi,t1{C;H}NPLi,t{C;H}),A2.44

in t + 1 = 2021.

Note that we ignore the net impact of the write-offs of corporate loans on the credit risk weight of the loan portfolio. Empirical evidence presented in Jobst, Ong and Schmieder (2013) suggests that the risk weights of defaulted loans tends to be 2.5 times higher than that of the average loan, which would mean that PD˜i,t{C;H} in equation (A2.43) above would be slightly lower all else equal.

Annex 3. Impact of COVID-19 on the Capital-to-Asset Ratio

In addition to the analysis of the CET1 ratio as a measure of bank solvency, we also consider the capital-to-asset ratio. This simpler measure of bank capital does not capture risk-weighted assets and is a proxy for the “leverage” ratio, and EU banks will need to maintain a minimum leverage ratio of 3 percent from June 2021. The leverage ratio is only a backstop to the regulatory risk-weighted capital ratios, such as CET1 ratio, and has been put in place to ensure that risk-weight modeling uncertainties and errors do not lead to excessively swollen balance sheets of banks. The leverage ratio leads to a simpler connection between banks’ profitability and capital accumulation.

Similar to the results for the CET1 ratio, the capital-to-assets is projected to fall by about 2 percentage points through 2021 under the baseline scenario with policies. For this exercise, the methodology in Annex 2 is used for the profitability and the asset channels only. Starting from a relatively high level of 6.9 percent in 2019, European banks, on average, lose retained earnings with lower ROA through 2020, and are not able to replenish capital buffers through retained earnings even with the recovery in GDP growth envisaged for 2021 even with policies. This is because higher NPLs in 2020 weigh on ROA in 2021, and corporate bankruptcies require write-offs reducing capital against existing and projected provisions. While policies help, the capital-to-asset ratio does not benefit from reduced risk weights against a no-policy counterfactual. The sensitivity analysis of the capital-to-asset ratio suggests that continued dividend retention would provide mitigate the projected shock by 0.2 percentage points over the two-year time horizon. Excluding the single factor shock from a surge of corporate bankruptcies would not significantly change the results given the delayed insolvency procedures, resulting most of the rising provisioning expenses being absorbed by higher profitability during the recovery in 2021.

Annex Figure 3.1.
Annex Figure 3.1.

Euro Area Banks: Change of Capital-to-Asset Ratio under Different Assumptions (Extended Coverage)

(Percentage points, end-2021 relative to end-2019)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; FitchConnect; S&P Global Market Intelligence; and IMF staff calculations.
Annex Figure 3.2.
Annex Figure 3.2.

Euro Area Banks (Extended Coverage): Solvency Stress Test

(Capital-to-asset ratio, percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; European Systemic Risk Board; FitchConnect; and IMF staff estimates.Note: The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. The analysis covers the two main channels affecting both capital and assets under stress—changes in profitability (net interest income and provisions) and effective changes in total assets (net lending and write-offs after reserves). The crisis-specific risk drivers of these channels are (1) write-offs due to the projected insolvency of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks), (2) the profitability impact of policy measures (lower provisions for guaranteed loans to solvent firms, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria), In addition, there is a general change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for noncorporate exposures. The calculation does not consider changes in unexpected losses, which are reflected in the risk-weighting of asset exposures in the computation of capital adequacy.1 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, delayed insolvency proceedings, and dividend restrictions (only in 2020).
Annex Figure 3.3.
Annex Figure 3.3.

