## 1deuea2022009 ## Source details **Canonical URL:** [1deuea2022009](https://www.imf.org/-/media/files/publications/cr/2022/english/1deuea2022009.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2022/english/1deuea2022009.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2022/english/1deuea2022009.pdf.json) --- ### Executive summary — background and macro‑financial context - GDP contracted by 4.6 percent in 2020. - Total assets of German banks increased by 10.5 percent (€0.9 trillion, equivalent to 24 percent of annual GDP) between December 2019 and December 2021. - About two-thirds of this increase (€560 billion) was funded by interbank liabilities, including toward the ECB. - Customer deposits increased by €365 billion. - Bank lending rose 5.4 percent in real terms; over half of the increase in loans was driven by savings and cooperative banks. - Housing mortgages increased 7.1 percent y-o-y to September 2021. - In 2022, financial conditions started tightening with 10-year Bund yields rising 100 basis points. - WEO baseline: gross impact of the Russia-Ukraine war estimated at 2.5 percent of GDP in 2022 (about 0.5 percent to be offset by fiscal relief measures). - Surge in energy and food prices likely to translate into inflation of around [5.5] percent on average for 2022. - Eurostat: 50-75 percent of Germany’s natural gas imports came from Russia in 2021 Q1. ### Key vulnerabilities and risks - Structural vulnerabilities that have become more prominent: - Low bank profitability (limited non-interest revenues, cost factors, and competition). - Misalignments in real estate sector prices. - Real estate tail risks: - Standard indicators suggest residential real estate (RRE) overvaluation of 21-37 percent deviation from long-run averages as of end-2021. - An econometric model accounting for real interest rates suggests RRE overvaluation of about 10-15 percent as of 2021Q3. - CRE tail risks increased since the pandemic; probability of negative one‑year ahead real price growth: RRE 2.2 percent (from 0.7 percent end-2019), CRE 66 percent (from 24 percent end-2019). - Under severe adverse scenarios (5th percentile, medium-term): RRE could fall by 14 percent; CRE could fall by 30 percent. - Banking sector’s direct exposures to Russia are limited; broader economic fallout could affect institutions, NPLs, and house prices. - NFC sector: - NFC total income declined by €220 billion (3.1 percent) in 2020; corporate profits before taxes declined by 3.8 percent. - Gross consolidated NFC debt rose by 5 percentage points of GDP in 2020 to 62.5 percent of GDP. - Net consolidated NFC debt barely increased in 2020. - By end-2020, debt at risk (listed firms with ICR<1) was 16 percent of total corporate debt (up from 4 percent at end-2019). - NFC non-performing bank loans were €39 billion and 2.3 percent of gross loans by September 2021. ### Solvency stress‑test findings and methodology - Scope and data: - SIs: end-2021 data (16 SIs under IFRS9 included in solvency test; 5 SIs under national GAAP excluded). - LSIs: 2021:Q3 data; 1,293 LSIs covered in solvency perimeter. - Coverage: 16 SIs and 1,293 LSIs account for about 87 percent of bank assets; 89.5 percent of SIs’ assets included in stress test given exclusions. - Hurdle rate for CET1: 8.25 percent (4.5 percent minimum CET1 + 2.5 percent conservation buffer + 0.75 percent CCyB starting 2023 + 0.5 percent systemic risk buffer). G‑SII, O‑SII, Pillar 2R and Pillar 2G excluded. - Key modelling features: - Static balance sheet assumption; no managerial action in baseline simulations. - Conservative assumption: shocks to policy/short-term rates are fully passed through to banks’ funding costs; deposit-to-lending pass-through implied smaller than one by FSAP NII satellite model. - IFRS9 transition matrices, PiT and TTC PDs used for SIs; EBA 2021:Q3 aggregates used for LSIs. - Credit risk: bank-level panel regressions for NFCs; household micro-simulation based on ECB HFCS (2017) for household default. - Baseline solvency results (SIs): - Aggregate CET1 remains above 13.6 percent. - Regulatory capital increased to 18.8 percent (end-2021); Tier 1 to RWA 16.8 percent. - Adverse scenario solvency (SIs): - Aggregate depletion of CET1 reaches 5.2 percent of RWAs by end-2023 from 14 percent and increases subsequently (no write-offs assumption). - Aggregate capital shortfall for SIs remains small at 0.3 percent of GDP (euros 20 bn) under conservative assumptions. - Three SIs out of fifteen in the sample fall below the hurdle rate assuming a binding CCyB in 2023-24, but all remain above minimum CET1 ratio. - If banks write-off non-performing exposures: CET1 depletion 5.0 percentage points of RWAs by 2023; capital shortfall euros 18.7 bn (0.52 percent of GDP). - If CCyB released: capital shortfall reduces to euros 14 bn (0.39 percent of GDP) no write-offs; euros 200 mil (0.01 percent of GDP) with write-offs. - LSIs solvency results: - Baseline aggregate CET1 rises to 17.7 percent of RWAs. - Under baseline 8-9 LSIs fail hurdle rate (accounting for 1.55-1.7 percent of LSI assets). - Adverse scenario aggregate CET1 about 16.3 percent of RWAs; 18-21 LSIs fail hurdle rate (accounting for 2.83-3.04 percent of LSI assets). - Recapitalization needs: baseline 0.016-0.018 percent of GDP; adverse 0.047-0.051 percent of GDP. ### Liquidity stress‑test findings and methodology - Scope and sample: - All 17 SI banks included; LSI sample: 40 randomly selected banks across categories (10 commercial, 10 savings, 15 cooperative, 5 building & mortgage). - LSI sample assets represent about 3 percent of total LSI sector assets. - Liquidity horizon: one year from reference date (Sept 2021 for LSIs; Dec 2021 for SIs). - Key liquidity indicators (Sep 2021 snapshot): - Average liquid assets-to-total assets (LATA): 28.3 percent. - LCR: 163 percent. - NSFR: 125 percent. - Average LATA of commercial banks and Landesbanken at end-2019: 30.6 percent. - Scenarios and parameters: - Baseline (benchmark), Scenario A (“new wave”), Scenario B (“risk aversion”), Scenario C (“persistent inflation”). - Selected roll-off and haircut parameters preserved exactly in scenario parameter table (see source). - Cash‑flow stress‑test results — LSIs: - Up to 3 months: - Scenario A: 4 out of 40 LSIs become illiquid. - Scenario B: 5 out of 40. - Scenario C: 8 out of 40. - At 12 months: - Scenario C: 12 out of 40 LSIs become illiquid. - Worst-case cumulative liquidity shortfall (Scenario C at 12 months): less than 1 percent of total assets of sample banks. - Average CBC holdings: about 20 percent of assets; CBC composition: cash, central bank reserves and level 1 tradable assets make up about two thirds. - CBC declines (average LSI Scenario C): use about one-half of initial CBC in first month; three-fourths by end of year. - Cash‑flow stress‑test results — SIs: - No SI becomes illiquid before six months under any scenario. - Within one year: - Scenario B: two SI banks exhaust liquidity buffers. - Scenario C: three SI banks exhaust liquidity buffers. - Scenario A: none exhaust buffers. - Shortfalls for affected SIs: about 0.1 percent of sample assets and 1-2 percent of equity; for affected SIs average shortfall about 1.5 percent of assets. - CBC usage (SIs): Scenario A <25% of initial CBC in 12 months; Scenario B >50%; Scenario C ≈60%. - US dollar liquidity: - Many banks report USD as significant; about ¾ of reporting banks have USD LCR below 80 percent. - Some banks have insufficient USD CBC; under Scenario C within 5 days USD shortfall could reach 0.3 percent of assets for affected SIs and 1 percent of assets for affected LSIs; after one month values rise to about 1½ percent of assets. - Recommendation: strengthen monitoring of USD liquidity exposures and contingency access to FX swap lines as needed. - Historical liquidity performance: - Aside from temporary strains in March–April 2020, German banks did not experience significant liquidity strains during the pandemic; average loan-to-deposit ratio around 80 percent. ### Interconnectedness and contagion analysis - Interbank structure and concentration: - Germany’s interbank system strongly interconnected; a relatively small number of banks account for a large share of interconnections. - Interbank market segmented among SIs and among LSIs. - Contagion simulation key findings: - Directional contagion losses: - 51 percent of contagion losses caused by contagion from SIs to LSIs. - 36.06 percent from LSIs to SIs. - 13.03 percent from contagion among SIs or among LSIs. - A few large banks account for most contagion risks; top 10 most contagious banks account for most losses. - Amplification: first-round losses dominate; higher-order rounds increase with shock severity, especially when small bank is trigger. - Sectoral