EU Banks (Extended Coverage): Solvency Stress Test

(Capital-to-asset ratio, percent)

Citation: Departmental Papers 2021, 008; 10.5089/9781513572772.087.A999

Sources: European Banking Authority; European Central Bank; European Systemic Risk Board; FitchConnect; and IMF staff estimates.Note: The grey shaded area of the boxplots shows the interquartile range (25th to 75th percentile), with whiskers at the 5th and 95th percentile of the distribution. The analysis covers the two main channels affecting both capital and assets under stress—changes in profitability (net interest income and provisions) and effective changes in total assets (net lending and write-offs after reserves). The crisis-specific risk drivers of these channels are (1) write-offs due to the projected insolvency of illiquid and insolvent firms (weighted by outstanding debt and mapped to the sector-by-sector corporate exposure of sample banks), (2) the profitability impact of policy measures (lower provisions for guaranteed loans to solvent firms, loss forbearance on eligible loans under moratoria, and decline in interest income due to duration of debt moratoria), In addition, there is a general change in net operating income after general provisions and losses on other noninterest income due to lower GDP growth and higher unemployment rate, including impairment charges for noncorporate exposures. The calculation does not consider changes in unexpected losses, which are reflected in the risk-weighting of asset exposures in the computation of capital adequacy.1 Debt repayment relief (moratoria) for businesses and households, corporate credit guarantees, deferred insolvency proceedings, and dividend restrictions (only in 2020).

Annex 4. Accounting and Prudential Treatment of Forborne Loans in Normal Times

Banks maintain (largely overlapping) sets of accounting and prudential provisions:

  • Accounting provisions focus on the proper measurement of asset values, and thus net worth, as attested by auditors. They form a key input to standardized, comparable, published financial statements to inform investors and other market participants. Accounting provisions seek to reduce the net value of loans to something approximating their market (but not fire sale) value; they flow through the income statement and appear as negative entries on the asset side of the balance sheet. In the EU, the applicable accounting standard is transitioning from IAS 39 (incurred loss) to IFRS 9 (expected loss), with a phase-in through the end of 2022. Accounting provisions may neither fall short of required amounts (failing to capture risk) nor exceed them (potentially raising questions of tax avoidance or profit smoothing).

  • Prudential provisions add an extra layer of protection at the behest of regulators and supervisors. These may supplement but not interfere with accounting provisions; the term “provision” is something of a misnomer here because, in practice, the process involves deductions from qualifying regulatory capital, which in turn require banks to add a microprudential overlay of additional capital and reserves. Some prudential provisioning requirements apply to all banks (Pillar 1), while others are set by supervisors on a bank-by-bank basis (Pillar 2). In the EU, prudential provisions are specified by the Capital Requirements Regulation (CRR) and Directive (as transposed), EBA technical standards, and EBA and ECB guidelines.

Under IFRS 9, banks must model expected losses based on historical, current, and forward-looking information, including macroeconomic forecasts. Although it is not necessary for a credit event to occur before an expected credit loss is recognized, a missed payment is taken as an indication of increased risk, triggering a higher provision. At origination, the provision must cover expected loss resulting from possible default within 12 months (Stage 1). If thereafter the bank determines that a material increase in credit risk has occurred, the provision must also cover the lifetime expected loss on the loan (Stage 2). Finally, if credit risk is determined to have increased to the point where the loan is impaired (usually at more than 90 days past due), the provision must add coverage of future accrued interest at amortized cost (Stage 3).

CRR, in turn, specifies what measures constitute forbearance, which may trigger reclassification of the loan to nonperforming, resulting in higher prudential provisions. Loan classification criteria are further elaborated in EBA and ECB guidelines issued in 2016 and 2017, respectively, with the former stipulating more than 90 days past due as the default threshold. An ECB addendum on prudential provisioning issued in 2018 prompted accusations of Pillar 2 supervisory powers being misused to achieve Pillar 1 regulatory outcomes, necessitating clarifications and some adjustments by the ECB.

References

  • Aiyar, Shekhar, Wolfgang Bergthaler, Jose M. Garrido, Anna Ilyina, Andreas Jobst, Kenneth Kang, Dmitriy Kovtun, Yan Liu, Dermot Monaghan, and Marina Moretti. 2015. “A Strategy for Resolving Europe’s Problem Loans.” IMF Staff Discussion Note 15/19, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Altavilla, Carlo, Paul Bochmann, Jeroen De Ryck, Ana-Maria Dumitru, Maciej Grodzicki, Heinrich Kick, Cecilia Melo Fernandes, Jonas Mosthaf, Charles O’Donnell, and Spyros Palligkinis. 2021. “Measuring the Cost of Equity of Euro Area Banks.” Occasional Paper Series No. 254, European Central Bank, Frankfurt am Main.