transmission: large shares of contagion losses occur within cooperative and savings networks (e.g., Credit Cooperatives to Credit Cooperatives 44.7 percent; Landesbanken to Savings Banks 21.3 percent; Savings Banks to Landesbanken 15.2 percent). - Data and scope: - Network constructed from Bundesbank credit registry as of 2021:Q3; sample sizes: 1,297 lending banks / 1,277 borrowing banks. ### Corporate sector (NFC, SMEs) risk analysis and sensitivity testing - Pre-pandemic NFC performance: - NFC sector on average performed slightly better in 2000–2019 than in 1989–2000; SMEs’ RoA trended upward since GFC. - Capital ratio of large enterprises rose from 28 percent in 2000 to average 32 percent in the ten years preceding pandemic. - SME equity capital ratio: 29 percent in 2019. - Pandemic impact and resilience: - NFC total income decline €220 billion (3.1 percent) in 2020; expenses fell similarly; annual result before taxes remained 4.1 percent of sales. - KfW SME panel: aggregate SMEs broadly unchanged return on sales; smallest SMEs faced particular difficulty and drew on buffers. - By mid-2021 gross unconsolidated NFC debt 72 percent of GDP. - Sensitivity analysis (Tressel and Ding (2021)) on listed firms: - Sample: 263 listed firms; analysts’ forecasts implied 2020 overall sales 17 percent lower than start-year forecasts. - Pre-shock (end-2019): 23 percent of listed firms had ICR<1; these accounted for 4 percent of total debt. - Under sales shocks with varying production cost offsets (y): - If y = 100%: share of firms with ICR<1 rises to 60 percent; debt-at-risk rises to 41 percent of total NFC debt. - If y = 0%: share of firms with ICR<1 rises to 85 percent; debt-at-risk rises to 83 percent of total NFC debt. - Observed outcome 2019–2020 for sample: sales decline contained to 8 percent; debt-at-risk rose to 16 percent; median ICR declined from 7.2 to 5.7. - Dynamic scenario-based firm stress tests: - Baseline: ICRs improve over time; share of debt in firms with ICR<1 and share of debt in firms with cash<0 fall. - Adverse scenario: contraction in 2023 drives ICR deterioration, share of debt in ICR<1 rises in 2023; share of debt in cash<0 remains more elevated; probability of default rises in 2023 (from a low base). ### Real estate sector risks and bank exposures - RRE developments: - Since 2010, nominal RRE prices increased about 91 percent by end-2021 (63 percent in real terms during 2010-21). - RRE price growth in largest 7 cities: 144 percent nominal (109 percent real) during 2010-21. - Outstanding bank loans for house purchases at 2021Q3: exceeded EUR 1.6 trillion (46 percent of GDP). - Growth of housing loans to domestic enterprises and resident individuals: 7.1 percent y/y in 2021Q3. - CRE dynamics: - CRE prices declined 0.8 percent in 2021 (office prices -0.2 percent in 2021 after H1 decline; retail property -3.1 percent in 2021). - Banks’ exposures to CRE sectors about EUR 602 billion (6 percent of assets and 19 percent of total loans to enterprises and households). - PaR and tail-risk findings: - RRE one-year-ahead negative growth probability increased to 2.2 percent (from 0.7 percent end-2019). - CRE one-year-ahead negative growth probability increased to 66 percent (from 24 percent end-2019). - Adverse scenarios: RRE cumulative fall about 14 percent over 3 years; CRE cumulative fall about 30 percent over 3 years. - Data gaps and supervisory actions: - Authorities lack comprehensive borrower-based data (e.g., LTV, DSTI); Deutsche Bundesbank to begin regular data collection on lending standards starting 2023Q1. - BaFin introduced macroprudential policy package including a sectoral systemic risk buffer on RRE exposures. - Banking sector soundness indicators (September 2021 snapshot, selected): - All banks aggregate: Tier 1 Capital Ratio to RWA 16.8; CAR 18.8; Liquid assets to total assets 26.1; LCR implicit in sector averages above 100; ROE 3.9; NPL to gross loans 1.4; Provisions to NPLs 35.4. ### Policy recommendations and supervisory priorities - Continue close monitoring of banks’ prudential ratios, particularly large SI commercial banks. - Establish microprudential buffers (Pillar 2 guidance) for less capitalized banks as needed. - Strengthen LSIs interest rate risk monitoring by collecting data on: - Remaining maturity of retail deposits, - Wholesale funding, and - Interest-bearing assets to perform top‑down interest rate stress tests. - Strengthen monitoring and analysis of domestic and cross‑border interconnectedness, focusing on key domestic interbank market institutions and markets with concentrated exposures. - Strengthen data sharing between the Bundesbank and the ECB to support risk monitoring and analysis. - Improve borrower-based data collection (LTV, DSTI) to better assess mortgage vulnerabilities; Deutsche Bundesbank to start regular data collection on lending standards in 2023Q1. - Monitor and, if needed, ensure access to central bank swap lines for U.S. dollar liquidity pressures. - Incorporate adverse PaR scenarios (RRE cumulative fall ~14 percent over 3 years; CRE cumulative fall ~30 percent over 3 years) into supervisory stress testing and contingency planning. - For selected banks with liquidity weaknesses, recommend lengthening tenors of deposits from non-financial corporations and adjusting/increasing CBC composition. *Prepared by Gerard Almekinders, Dan Cheng, Alla Myrvoda, Marco Pani, Thierry Tressel, and Sebastian Weber.* ### EXECUTIVE SUMMARY __________________________________________________________________________ 7 ### EXECUTIVE SUMMARY ### Background and macro-financial context - GDP contracted by 4.6 percent in 2020, less than most European peers. - Total assets of German banks increased by 10.5 percent (€0.9 trillion, equivalent to 24 percent of annual GDP) between December 2019 and December 2021. - About two-thirds of this increase (€560 billion) was funded by interbank liabilities, including toward the ECB. - Customer deposits increased by €365 billion. - Bank lending rose 5.4 percent in real terms; over half of the increase in loans was driven by savings and cooperative banks. - Housing mortgages increased 7.1 percent y-o-y to September 2021. - In 2022, financial conditions started tightening with 10-year Bund yields rising 100 basis points. - The WEO baseline estimates the gross impact of the Russia-Ukraine war at 2.5 percent of GDP in 2022 (about 0.5 percent to be offset by fiscal relief measures). - The surge in energy and food prices is likely to translate into inflation of around [5.5] percent on average for 2022. - Eurostat: 50-75 percent of Germany’s natural gas imports came from Russia in 2021 Q1. ### Key vulnerabilities and risks - Two structural vulnerabilities identified in the 2016 FSAP have become more prominent: - Low bank profitability (limited non-interest revenues, cost factors, and competition). - Misalignments in real estate sector prices. - Tail risks in real estate have increased since the onset of the pandemic, particularly in the CRE market. - Standard indicators suggest residential real estate (RRE) overvaluation of 21-37 percent deviation from long-run averages as of end-2021. - An econometric model that accounts for real interest rates suggests RRE overvaluation of about 10-15 percent as of 2021Q3. - The banking sector’s immediate direct exposures to Russia are limited, making direct risks manageable; however, economic fallout from the war could affect individual institutions, non-performing loans, and house prices. - NFC sector: - The pandemic caused a large shock to enterprises’ sales. - In the baseline scenario enterprises’ capacity to service debt (ICR) improves over time, with the share of debt in firms with an ICR<1 and the share of debt in firms with cash<0 falling over time. - In the adverse scenario, a significant contraction in 2023 causes the share of debt in firms with an ICR<1 to rise again; the share of debt in firms with cash<0 remains more elevated than in the baseline and the probability of default rises in 2023 (albeit from a low base and remaining relatively low in absolute terms). ### Solvency stress-test findings - SIs and LSIs are resilient under the baseline and adverse scenarios overall. - Under the V-shaped adverse scenario with conservative assumptions on interest rate pass-through: - Aggregate capital shortfall for SIs remains small at 0.3 percent of GDP. - Three SIs out of fifteen in the sample fall below the hurdle rate assuming a binding CCyB in 2023-24, but all banks remain above the minimum CET1 ratio. - LSIs’ aggregate capital remains very high; only 21 very small banks (with up to 3 percent of total LSIs assets) fall below the hurdle rate. - The stress-test analysis is performed under the conservative assumption that shocks to policy rates are fully passed through to funding costs of banks, and the speed of pass-through is consistent with supervisory reporting in the ECB “Interest Rate in the Banking Book” template. The FSAP NII satellite model implies pass-through from deposit rates to lending rates is smaller than one; these assumptions lead to conservative IRRBB estimates. ### Liquidity stress-test findings - The banking system appears generally resilient to liquidity stress. - Under the liquidity severe adverse scenario, selected banks’ U.S. dollar exposures could pose a risk and require access to the central bank swap line. - The FSAP recommends strengthening monitoring of U.S. dollar liquidity exposures and contingency access to FX swap lines as needed. ### Interconnectedness and contagion - Germany’s interbank system is strongly interconnected; a relatively small number of banks account for a large share of interconnections. - The interbank market appears segmented among SIs and among LSIs. - Domestic interbank contagion risks flow mostly: - From SIs to LSIs; and - From LSIs as a group to SIs. - Germany’s financial system is highly interconnected across borders through financial claims and liabilities. - The FSAP recommends continued strengthening of monitoring and analysis of domestic and cross-border interconnectedness, focusing on key domestic interbank market institutions and markets where exposures are concentrated. ### Policy recommendations and supervisory priorities - Continue closely monitoring banks’ prudential ratios, particularly large SI commercial banks. - Establish microprudential buffers (Pillar 2 guidance) for less capitalized banks as needed. - Strengthen LSIs interest rate risk monitoring, including by gathering data on: - Remaining maturity of retail deposits, - Wholesale funding, and - Interest-bearing assets to perform top-down interest rate stress tests. - Strengthen data sharing between the Bundesbank and the ECB to support risk monitoring and analysis. - Monitor and, if needed, ensure access to central bank swap lines for U.S. dollar liquidity pressures. ### Real estate and bank exposures - High RRE valuations and CRE tail risks imply potential pockets of vulnerability in bank exposures to real estate, particularly in larger cities. - The FSAP welcomes progress toward closing residential real estate data gaps to support risk monitoring and calibration of macroprudential tools. *Prepared by Gerard Almekinders, Dan Cheng, Alla Myrvoda, Marco Pani, Thierry Tressel, and Sebastian Weber.* ### 4. The banking sector’s structure is large and complex. The banking sector is comprised of ### 4. The banking sector’s structure is large and complex. The banking sector is comprised of ### Banking sector structure and size - More than 1400 entities, including local subsidiaries and branches of foreign banks. - As of end-2021: - Total assets: euro 9.2 trillion (equivalent to about 2 ½ times annual GDP). - Total equity invested in the sector: euro 565 billion (about 15 percent of GDP). - Average capital/assets ratio: 6.1 percent. - The LSI sector is large, with total assets amounting to about 60 percent of total banking system assets. - Table summary (December 2021, aggregate counts from source): - SIs: 17 - LSIs: 1,321 - Total institutions reported: 1,338 - Note: Five SIs with national GAAP accounts are not included in the solvency stress test. Excluding 109 branches of foreign banks and four SIs which are holdings. Excluding 4 systemically important entities classified as “financial holdings” by the ECB. ### Three pillars of the banking system (Table 2 and descriptive text) - Private “pillar”: - About 170 banks (about 140 commercial banks and the others real estate and specialized banks). - Accounts for 44 percent of the assets of the banking sector (about 100 percent of GDP). - Includes systemically important institutions, including Deutsche Bank with 1.3 trillion EUR in assets ([37] percent of GDP, 14.3 percent of the banking sector). - Public “pillar”: - Includes savings banks (Sparkassen) and regional banks (Landesbanken). - Savings banks: owned by local governments, pursue social objectives, do not distribute dividends but capitalize net earnings as net equity. - Regional banks: publicly-owned, provide wholesale services to savings banks, hold a large share of savings banks’ liquidity in deposits, support local businesses and development projects. - Cooperative “pillar”: - About 800 “primary” banks and umbrella organizations (Volksbanken, Raiffeisenbanken, Sparda banks, PSD banks). - Organized in financial networks and supported by institutional network organizations; notable central organization: DZ (acts as liquidity manager). ### Banks by category (December 2021, supervisory data) - Aggregate totals: - Total banks: 1,337 - Total assets (reported table): 8,777.3 (billions) - Average assets per bank: 6.6 (billions) - Selected category breakdown (from Table 2; preserve numbers and labels): - Private — Commercial: Number of banks 142; Total assets percent 10.6; Total assets billions 3,351.1; percent 38.2; Average assets per bank 23.6 (billions). - Private — Real estate: Number of banks 27; Total assets percent 2.0; Total assets billions 482.1; percent 5.5; Average assets per bank 17.9 (billions). - Public — Regional banks (Landesbanken): Number of banks 6; Total assets percent 0.4; Total assets billions 804.8; percent 9.2; Average assets per bank 134.1 (billions). - Public — Savings banks: Number of banks 371; Total assets percent 27.7; Total assets billions 1,550.5; percent 17.7; Average assets per bank 4.2 (billions). - Cooperative — Cooperative banks (excl. DZ): Number of banks 772; Total assets percent 57.7; Total assets billions 830.4; percent 9.5; Average assets per bank 1.1 (billions). - Cooperative — Other banks (Includes DZ): Number of banks 20; Total assets percent 1.4; Total assets billions 1,758.4; percent 20.0; Average assets per bank 92.5 (billions). - Note: Excluding 109 branches of foreign banks and 4 SIs which are holdings. The data for DZ Bank included in the total are indicative Fund staff estimates. ### Cooperative banking sector (Box 1 highlights and statistics) - Legal form: “registered cooperative” (eingetragene Genossenschaft, e.G.). - Organization: National Association of German Cooperative Banks (BVR e.V.), five regional associations, two special associations, the Genossenschaftliche Finanzgruppe Volksbanken Raiffeisenbanken (GFVR), and DZ Bank central group. - As of December 31, 2021, the network includes 772 “primary” cooperative banks. - Box Table composition (as of December 2019; quintiles refer to asset distribution of all German banks): - DZ Bank: Number of banks 1; % of total 0.1; Total Assets 284; % of total 21.9. - Coops in top quintile: Number of banks 52; % of total 7.1; Total Assets 436; % of total 33.7. - Coops in 4th quintile: Number of banks 113; % of total 15.4; Total Assets 263; % of total 20.3. - Coops in 3rd quintile: Number of banks 171; % of total 23.3; Total Assets 192; % of total 14.8. - Coops in 2nd quintile: Number of banks 226; % of total 30.7; Total Assets 95; % of total 7.3. - Coops in 1st quintile: Number of banks 171; % of total 23.3; Total Assets 24; % of total 1.8. - Total (reported): Number of banks 734; % of total 100.0; Total Assets 1,294; % of total 100.0. - Cooperative sector operational statistics: - About 18.4 million members. - Serve about 30 million customers. - Employ 170,000 people in about 8,000 branches. - Hold assets for about €1,140 million (12 percent of the German banking sector). - Hold about one-fifth of customer deposits (but more than 30 percent of savings deposits). - DZ Bank (holding) with €310 billion of assets and about 13,000 employees; accounts for about one-fourth of total assets of the German cooperative banking sector. - Liquidity features and implications: - Many cooperative banks hold a large share of liquidity as claims on DZ Bank and other central institutions. - BVR recommends cooperative banks place at least 7.5% of liquidity collected from customer deposits in claims on DZ Bank. - Central organizations can provide liquidity support, including lending securities usable as collateral to borrow from the central bank. - Individual cooperative banks’ liquidity buffers, if defined to include interbank deposits at DZ Bank, may underestimate available liquidity under idiosyncratic strain; but common stress would require netting interbank deposits as a liability for DZ Bank. ### Banking sector developments since COVID-19 (Section C highlights) - Overall indicators remain sound, reflecting supporting measures. - Capital and solvency (end-2021): - Regulatory capital increased to 18.8 percent. - Regulatory Tier 1 capital to risk-weighted assets increased