    • Search Google Scholar
    • Export Citation
  • Bhatia, Ashok Vir, Srobona Mitra, Anke Weber, Shekhar Aiyar, Luiza Antoun de Almeida, Cristina Cuervo, Andre O. Santos, and Tryggvi Gudmunds-son. 2019. “A Capital Market Union for Europe.” IMF Staff Discussion Note 19/07, International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Bhatia, Ashok Vir, Andreas A. Jobst, Nazim Belhocine, Srobona Mitra, Ber-gljot Barkbu, and Jan-Martin Frie, 2021, “Guarantees, Moratoria, Other Firm Supports, and Risk Management in European Banking,” IMF Working Paper (forthcoming), International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Bluedorn, John C., Shekhar Aiyar, Romain A. Duval, Davide Furceri, Daniel Garcia Macia, Yi Ji, Davide Malacrino, Haonan Qu, Jesse Siminitz, and Aleksandra Zdzienicka. 2019. “Strengthening the Euro Area: The Role of National Structural Policies in Building Resilience.” IMF Staff Discussion Note 19/05, International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation
  • Boot, Arnoud, Elena Carletti, Hans-Helmut Kotz, Jan Pieter Krahnen, Loriana Pelizzon, Marti Subrahmanyam. 2021. “Corona and Banking – A Financial Crisis in Slow Motion? An Evaluation of the Policy Options.” SAFE White Paper No. 79, Goethe University, Frankfurt am Main.

    • Search Google Scholar
    • Export Citation
  • Borsuk, Marcin, Katarzyna Budnik, and Matjaz Volk. 2020. “Buffer Use and Lending Impact.” Macroprudential Bulletin, Issue 11, European Central Bank, Frankfurt am Main.

    • Search Google Scholar
    • Export Citation
  • Federico J. Díez, Romain Duval, Jiayue Fan, Jose Garrido, Sebnem Kalemli-Özcan, Chiara Maggi, Soledad Martinez-Peria and Nicola Pierri. 2021. “Insolvency Prospects Among Small-and-Medium Sized Enterprises in Advanced Economies: Assessment and Policy Options.” Staff Discussion Note (Washington, D.C.: International Monetary Fund), forthcoming.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ebeke, Christian H., Nemanja Jovanovic, Laura Valderrama, and Jing Zhou. 2021. “Corporate Liquidity and Solvency in Europe during COVID-19: The Role of Policies.” IMF Working Paper 21/056, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Elekdag, Selim, Sheheryar Malik, and Srobona Mitra. 2020. “Breaking the Bank? A Probabilistic Assessment of Euro Area Bank Profitability.” Journal of Banking and Finance 120: 105949.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • European Banking Authority (EBA). 2019. “Basel III Reforms: Impact Study and Key Recommendations – Macroeconomic Assessment, Credit Valuation Adjustment and Market Risk.” Report, European Banking Authority, Paris.

    • Search Google Scholar
    • Export Citation
  • European Banking Authority (EBA). 2020a. “2021 EU-Wide Stress Test: Methodological Note.” European Banking Authority, Paris.

  • European Banking Authority (EBA). 2020b. “EBA Report on the Implementation of Selected COVID-19 Policies.” Report 2020/39, European Banking Authority, Paris.

    • Search Google Scholar
    • Export Citation
  • European Banking Authority (EBA). 2020c. “Report on Benchmarking of National Insolvency Frameworks Across the EU.” European Banking Authority, Paris.

    • Search Google Scholar
    • Export Citation
  • European Banking Authority (EBA). 2020d. “Risk Assessment of the European Banking System.” European Banking Authority, Paris.

  • European Banking Authority (EBA). 2020e. “The EU Banking Sector: First Insights into the COVID-19 Impacts.” Report 2020/17, European Banking Authority, Paris.

    • Search Google Scholar
    • Export Citation
  • European Banking Authority (EBA). 2020f. “Tematic Note on Moratoria and Public Guarantees.” European Banking Authority, Paris.