to 16.8 percent. - All types of banks have buffers well above the regulatory minimum. - Earnings retention partly explains the positive trend. - Asset quality: - NPLs to total gross loans: 1.5 percent in 2021 (low and declined since last FSAP). - The unwinding of support may uncover pockets of vulnerabilities. - FX net foreign position to capital: - Slight decline to 3.4 percent in 2020 from 4 percent in 2016; reversed to 4.4 percent by end-2021. ### Scope and scenarios of the risk analysis (Section D) - Risk analysis components: - Solvency and liquidity stress tests of the banking sector. - Interconnectedness and contagion analysis. - Special topics: non-financial corporate risk analysis and real estate market risk analysis. - Data vintage used: - Banking risk analysis: end-2021 data for Significant Institutions (SIs); 2021:Q3 data for Less Significant Institutions (LSIs). - Liquidity stress tests: - Forward-looking, cash-flow-based, up to a one-year time horizon from the reference date (September 2021 for LSIs and Decemeber-2021 for SIs). - Based on a “going concern” hypothesis and contemplates four different scenarios. - Sample: random sample of LSI banks and 17 SI banks. Tests exclude 4 SI banking institutions classified as “financial holdings.” - Contextual caveats: - Analysis performed amid the pandemic and the development of the war in Ukraine. - Results based on a severe adverse scenario provide conditional quantifications of downward shocks and channels of contagion. ### Risk and vulnerabilities — NFC balance sheets and SMEs (Section E highlights) - Pre-pandemic NFC sector performance was strong overall; SMEs particularly stood out. - Comparative periods: - In the 10 years between the GFC and the start of the pandemic in 2020, NFC sector on average achieved slightly better results than during the 11 years before the GFC. - SMEs performance: - SMEs’ return on assets (RoA) trended upward since the GFC. - Large firms’ RoA fluctuated around the longer-term average of 3.5 percent. - SMEs’ performance improvement most pronounced outside manufacturing. - Capitalization and indebtedness: - Capital ratio of large enterprises rose from 28 percent in 2000 to an average of 32 percent during the ten years preceding the pandemic. - Sample of about 230 non-financial enterprise groups listed in Germany in the Prime Standard segment maintained an average capital ratio of 29.5 percent. - SME sector’s equity capital ratio: 29 percent in 2019 (close to that of large firms). - Rising capitalization enabled SMEs to deleverage and reduce the share of bank loans in total liabilities. - Germany’s NFC sector capital ratio caught up with leading European peers. - Interest coverage and indebtedness trends: - SMEs’ interest coverage ratio (ICR = EBIT over interest payments) rose after the GFC due to improving EBIT and falling interest burden. - Large firms’ aggregate ICR hovered around the long-term average of 3.5. - Deutsche Bundesbank defines SMEs as enterprises with sales of less than €50 million; KfW defines SMEs as enterprises with turnover of up to €500 million. - Zombification context: - Deutsche Bundesbank granular analysis suggests prevalence of zombification broadly stable at about 6 percent of all NFC firms over the past decade. - Banerjee and Hofmann (2021), looking only at listed firms, finds zombie share broadly stable at about 10 percent of firms. *Source: IMF staff elaboration based on Bundesbank, ECB, supervisory data, and other public sources as presented in the provided chapter content.* ### 12. The various support measures implemented by the authorities alongside the resilience ### 1deuea2022009 - 12. The various support measures implemented by the authorities alongside the resilience ### Nonfinancial corporate (NFC) sector performance and resilience - NFCs’ total income declined by €220 billion, or 3.1 percent in 2020 (preliminary Deutsche Bundesbank (2021) data). - Enterprises’ total expenses before taxes on income were lower by a similar amount; corporate profits before taxes declined by 3.8 percent. - Annual result before taxes on income remained virtually unchanged at 4.1 percent of sales (measured relative to reduced revenue figures). - KfW SME panel (9,889 enterprises with annual turnover of less than €500 million) suggests aggregate SMEs managed a broadly unchanged return on sales; the smallest SMEs experienced particular difficulty and drew on buffers, which may be thin if operating in sectors also weak in 2021. - Distribution of listed NFCs’ RoA fell to levels last observed during the GFC, but the 2020 RoA downturn is milder than in 2000/01 and 2008/09; a slide had started in 2018 due to weakening demand from China. - Of 57 listed firms with an ICR below 1 in 2019 and present in the 2020 sample, 17 firms with total debt of €17.5bn improved earnings in 2020, pushing their ICR above 1; these firms operate in sectors such as (solar) energy, telecom, food, pharmaceuticals, e-commerce, construction, and salt and sugar production. - Firms improved debt maturity profiles: falling share of short-term debt reduced rollover risks at the aggregate level. ### Debt, cash holdings, insolvencies, and nonperforming loans - Gross consolidated NFC debt rose by 5 percentage points of GDP in 2020, to 62.5 percent of GDP, equaling the 2009 GFC high. - About two thirds of the increase in the debt ratio was due to increased borrowing and the remaining one third was due to the 3 percent decline in nominal GDP in 2020. - The 5 percentage points rise partly reflected drawdown of credit lines for liquidity and was broadly matched by rising NFC cash holdings. - Net consolidated NFC debt (nets out cash holdings) barely increased in Germany in 2020; Germany’s net NFC debt to GDP ratio and NFC debt to income ratio remain below peers. - German NFCs’ unconsolidated gross debt was 72 percent of GDP in mid-2021 (median for G20 countries). - High-yield corporate bonds outstanding: 1.7 percent of GDP (less than half euro area peers and about a fifth of the United States). - Investment-grade corporate bonds: 16.4 percent of GDP (compared to 21.5 percent of GDP in the euro area and about 29 percent of GDP in the US). - Bankruptcies of consumers and self-employed jumped in 2021; corporate insolvencies fell for the 12th year in a row in 2021 to the lowest level since the adoption of the 1999 Insolvency Act. The May 2021 full reinstatement of the obligation to file insolvencies after suspension has not yet led to corporate insolvencies picking up. - By end-2020, debt at risk (total debt of listed firms with ICR below 1 as a percentage of total debt of all listed firms) was 16 percent of total corporate debt (up from 4 percent at end-2019). - Preliminary data for the first nine months of 2021 suggests recovery of listed NFCs’ earnings in manufacturing and services, expected to stem the rise in debt at risk. - By September 2021, non-performing bank loans to the NFC sector were €39 billion and 2.3 percent of gross loans, back at pre-pandemic lows. ### Listed non-financial groups (Bundesbank data) - On the eve of the pandemic, listed non-financial groups’ total revenues were about 51 percent of GDP per annum; their debt had risen to 28 percent of GDP by end-2019. - Debt-to-equity ratio reverted to the long-term average in H1 2021 thanks to surging equity. - Pandemic extended the 2018/19 slide in profits; groups focused on production saw weighted average profitability fall to 0 in the first half of 2020; services-focused groups experienced a less deep decline and a stronger recovery in subsequent semesters. ### Banking sector profitability and vulnerabilities - Aggregate ROE of German banks: 2.7 percent in 2020 (fell short of cost of capital estimated in the range of 8-12 percent by European banks). - German banks reported lower return on total assets, on risk-weighted assets, and on equity than EU averages in 2020. - Drivers: low interest rates squeezing net interest income; limited growth in fee and commission income due to customer risk aversion, preference for savings, limited customer experience with fee-based products, strong competition, and dense branch networks. - Heterogeneity across bank types: savings banks and cooperatives outperformed private banks; all bank groups have experienced declining profitability over time. - Savings and cooperative banks’ higher profitability was supported by increased lending volumes (including real estate) and regional focus shielding them from competition; this raised their market share in lending to households and enterprises (particularly SMEs). - Rising risks challenge sustainability of saving and cooperative banks’ business models: - Increased lending, particularly for residential and commercial real estate, supported revenues but raises vulnerabilities. - BaFin introduced a macroprudential policy package, including a sectoral systemic risk buffer on RRE exposures. - Other rising risks: interest rate changes, slow digitalization, cyber risks, geopolitical developments. - Recommendation: accelerate efforts to collect data on bank exposures to various risks for ongoing monitoring. ### Real estate market dynamics - Since 2010 upswing, RRE prices increased by about 91 percent by end-2021 in nominal terms (63 percent in real terms during 2010-21). - Prior to 2010 (1990-2009), nominal RRE prices increased about 25 percent. - RRE price growth in the largest 7 cities: 144 percent nominal (109 percent real) during 2010-21. - CRE dynamics diverged during the pandemic: - Overall CRE prices declined by 0.8 percent in 2021 (they had increased by 6.4 percent in 2019). - Office prices: after declining in 2021H1, grew by 0.2 percent in 2021 (9.6 percent in 2019). - Retail property prices (largely nonfood): continued to fall, -3.1 percent in 2021, pressured by e-commerce and aggravated by the pandemic. - CRE sub-sectors vary by location: prime-location offices in larger cities performed better than properties in less populated areas; food-related retail properties, particularly in prime locations, were in greater demand than non-food retail such as shopping malls. ### Macro-financial scenarios and solvency stress tests - Key macro-financial risks assessed (RAM: Table 6): - Escalation of the conflict associated with a Russian gas shut off, higher commodity prices. - Global resurgence of COVID-19 with extended supply chain disruptions. - De-anchoring of inflation expectations in the U.S. and advanced Europe, leading to rising core yields and risk premia. - Global resurgence of COVID-19 with extended supply chain disruptions could cause greater scarring, pressure on capital buffers and margins, credit tightening, increase of zombie corporates, wave of bankruptcies, and higher NPLs. - De-anchoring of inflation expectations (and scarcity of gas and oil) in adverse scenario: - Commodity price hikes, higher-than-expected above-target inflation, real GDP decline with significant damage to large manufacturing sector in Germany. - Unexpected monetary tightening in the U.S. and euro area leads to tightening financial conditions, rise in risk premia and funding costs, and a global housing bust. - Higher rates and supply constraints cause loss of confidence, drop in demand, recessionary pressure in Germany, and increases in corporate and household NPLs. - Lower investor confidence against some high-debt euro area countries could raise sovereign yields and affect German banks’ and other financial institutions’ holdings in these countries. - FSAP stress test design: - Assessed German banking system resilience against March 15 WEO baseline and a RAM-based adverse scenario (Appendix IA, STeM). - Included three global shocks (global resurgence of the pandemic, de-anchoring of inflation expectations in the U.S., geopolitical risks related to the Ukraine-Russia war), a high-debt countries euro area shock, and a structural shock related to digitalization. - Adverse scenario severity: based on statistical approach (3 standard deviations from the baseline over two years); scenario is V-shaped with a deeper trough. - Adverse scenario outcomes by 2024: - 13.4 percent decline in real GDP by 2024. - House prices decline by 23.35 percent by 2024. - Stock prices decline by 28.6 percent by 2024. - Unemployment increases by 6.6 percentage points above the baseline and remains 0.8 percentage points above the baseline at the end of the projection horizon. - The adverse scenario combines three G-RAM risks; global layers explain about two-thirds of the severity and domestic layers the remaining third. *Source: IMF staff analyses and referenced Bundesbank, KfW, KfW SME panel, vdpResearch, Deutsche Bundesbank, FSC, and BaFin data as presented in the supplied content.* ### 22. The liquidity stress test focuses on banking institutions headquartered in Germany. ### 22. The liquidity stress test focuses on banking institutions headquartered in Germany ### Scope and sample - The liquidity stress test focuses on banking institutions headquartered in Germany. - German subsidiaries of foreign banking groups are included if their legal headquarters are in Germany; local branches of foreign banks are not included. - The liquidity stress tests are based on a randomly collected sample of LSI banks that include banks of various categories and of all 17 SI banks. - For the solvency stress tests: the largest 16 SIs with accounts under IFRS9 and 1293 LSIs with accounts under national GAAP are covered. - The 16 SIs and the 1293 LSIs account for about 87 percent of bank assets. - The five SIs under national GAAP accounting were not included in the solvency stress tests; as a result, 89.5 percent of SIs’ assets are included in the stress test. ### Stress scenarios (liquidity and macro-financial shocks) - Three different stress scenarios were considered to assess liquidity resilience: - Baseline (benchmark): strain on bank liquidity under otherwise “normal” conditions as a result of cyclical fluctuations and other occasional but not altogether exceptional events. - First stress scenario (less likely, short-lived): renewed pandemic wave triggering precautionary measures and restrictions that induce a new decline in economic activity. - Two most severe stress scenarios (more persistent; likelihood increased by recent inflation and geopolitical developments): - A new surge in risk aversion toward highly leveraged euro area countries. - Persistent high inflation rates that trigger a monetary policy response. - The two most severe scenarios could stem from the same factors, entail very similar effects, and yield similar results. - Selected macro-financial simulated outcomes under the Adverse Scenario (IMF staff GFM simulations): - Real GDP falls to 13.4 percent below baseline by 2024. - Inflation shoots up 1.4 percentage points above baseline in 2023 and then falls in negative territory thereafter. - The ECB tightens monetary policy in the near term, and loosens in outer years. - Long-term rates rise in Germany, but by less than short-term rates, resulting initially in a flattening of the yield curve. - Simulated House Prices: a global housing bust results in a decline in residential real estate prices of about 23 percent by 2024. - Simulated Unemployment Rate: unemployment rises 6 percentage points above baseline by 2024. ### Solvency stress test design and methodology - FSAP stress tests are top-down with projections generated by in-house models; different in granularity and calibration from the supervisory (EBA-SSM-ECB) bottom-up exercise; results are not directly comparable. - Cut-off dates: - SIs: end-2021 (balance sheets and P&L). - LSIs: end-2021:Q3 (balance sheets and P&L). - Data sources: - SIs: ECB-SSM (FinRep and CoRep templates, and STE files on IRRBB). - LSIs: Bundesbank (FinRep and CoRep templates). - Default rates and loss rates for loans to NFC: Bundesbank credit registry, aggregated at the economic sector level. - Consolidation and exposures: - For SIs, stress tests follow a consolidated balance sheet approach and consider foreign exposures to France, Italy, Spain, the UK and the US, in addition to domestic exposures. Stressed foreign exposures include non-financial corporates, households, financial institutions and sovereign exposures. - For LSIs, only domestic exposures are considered and a domestic consolidated approach is used. - Accounting and regulatory metrics: - SIs: IFRS9 accounting and regulatory capital ratios; performance assessed on Common Tier one (CET1) capital ratio. - Hurdle rate for CET1: 8.25 percent, comprising: - minimum CET1 of 4.5 percent, - capital conservation buffer of 2.5 percent, - countercyclical capital buffer of 0.75 percent starting 2023, - systemic risk buffer of 0.5 percent. - The choice of a common hurdle rate excludes G-SII buffer, O-SII buffer and any Pillar 2R and Pillar 2G buffers for confidentiality reasons. - Balance sheet assumptions: - Static balance sheet assumption: no write-offs and new originations equal maturing loans; bank assets and liabilities stay broadly constant. - No managerial action and no portfolio rebalancing. - Share of write-offs and maturing loans at asset class level consistent with recent historical supervisory data. - IFRS9 staging and provisioning: - Exposures enter Stage 1 upon origination; may migrate to Stage 2 or Stage 3 (non-performing). - Stage 1 provisions: 12-month horizon; Stage 2 and Stage 3 provisions: life-time horizon. - Transition matrices modeled with estimated transition rates; where long historical series are absent, transition