  • European Banking Authority (EBA). 2020g. “Basel III Reforms: Updated Impact Study.” Report No. 2020/34. European Banking Authority, Paris.

    • Search Google Scholar
    • Export Citation
  • European Banking Authority (EBA). 2021a. “EBA Announces Timing for the Launch of its 2021 EU-wide Stress Test Exercise.” Press Release, European Banking Authority, Paris.

    • Search Google Scholar
    • Export Citation
  • European Banking Authority (EBA). 2021b. “Risk Parameters.” EBA Risk Dashboard, European Banking Authority, Paris.

  • European Commission (EC). 2020. “Temporary Framework to Support the Economy in the Context of the Coronavirus Outbreak: Informal Consolidated Version of the Temporary Framework as Amended on 3 April 2020, 8 May, 29 June and 13 October 2020.” European Commission, Brussels.

    • Search Google Scholar
    • Export Citation
  • European Commission (EC). 2021a. “Targeted Consultation on the Review of the Crisis Management and Deposit Insurance Framework.” European Commission, Brussels.

    • Search Google Scholar
    • Export Citation
  • European Commission (EC). 2021b. “Corporate Solvency of European Enterprises: State of Play.” Note to the Eurogroup Working Group, European Commission, Brussels.

    • Search Google Scholar
    • Export Citation
  • European Commission/European Central Bank/Single Resolution Board (EC/ECB/SRB). 2020. “Monitoring Report on Risk Reduction Indicators.” European Commission, Brussels.

    • Search Google Scholar
    • Export Citation
  • European Central Bank (ECB). 2017. “Guidance to Banks on Non-performing Loans.” European Central Bank, Frankfurt am Main.

  • European Central Bank (ECB). 2019. “Euro Area Bank Profitability: Where Can Consolidation Help?Financial Stability Review, November.

    • Search Google Scholar
    • Export Citation
  • European Central Bank (ECB). 2020a. “COVID-19 Vulnerability Analysis: Results Overview.” Presentation, Banking Supervision, July 28.

    • Search Google Scholar
    • Export Citation
  • European Central Bank (ECB). 2020b. “COVID-19: Recovery and Regulatory Response.” Fireside Chat, Conversation between Kerstin af Jochnick, Member of the Supervisory Board of the ECB, and Tim Adams, President and CEO of the Institute of International Finance, IIF 7th Annual European Banking Union Colloquium, November 17.

    • Search Google Scholar
    • Export Citation
  • European Central Bank (ECB). 2020c. “Euro Area Banking Sector.” Financial Stability Review, November 22.

  • European Central Bank (ECB). 2020d. “Financial Market Pressure as an Impediment to the Usability of Regulatory Capital Buffers.” Macroprudential Bulletin, October.

    • Search Google Scholar
    • Export Citation
  • European Central Bank (ECB). 2020e. “Dividend Distributions During the COVID-19 Pandemic and Repealing Recommendation ECB/2020/35.” Recommendation ECB/2020/62, European Central Bank, Frankfurt am Main.

    • Search Google Scholar
    • Export Citation
  • European Central Bank (ECB). 2021a. “2020 SREP Aggregate Results.” European Central Bank, Frankfurt am Main.

  • European Central Bank (ECB). 2021b. “Guide on the Supervisory Approach to Consolidation in the Banking Sector.” European Central Bank, Frankfurt am Main.

    • Search Google Scholar
    • Export Citation
  • European Central Bank (ECB). 2021c. “FAQs on ECB Supervisory Measures in Reaction to the Coronavirus.” European Central Bank, Frankfurt am Main.

    • Search Google Scholar
    • Export Citation
  • European Systemic Risk Board (ESRB). 2020. “Policy Measures in Response to the COVID-19 Pandemic.” Online Database, European Systemic Risk Board, Frankfurt am Main.

    • Search Google Scholar
    • Export Citation
  • European Systemic Risk Board (ESRB). 2021a. “Macro-financial Scenario for the 2021 EU-wide Banking Sector Stress Test.” European Systemic Risk Board, Frankfurt am Main.

    • Search Google Scholar
    • Export Citation
  • European Systemic Risk Board (ESRB). 2021b. “Financial Stability Implications of Support Measures to Protect the Real Economy from the COVID-19 Pandemic.” Report, European Systemic Risk Board, Frankfurt am Main.

    • Search Google Scholar
    • Export Citation
  • Financial Stability Institute (FSI). 2019. “Definition of Capital in Basel III.” Executive Summary, Financial Stability Institute, Basel.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (IMF). 2018. “Euro Area Policies: Financial Sector Assessment Program-Technical Note-Bank Resolution and Crisis Management.” International Monetary Fund, Washington, DC, July.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (IMF). 2020a. “Kurzarbeit: Germany’s Short-Time Work Benefit,” IMF Country Focus, International Monetary Fund, Washington, DC, July.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (IMF). 2020b. “Bank Capital: COVID-19 Challenges and Policy Responses.” in Global Financial Stability Review, International Monetary Fund, Washington, DC, October.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (IMF). 2020c. “Corporate Liquidity and Solvency in Europe during the Coronavirus Disease Pandemic: The Role of Policies.” In Regional Economic Outlook, International Monetary Fund, Washington, DC, October.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (IMF). 2021a. “Policy Support and Vaccines Expected to Lift Activity.” In World Economic Outlook Update, International Monetary Fund, Washington, DC, January.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (IMF). 2021b. “Vaccines Inoculate Markets, but Policy Support Is Still Needed.” In Global Financial Stability Report Update, International Monetary Fund, Washington, DC, January.

    • Search Google Scholar
    • Export Citation
  • Jobst, Andreas A., and Li Lian Ong. 2020. Stress Testing: Principles, Concepts, and Frameworks. Washington, DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Jobst, Andreas A., Li Lian Ong, and Christian Schmieder. 2013. “A Framework for Macroprudential Bank Solvency Stress Testing: Application to S-25 and Other G-20 Country FSAPsIMF Working Paper 13/68, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jobst, Andreas A., and Anke Weber. 2016. “Profitability and Balance Sheet Repair of Italian Banks.” IMF Working Paper 16/175, International Monetary Fund, Washington, DC.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lozano Guerrero, Silvia, Julian Metzler, and Alessandro D. Scopelliti. 2020. “Developments in the Sovereign-bank Nexus in the Euro Area: The Role of Direct Sovereign Exposures.” Financial Stability Review, European Central Bank, Frankfurt am Main, November.

    • Search Google Scholar
    • Export Citation
  • Mojon, Benoît, Daniel Rees, and Christian Schmieder. 2021. “How Much Stress Could Covid Put on Corporate Credit? Evidence Using Sectoral Data.” BIS Quarterly Review, Bank for International Settlements, Basel.

    • Search Google Scholar
    • Export Citation
  • Organisation for Economic Co-operation and Development (OECD). 2020a. “Supporting Businesses in Financial Distress to Avoid Insolvency During the COVID-19 Crisis.” OECD Policy Responses to Coronavirus (COVID-19), Organisation for Economic Co-operation and Development, Paris, May.

    • Search Google Scholar
    • Export Citation
  • Organisation for Economic Co-operation and Development (OECD). 2020b. “Insolvency and Debt Overhang Following the COVID-19 Outbreak: Assessment of Risks and Policy Responses.” OECD Policy Responses to Coronavirus (COVID-19), Organisation for Economic Co-operation and Development, Paris, November.

    • Search Google Scholar
    • Export Citation
  • Schmitz, Stefan W., Viola Nellessen, Michaela Posch, Peter Strobl. 2021. “Bufer Usability and Potential Stigma Effects.” SUERF Policy Note, Issue No. 219, Österreichische Nationalbank, Vienna.