flows estimated using the beta-linking approach. - PiT PDs and TTC PDs are those reported by each SI in supervisory templates. ### Auxiliary and satellite models used - The solvency balance sheet model relies on satellite models for: - Credit risk (bank-level, four sectors: non-financial corporates, households, financial institutions, sovereign). - Interest rates and income components. - Accounting model: - Generates dynamics of IFRS9 transition matrices and calculates provisions under expected credit loss approach. - Produces paths for PDs, lending rates, deposit rates, pre-provisioning income, and sovereign spreads relative to the repo rate for baseline and adverse scenarios. - RWAs result from shocked risk parameters (PDs, LGDs) and simulated provisions; risk densities for other RWAs are assumed constant. - Income and expense modeling: - Net interest income computed from simulated lending and deposit rates, sovereign yields (bank-specific), and projected evolution of stocks of loans, deposits and sovereign debt holdings. - Net fees and commissions, non-interest rate expenses and other income are assumed to grow in line with total assets. - In the adverse scenario, net fees and commission are assumed to drop by 10 percent in 2022, stay constant in 2023, and grow again during 2024-2026. - Lending and deposit rate modeling: - Interest rate levels are bank-specific based on end-2020 starting values; changes in interest rates are common to all banks and determined by regression-predicted changes. ### Credit risk modeling specifics - Non-Financial Corporates (NFC): - Satellite models use bank-level panel regressions of default rates and loss rates to generate quarterly PDs (averaged to annual PDs) and scenario-dependent LGDs. - Estimation period: 2008:Q1 – 2019:Q4 (includes Global Financial Crisis; excludes pandemic period). - Preferred specification: real GDP growth as macroeconomic control variable showed best performance. - Key quantitative implication from pooled panel regressions (column 1 of Table 13.A): - A 3 standard deviation shock to real GDP growth results in a 0.25 percentage points average increase in NFC annualized default rates, starting from a 0.53 percent annualized average default rate for the banking system. - This implies a 47 percent increase in the NFC default rate and a 12 percent increase in the loss rate by 2024. - Sectoral differences: manufacturing and retail and wholesale trade more sensitive to macroeconomic cycle than other sectors. - For LSIs, average coefficient on real GDP for Savings banks and credit cooperatives used; for SIs, average coefficient for big banks and Landesbanken used. - Households: - Credit risk estimated from a micro-simulation model based on the ECB Household and Finance Consumption Survey (2017) and the US PSID. - Default logic: households draw down financial assets to repay debt; default occurs if financial assets are fully depleted and other income insufficient; default possible only if household becomes unemployed. - Simulations incorporate unemployment replacement rates and likelihood of return to employment. - For SIs, PD and LGD evolution benchmarked on starting PD and LGD reported by each bank in supervisory templates; for LSIs, benchmarked on 2021:Q3 aggregate PD and LGD published by the EBA on the risk dashboard. - Note on data vintage: reliance on 2017 HFCS may underestimate risks if more recent mortgage vintages are riskier. *Source: IMF staff (Germany FSAP chapter).* ### Box 3. Germany: Micro-Simulation Model (Concluded) ### Box 3. Germany: Micro-Simulation Model (Concluded) ### Model mechanics and household default conditions - A household defaults if it remains unemployed and its gross financial assets are entirely depleted at date t (condition expressed as 퐹퐹..._퐴퐴푠푠푠푠푈푈푈푈푠푠 푖푖,푡푡 < 0). - Financial assets dynamics when employed (equation (6)): - 퐴퐴푖,푡 = (1 + 푃푃) × 퐴퐴푖,푡−1 + 푃푃푈푈ℎ푈푈푃푃...푖,푡 + 푊푊푃푃푊푊 푈푢 − 푐푐푈푈푃푃푈푈_푠푠푈푈푃푃_푠푠푐푐푐푐푈푈 − 푆푆푈푈푈푈푈푈푐푐푐푐푈푈푊푊_퐺퐺푆푆. - Financial assets dynamics when unemployed (equation (7)): - 퐴퐴푖,푡 = (1 + 푃푃) × 퐴퐴푖,푡−1 + 푃푃푈푈ℎ푈푈푃푃...푖,푡 + 푈푈푈푈 − 푐푐푈푈푃푃푈푈_푠푠푈푈푃푃_푠푠푐푐푐푐푈푈 − 푆푆푈푈푈푈푈푈푐푐푐푐푈푈푊푊_퐺퐺푆푆. - Unemployment benefits 푈푈푈푈푖,푡 given by (equation (8)): - 푈푈푈푈푖,푡 = 푁푁푈푈푈푈_푃푃푈푈푈푈푈푈푃푃푐푐...푡,푟 × 푊푊푃푃푊푊 푈푢 - 푁푁푈푈푈푈_...푡,푟 is the net replacement rate at date 푈푈, 푃푃 quarters after becoming unemployed. - Spending 푆푆푈푈푈푈푈푈푐푐푐푐푈푈푊푊_퐺퐺푆푆푖 truncated at the 10th percentile (lower bound) and the median (upper bound) of the survey distribution. ### Loss given default (LGD) and property valuation - LGD at time of default given by 푀푀푃푃푥푥{0, LTV − 1}, where LTV = remaining stock of debt at default / value of property at default. - Property value at default = initial survey value × growth rate of real estate prices assumed in the macroeconomic scenario. - Replacement rates and likelihood of exit from unemployment 푃푃 quarters after becoming unemployed are estimated from OECD macroeconomic data. ### Default risk projections (NFCs, households, aggregate) - In the adverse scenario, default risk doubles by 2024 in the banking system. - Baseline WEO projections: - Default risk for NFCs moderately increases. - Default rate for households remains broadly stable initially and then declines slightly by the end of the projection horizon. - Adverse scenario projections: - Default risk for NFCs increases by 46-47 percent in 2023 and 2024. - Default risk for households increases and reaches about 138 percent by 2024. - Based on loan shares to NFCs and households for the banking system as a whole, overall default risk for loans to the non-financial private sector increases by 106 percent by 2024. ### Sovereign and corporate bond PDs - PDs for sovereign and corporate bond exposures estimated from projections of sovereign bond yields and short-term rates. - A Merton-based transformation converts the spread between 10-year sovereign yields and the repo rate (푆푆푡) into a PD proxy: - Implied risk neutral PD: 푃푃푃푃푡 = 1 − 푈푈 − 푆푆푡(푇−푡)퐿퐿퐺퐺푃푃푡. - Residual maturity of banks’ fixed-income securities exposures from macro data (Haver). - LGD assumption for bond exposures: 45 percent. ### IFRS9 modeling, transition matrices, PDs and LGDs - LGDs and PDs informed by banks’ supervisory data (PiT IFRS9 transition rates, PiT PDs, TTC PDs, LGDs for IRB banks). - Adverse scenario uses TTC LGDs and PDs provided by each bank; exposures to households shocked according to RRE price projections. - For LSIs, starting LGDs and PDs from the 2021:Q3 EBA Risk Dashboard. - Provisions under IFRS9 computed based on expected lifetime loss of new net flows into Stage 2 and Stage 3 buckets; lifetime horizon M truncated at 5 years maximum. - Residual probability of default 푃푃푃푃∗ defined as conditional default probability (formula preserved in source). - Transition matrices linked to projected PDs via beta-linking approach; elasticities 훽 used are 0.5 and −0.5 (same as France and Canada FSAP). - Example: change in transition from Stage 1 to Stage 2: ∆푇푇푇푇12 = 훽12 × ∆푃푃푃푃. - Bank-by-bank starting annual transition matrices constructed from FINREP supervisory templates for 2021. ### Provisions (LSIs and SIs) - LSIs under national GAAP: - Loan loss provisions 푃푃푃푃푃푃... = 푃푃푃푃... × 퐿퐿퐺퐺푃푃 × 퐸퐸푥푥푈푈푃푃... - SIs under IFRS9: - Flows of provisions determined by annual expected defaulted Stage 1 exposures and lifetime expected losses for Stage 2 and Stage 3 exposures (see Box 4). ### Provisioning under IFRS9 (Box 4) - Stock of provisions = expected credit losses for S1, S2 and S3 exposures: - 푃푃푃푃푃푃_푡 = 푃푃푃푃푃푃_푆1,푡 + 푃푃푃푃푃푃_푆2,푡 + 푃푃푃푃푃푃_푆3,푡. - S1 stock = expected losses during the year (푇푇푇푇1−3 × 퐿퐿퐺퐺푃푃 × S1). - S2 stock = lifetime expected credit losses summed over u = t+1 to t+M (with discounting by (1 + 푃푃)^(u−t)). - S3 stock = non-recoverable part of defaulted exposures = 퐿퐿퐺퐺푃푃t × S3t. - M is the horizon for lifetime expected credit losses; 푃푃 is discount rate. ### RWAs, IRB/STA treatment and scaling - RWAs projections: - For IRB exposures, credit risk evolves with EAD, PD, and LGD; updated weighted average through-the-cycle (TTC) PDs used with smoothing parameter 푥푥 = 1/10: - ∆TTC_PD = 푥푥 ∙ ∆푃푃푐푐푇푇_푃푃푃푃. - For STA exposures, deterioration reflected in higher specific and collective allowances and higher capital requirements from downgrades. - For SIs, Basel III adjustments: - Apply scaling factor 1.06 to credit RWAs. - Use 1.25 multiplier to the correlation parameter for exposures to large regulated and unregulated financial institutions. - Differences in granularity considered by applying original scaling factor = ratio of model-calculated RWAs to reported RWAs at time t0. - IRB RWAs dynamics for bank c: - 푇푇푊푊퐴퐴...푖,푡 = 푇푇푊푊퐴퐴...푖,푡−1 × (1 + 푊푊) × (1 + 퐸퐸푈푈), where 푊푊 = asset growth rate, 퐸퐸푈푈 = rate of increase of unexpected losses. - STA RWAs dynamics for LSIs: - 푇푇푊푊퐴퐴...