    • Search Google Scholar
    • Export Citation
1

The IMF October 2020 Global Financial Stability Report (GFSR) includes a broad stress testing exercise of the largest banks globally, which also covers some European banks. Despite a similar top-down stress-testing approach, the GFSR and this paper differ along many dimensions that affect the results. For example, the GFSR uses macroeconomic projections from the October 2020 WEO, whereas this paper uses the January 2021 WEO, by which time growth projections had generally been revised upwards. While the GFSR explicitly accounts for loan guarantee programs and capital relief only, this paper accounts for a broader set of policy measures. Moreover, while the GFSR assumes a static bank balance sheet, this paper uses a two-period model in which bank balance sheets evolve dynamically, with new lending conditioned on the capital position of the previous period.

2

In fact, the aggregate CET1 capital ratio of euro area banks that are directly supervised by the ECB increased by almost 30 basis points from end-2019 to end-September 2020.

1

CET1 capital is the highest quality of regulatory capital, as it absorbs losses immediately when they occur (FSI 2019).

1

Ebeke and others (2021) lay out the technical details for estimating the impact of the pandemic on corporate balance sheets, which varies significantly across countries. The analysis relies on the results from the top-down impact analysis published in the October 2020 REO for Europe to ensure tractability.

2

A recent report by the OECD (2020) also finds that the share of insolvent firms is twice as high for young firms (less than 5 years) compared to the average.

3

The REO finds that policies tend to reduce larger firms’ liquidity and solvency gaps to a greater extent than SMEs’ for a variety of reasons.

1

The measures aimed to avoid the automatic reclassification of loans into forborne or defaulted status in case of generalized moratoria. On March 20, the ECB introduced supervisory flexibility regarding the classification of debtors as “unlikely to pay” on public guarantees granted. It recommended that all banks avoid procyclical assumptions in their models to determine provisions. On March 25, EBA published guidance on the definition of default, forbearance and the application of IFRS 9 in the context of COVID-19, and the European Securities and Markets Authority (ESMA), in coordination with EBA, published a statement providing its opinion on how banks should enforce IFRS 9 accounting standards in light of the crisis.

2

On January 28, 2021, the European Commission announced an increase in the ceiling to €1.8 million per company.

3

In bank accounting, which is on an accrual basis, interest payments that are deferred but not cancelled would still be counted as interest income. However, it is assumed that the moratoria delay the cash receipt of accrued income to a time beyond the stress test horizon, and, thus, imposes an economic loss.

4

The EBA published on April 2, 2020, the Guidelines on Legislative and Non-legislative Moratoria on Loan Repayments (EBA/GL/2020/02), laying out the conditions under which exposures covered by the moratoria should not necessarily be classifed as forborne and, consequently, would not have to be automatically assessed as distressed restructuring under the definition of default. The ECB issued similar guidance to the significant institutions under its direct supervision, advising that undue volatility in provisioning be avoided by focusing on the full life cycle of each loan. Such guidance applies only to the regulatory definition of default and the regulatory classification of forbearance (as well as any related supervisory assessment via the SREP) but does not extend to accounting requirements determining adequate provisioning.

5

Delayed insolvency proceedings do not result in lower provisioning and loss recognition (which take already place at the point of provisioning not exclusively at the point of write-of). However, it is assumed that banks incur losses due to under-provisioning when write-offs occur. In addition, they delay banks’ ability to foreclose and restructure loans after borrowers have defaulted on their obligations, and thus reduce the recovery value of collateral.

1

The exercise in this paper differs from stress tests in FSAPs on several fronts. First, while this paper relies on publicly available data, FSAP stress tests draw heavily on confidential supervisory data. Second, this paper’s exercise focuses on declines in bank capital ratios primarily due to credit risk and does not include interconnectedness within the financial sector or detailed liquidity stress tests, which FSAP stress tests cover.

2

For instance, the specification in Table 2 would suggest negative bank profitability due to the deep economic contraction in 2020—albeit with only a muted rise in unemployment—while euro area banks at end-September 2020 still reported a slightly positive ROA.

3

It is assumed that banks under-provision impaired loans relative to the net present value of collateral and, thus, generate losses. This could occur if debt moratoria delay the timely recognition of deteriorating borrower quality. Moreover, even when debt moratoria expire, a, sudden rise in pent-up bankruptcies could delay debt enforcement and overwhelm the capacity of the court system to manage insolvencies efficiently.

4

In this paper, the debt-weighted share of firms rather than the number of firms is used to quantify potential losses. This is consistent with ECB (2020d) estimates, which show that the share of loans to firms facing liquidity shortfalls is significantly lower than the share of such firms in the count of all firms, due to the loan book being skewed towards larger and less vulnerable firms.

5

Note that this approach generates a point-in-time (PIT) measure of default risk (since the updating is linked to the change in provisions under IFRS-9, which requires the use of PIT PDs). However, the calculation of the regulatory capital only requires through-the-cycle (TTC) parameters, which would result in a lower estimate of PDs under stress.

6

Higher provisions reflect the increase of general default risk of all exposures, including additional impairments of corporate loans due to corporate insolvencies.

7

The extent of capital relief differed greatly among countries, reflecting in part the size of capital buffers prior to the pandemic. For example, the counter-cyclical capital buffer in Ireland was 1 percent, while that in France and Germany was 0.25 percent. In Denmark, following the reduction of the counter-cyclical capital buffer to zero at the onset of the pandemic, it was raised twice (to 1.5 percent in June 2020 and 2.0 percent in December 2020).

8

Note that the most recent macroeconomic projections by the ECB in March 2021 suggest that euro area GDP fell by 6.9 percent in 2020.

1

At the end of 2020, close to one-quarter of the capital base of the large euro area banks comprised various forms of hybrid capital. This share is likely to increase as the European Commission has brought forward legislation to allow greater flexibility in the use of hybrid capital for Pillar 1 and 2 minimum capital requirements in its 2020 Banking Package.

2

Restoring capital buffers might also become more challenging in anticipation of the capital impact of impending regulatory changes. For instance, EBA estimates that banks’ transition to the “Basel IV” regime (and the EU implementation in the Single Rulebook) could reduce average CET1 capital ratios by about 2 percentage points on average (with significant variation across banks) (EBA 2019).

1

For banks in non-EU/European Economic Area countries, which fall outside the scope of the EBA Transparency Exercise, banks’ sectoral exposure is assumed to be the same as the average sectoral exposure in one or more neighboring countries whose banks are included in the EBA Transparency Exercise. For instance, the asset-weighted average exposure of Polish banks is used as proxy for the sectoral exposures of banks in the Czech Republic and the Slovak Republic. In practice, corporate losses could be larger for banks outside the EBA sample as smaller banks tend to lend more to highly affected firms (Diez and others 2021).

2

Given the nascent implementation of the MDA concept in non-EU CESEE countries, the paper does not provide estimates of potential capital shortfalls to this hurdle rate.

1

Looking forward, it is also important to note that some policy measures will have a permanent effect on bank capital even after borrower eligibility has expired. While the effect of debt moratoria will fade over time, in most cases, credit guarantees will remain effective until the maturity date of covered loans, which will permanently reduce LGDs, and, thus, lower expected and unexpected losses reflected in the determination of loan loss coverage through provisions and credit risk weights.

2

In addition to the MDA, the minimum required eligible liabilities (MREL) and the introduction of a binding leverage ratio in 2021 may also constrain the use of capital buffers.

3

The EU-proposed “parallel stacking” as a new approach to determining the output floor for the calculation of the minimum capital requirement could reduce the amount of capital buffers banks would need to replenish; but lowering the minimum capital requirement is also likely to make the MDA more binding for banks (EBA 2020g).

4

The large price decline of European banks’ hybrid capital instruments at the onset of the COVID-19 crisis underscores the importance of the MDA hurdle for the market valuation of banks and their cost of capital during times of stress.

5

This is in line with the ECB’s recent recommendation (ECB, 2020e) on the very restricted resumption of dividend payments; banks are generally encouraged to refrain from or limit shareholder payouts until September 2021.