푖,푡 = [푇푇푊푊퐴퐴...푖,푡−1 − 푈푈푈푈 푛푛_푈푈푃푃푃푃...] × (1 + 푊푊) + 푈푈푈푈 푛푛_푁푁푃푃퐿퐿..., where terms involve sums of LGD × PD × NU... (asset-class specific). ### Interest rate risk in the banking book (IRRBB) and net interest income (NII) satellite model - Conservative assumption: shocks to policy rates/short-term market rates are fully passed through to banks’ funding costs; pass-through from deposit rates to lending rates is smaller than one. - Empirical fixed-effect panel regressions (Fitch Connect, 2006-2019) model NII: - NIIi,t = α × NIIi,t−1 + β ∙ Macro t + Fi + εi,t. - Regression findings: - NII positively associated with slope of yield curve (spread between short-term repo rate and long-term bond yield). - Coefficients and statistics (LSIs | SIs): - Lagged dependent variable: 0.509*** | 0.409*. - Spread (long-term − short-term): 0.185*** | 0.104*. - Constant: 0.836*** | 0.806***. - Observations: 18,380 | 256. - R-squared: 0.3150 | 0.013. - Number of banks: 1,284 | 20. - Through to NII (measured by α/(1−β)) larger for LSIs (0.37) than for SIs (0.17). - Complete pass-through from deposit rates to lending rates assumed to occur in 5 years. - From regression mapping: - (rL − rD) ≈ NII. - (rL,t − rD,t) ≈ α × (PL,t−1 − PD,t−1) + β × sUPPc_t. - Pass-through change: ∆PL,t,t−1 = ∆PD,t,t−1 + β(1 − α) × ∆sUPPc_t,t−1 /. ### Timing, maturity transformation and data limitations - Repricing of assets is slower than liabilities; maturity transformation gaps can exacerbate short-term margin compression. - For SIs: - Scenario funding and lending rate projections mapped to banks’ assets/liabilities by product and counterparty using the short-term exercise (STE) IRRBB template of the ECB. - Template includes maturity ladders for fixed-rate instruments and repricing dates for floating-rate instruments. - For LSIs: - In absence of bank-level maturity ladder data, aggregate maturity structure of retail and wholesale deposits from Haver used to estimate banks’ liabilities and applied to assets, neutralizing maturity transformation impact but still allowing assessment of funding-shock pass-through effects. ### Market risk assessment for fixed-income securities - Assessment based on modified duration approach; fair value changes due to policy rates and credit spreads. - Analysis covers domestic sovereign, domestic corporate bonds, domestic financial institutions bonds; for SIs exposures to 5 sovereigns also included. - Repricing risk formula (preserved form in source) links change in fair value to change in risk-free rate and remaining duration. - Credit spread risk formula (preserved form in source) links change in fair value to change in credit spread and remaining duration. - For SIs under IFRS9: - Only FVOCI and FVPL bonds included in market risk scenario; AC bonds treated under PD/LGD expected loss approach in sovereign asset class. - Remaining duration approximated by aggregate Haver data applied to each bank. - For LSIs under national GAAP: - HFT bonds included in market risk analysis; when breakdown unavailable, assumed all exposures were HFT. - Remaining duration approximated by aggregate Haver data. ### Accounting and interest accrual assumptions - Non-accrual exposures (e.g., non-performing or S3 exposures) assumed not to earn interest income; net income “before stress” can decline due to accumulation of non-accrual exposures. - For SIs under IFRS9, only bonds classified as FVOCI and FVPL considered for market risk; AC bonds treated under PD/LGD. ### Solvency stress test approach - Economic approach: credit risk estimated on total exposure of each bank. - As customary in FSAPs, PDs, LGDs, provisions, RWAs and capital dynamics projected bank-by-bank using supervisory and macro data. - Loss given default assumption for bond exposures: 45 percent (used in FSAP practice). *Source: German Credit Registry, WEO and IMF staff estimates* ### 44. The simulations performed under the baseline show that the banking system is well ### 1deuea2022009 - 44. The simulations performed under the baseline show that the banking system is well ### Baseline solvency simulations (Significant Institutions) - Aggregate capitalization remains high, above 13.6 percent for the CET1 ratio. - A moderate decline occurs in 2022 due to interest and market risk, and some losses on FX exposures. - The increase in capital is driven by net income despite some moderate near-term losses caused by the interest rate increase (which leads to interest rate risk and repricing risks). - Credit risk parameters (aggregate PDs and aggregate LGDs) decline over time, contributing to net income before provision and to lower provisions. - Banks continue to pay dividends of 25 percent if net profit is positive. ### Adverse scenario solvency (Significant Institutions) - Aggregate depletion of CET1 capital reaches 5.2 percent of RWAs by the end of 2023 from 14 percent and increases in subsequent years (under the assumption that banks do not write-off non-performing exposures). - Capital depletion drivers: high funding shocks combined with maturity/repricing gap between assets and liabilities, interest rate risk, market risk, higher RWAs and credit risks resulting from the sharp recession, and a sharp drop of fee and commission income. - Three banks fall below the hurdle rate in 2022-23, but capitalization remains above the minimum CET1 ratio. - Aggregate capital shortfall remains moderate at around euros 20 bn (no write-offs). - Under the assumption that banks can write-offs non-performing exposures: - CET1 capital depletion is 5.0 percentage points of RWAs by 2023. - Capital shortfall is euros 18.7 bn (0.52 percent of GDP). - If the counter-cyclical capital buffer (CCyB) could be released under the adverse scenario: - Capital shortfall would become euros 14bn (0.39 percent of GDP) under no write-offs. - Capital shortfall would be euros 200 mil (0.01 percent of GDP) if banks write-offs non-performing exposures. ### Less Significant Institutions (LSIs) solvency - Under the baseline, aggregate CET1 capitalization remains strong and continues to rise to 17.7 percent of RWAs. - Under the baseline, 8-9 banks fail the hurdle rate depending on whether the CCyB is included or not in the hurdle rate; these banks account for 1.55-1.7 percent of total LSI assets and are weakly capitalized in 2021. - Under the adverse scenario, aggregate CET1 capitalization increases very moderately to about 16.3 percent of RWAs. - Under the adverse scenario, 18-21 banks fail the hurdle rate depending on whether the CCyB is included; these banks account for 2.83-3.04 percent of total LSI assets. - Recapitalization needs under the baseline reach 0.016-0.018 percent of GDP depending on CCyB inclusion. - Recapitalization needs under the adverse scenario reach 0.047-0.051 percent of GDP depending on CCyB inclusion. - Under the baseline, net income remains positive as a result of strong pre-impairment income and relatively limited stress. - Under the adverse scenario, higher provisions and RWAs due to impaired exposures, market risk and compression of interest margins drive negative contributions to capital accumulation. - Pre-impairment income is affected by a decline in accrual loans (non-accrual loans do not earn interest income) and shocks to fee and commission income in 2022. ### Liquidity stress test — overview and objectives - Liquidity stress tests analyze potential flows and effective liquidity buffers under stress to determine whether a bank’s liquidity buffers would enable it to continue normal activities (“going concern hypothesis”) without resorting to extraordinary liquidity support (ordinary central bank support is admitted). - Key conceptual challenge: balancing higher-yield/low-risk investments with sufficient idle/liquid assets to honor unexpected surges in cash demand; buffers may be inadequate in systemic or idiosyncratic tail events. ### Funding profiles and liquidity conditions (banking sector) - Commercial banks derive half of their funds from wholesale sources; real estate banks rely on Pfandbriefe (covered bonds). - Savings banks and cooperative banks are funded mainly by customer deposits. - Landesbanken collect few deposits and rely heavily on wholesale and interbank funds. - Other banks are heterogeneous but almost all their funding comes from wholesale and interbank sources. - The German banking sector displays a risk-averse approach to liquidity, maintaining ample safety buffers and potentially foregoing profit opportunities. - Liquidity indicators as of September 2021: - Average liquid assets-to-total assets (LATA) ratio: 28.3 percent. - LCR: 163 percent. - NSFR: 125 percent. - The banking system significantly increased liquidity buffers since the 2016 