6

In 2017, ECB-Banking Supervision called for clearly defined internal criteria to identify indicators of unlikeliness to pay (UTP). Banks should ensure that the definition of NPLs and the criteria for identifying UTP are implemented uniformly in all parts of banking groups (ECB, 2017).

7

See also Boot and others (2021) for an overview of potential policy measures to address the rising risk of corporate insolvencies and their impact on the banking sector in Europe.

8

Note that Figure 32 covers a wide range of European banks. Altavilla and others (2021) find that the cost of equity for ECB-supervised euro area banks is lower.

1

For simplicity, the country-specific index is omitted from all equations.

2

The minimum leverage ratio of 3 percent serves a backstop to the more relevant CAR and is not yet fully applicable for EU banks until the end of June 2021.

3

The usage rate of moratoria applies to all banks uniformly in a country and is not sector-specific.

4

The choice of clustering at the country-year level does not consider differences in business model by the banks in the same countries. Clustering also assumes that observations of banks within countries are interdependent for a given year but are independent in the year after.

5

We find that under the baseline, the amount of NPLs of euro area banks could increase to more than €900 billion by 2021 (up from about €500 billion in 2019), which would increase the NPL ratio to more than 6 percent.

6

For instance, the asset-weighted average exposure of Polish banks is used as proxy for the sectoral exposures of banks in the Czech Republic and the Slovak Republic. For other countries, the following proxies (in parentheses) have been chosen: Slovenia (for Bosnia-Herzegovina, Croatia, Montenegro, Serbia), Bulgaria and Romania (for Albania, Belarus, Russia, Turkey, Ukraine), Germany (for Switzerland), and Italy (for San Marino).

7

Note that this assumes that loans are not fully provisioned after accounting for the net present value of collateral, that is, banks carry the loans at positive net asset value.

8

This calculation of effective maturity applies if any element of the advanced IRB approach is used. If the effective maturity cannot be calculated according to the specification above, a more conservative measure of M may be chosen, so that M equals the maximum remaining time (in years). However, national supervisors may allow the effective maturity to be fixed at 2.5 years (that is, “fixed maturity treatment”) for facilities to certain smaller domestic corporate borrowers.

9

Note that the specification of credit risk weights using the advanced IRB formula in equation (A2.39) is concave on the PD value. Thus, using the average PD of the portfolio (instead of computing the weighted average of the credit risk weights calculated at loan level), may generate a conservative estimate.

10

The EBA Risk Dashboard is part of the regular risk assessment conducted by the EBA and complements the Risk Assessment Report.

COVID-19: How Will European Banks Fare?
Author: Mr. Shekhar Aiyar, Mai Chi Dao, Mr. Andreas A. Jobst, Ms. Aiko Mineshima, Ms. Srobona Mitra, and Mahmood Pradhan
  • View in gallery

    Euro Area Banks: Change of CET1 Capital Ratio under Different Assumptions

    (Percentage points, end-2021 relative to end-2019)

  • View in gallery

    Dispersion of Change in Asset Risk Weights (Baseline Scenario)

    (Percent)

  • View in gallery

    Euro Area Banks: CET1 Capital Ratio (Baseline Scenario), Extended Coverage

    (Percent)

  • View in gallery

    Solvency Stress Test—Dispersion of CET1 Capital Ratio (Adverse Scenario/Extended Coverage)

    (Percent)

  • View in gallery

    Euro Area Banks: CET1 Capital Ratio (Adverse Scenario), Extended Coverage

    (Percent)

  • View in gallery

    Euro Area: Potential Capital Need and Number of Banks below Thresholds

    (EUR billion/count)

  • View in gallery

    European Banks: Changes in Capital Ratio and GDP Growth, Extended Coverage

    (Percent)

  • View in gallery

    Euro Area Banks (Extended Coverage): Solvency Stress Test—Baseline Scenario

    (Percent)

  • View in gallery

    Euro Area Banks (Extended Coverage): Solvency Stress Test—Adverse Scenario

    (Percent)

  • View in gallery

    EU Banks (Extended Coverage): Solvency Stress Test—Baseline Scenario

    (Percent)