FSAP; German banks entered the Covid pandemic with large excess liquidity. - At end-2019: - Commercial banks LATA: 30.6 percent. - Cooperative and savings banks LATA: around 8 percent (noting LATA underestimates actual liquidity for these banks since interbank deposits are excluded from the numerator). - Average LCR for the sector: 158 percent (regulatory minimum of 100 percent). - Between December 2019 and December 2021, assets of the German banking system increased by 10 percent (€880 billion), equivalent to 24 percent of GDP. - Two-fifths of this increase went into savings and cooperative banks. - About one-third went into non-systemic commercial banks. - Increase funded mainly with interbank credit, which increased by €560 billion (one-half of which went into commercial banks). - Customer deposits increased by €365 billion and funded most of the remaining increase. - Remark: Excluding interbank lending, the assets of the banking sector increased by 316 billion, equivalent to 4.5 percent of the initial assets net of interbank credit. ### Pfandbriefe (Special Covered Bonds) — structural features and liquidity buffers - Pfandbrief issuance and cover pools: - Pfandbriefbanken: about 80 Pfandbriefbanken and about 120 Pfandbrief programs. - Cover pools typically include mortgage loans (valued at 60% of their Mortgage Lending Value, MLV), ship and aircraft finance loans (valued at 60% of their MLV), and public sector bonds (valued at 100%). - Cover pool may include up to 20% of Pfandbrief circulation in high quality liquid assets. - Cover pool must include a liquidity buffer equivalent to contractual outflows of outstanding covered bonds maturing in the following 180 days net of contractual inflows from cover assets falling due during that period. - Cover pools are audited every three years. - Recourse and protections: - Creditors have double recourse: general payment obligation of the issuer and, in issuer insolvency, claim on cover pool assets (with prior claim on other bank creditors); in case of shortfall, Pfandbrief creditors participate pro-rata in the issuer insolvency estate. - Mortgage Lending Value (MLV) features: - MLV reflects the “long-term, sustainable value of the property being mortgaged,” excluding cyclical fluctuations and nonsustainable speculative increases. - MLV assessed conservatively using a double methodology; if two methods differ by more than 20% the bank must explain. - Market value increases above MLV provide additional buffer but MLV cannot be revised upward except in very limited circumstances. - Risk mitigation and issuance practice: - Bonds must be issued to limit liquidity, sectoral, and regional concentration (e.g., distributing payment outflows over time to prevent spikes). - There is a trend to lengthen average maturity of outstanding stock. - About one-third of the aggregate volume of Pfandbrief cover pool consists of public sector debt; the rest are mortgages. - Bonds issued against the pool cannot exceed 60% of the MLV of the mortgages; issuers often provide voluntary over-collateralization to obtain favorable external ratings. - The required 180-day liquidity buffer and structural features make Pfandbriefe comparatively low solvency and liquidity risk instruments; market appetite has historically been resilient even in crisis conditions. *Source: IMF staff estimates.* ### 55. Aside for some temporary strains in March and April 2020, German banks, overall, did ### 1deuea2022009 - 55. ### Liquidity performance of German banks during the pandemic - Aside for some temporary strains in March and April 2020, German banks, overall, did not experience significant liquidity strains during the pandemic. - Since January 2020 liquidity indicators altogether have held up well or even marginally improved as banks sought to strengthen their precautionary liquidity buffers. - The average LATA ratio of commercial banks and Landesbanken has increased. - Only “other” banks experienced a marked decline in their LCR, which still remains on average comfortably above the 100 percent requirement. - The average loan-to-deposit ratio of German banks, which hovers around 80 percent, signals ample liquidity. - Figure 13: Germany: Selected Liquidity Indicators of the Banking Sector, 2020-21 (source: German Supervisory Authorities; and Fund staff elaboration). ### Liquidity strains in Spring 2020 and policy response - Some strains on liquidity emerged in March and April 2020 but they eased with public intervention. - In March and April some (mostly large commercial) banks experienced liquidity outflows as nonfinancial corporations drew on their committed credit lines; these demands eased when the government started to provide solvency and liquidity support to enterprises, and most of the liquidity thus mobilized was returned, often to the same bank, in the form of deposits. - Net outflows were dampened as nonfinancial corporations used drawn credit lines to build up precautionary liquidity buffers. - Market strains: securities issued by sectors most directly hit by the pandemic (automobiles, tourism, and energy) started to trade at high-risk premia; bid-ask spreads for bonds issued by European high-yield enterprises widened, resulting in higher margin calls and raising demand for liquid collateral, eroding banks’ effective liquidity buffers. - Banks initially responded by mobilizing lower quality collateral (e.g., pledging mortgages with the ECB via covered bonds or securitization). - Market conditions rapidly recovered after ECB actions: long-term financing operations (TLTRO-III and PELTRO), eased collateral eligibility constraints, and expanded asset purchase programs (APP and PEPP). - Footnotes and specifics: - The LCR ratios effectively declined in several banks in March and April due to this operation reducing some liquidity indicators as higher rolloff rates are applied to deposits than to undrawn committed credit lines. - Changes in the TLTRO included an increase in the maximum amount that could be borrowed. In the fourth TLTRO operation in June 2020 banks took about €1.3 trillion. - The Pandemic Emergency Long Term Operation (PELTRO) was launched in May 2020. - The ECB eased LCR requirements, allowing banks to reduce their ratios below 100 percent under stress and temporarily excluding holdings of central bank reserves from the LCR ratio. ### Box 6 — Liquidity strains in European financial markets, Mar–Jun 2020 (summary) - Timeline and drivers: - Episode most acute mid-March and eased gradually when the ECB intervened. - Strains started at end of February in the U.S. Treasury market and rapidly extended to other markets. - A global flight to safety triggered declines in share and bond prices, surging volatility and transaction volumes, and evaporating liquidity especially in derivatives markets such as futures. - Massive redemptions from investment and money market funds forced asset liquidations, aggravating downward asset-value spirals. - Most severely strained markets: - Commercial paper and certificates of deposit. - Corporate bond market, hampering corporations’ capacity to raise new funds. - Unsecured term money market experienced tensions (rates on commercial paper used as benchmark). - Market functioning and dealer capacity: - Many market dealers had limited balance sheet space to provide liquidity; operational difficulties (remote work) also contributed. - Conditions became most acute between March 16 and March 23 with sharp drops in asset prices, widening bid-ask spreads, and decline in market depth. - ECB and other central bank interventions: - Major central banks injected liquidity through asset purchases, liquidity facilities, currency swap arrangements, and temporarily eased some regulatory requirements. - Eurosystem intervention (TLTRO, PELTRO, APP, PEPP) restored market liquidity and contained volatility, bringing down rates though not fully restoring market depth. - After ECB announcements (including the €750 billion PEPP on May 18) benchmark liquidity indicators like bid-ask spreads narrowed significantly but did not return to pre-pandemic levels; total volume of transactions in these bonds did not recover. - Quantitative market signals: - Repo Funds Rates (RPRs) on German and French bonds fell by about 15 bps to -0.65 percent between March 12 and March 18, signaling widening of the “specialness premium”. - The premium declined and rates bounced back a few days after the PEPP announcement. - Transmission to banks: - Corporates mobilized outstanding credit lines, transmitting liquidity strain to banks’ balance sheets; much of this demand was precautionary and remained on banks’ balance sheets as corporate deposits. - Leveraged investors’ higher margin calls tapped bank deposit balances and committed credit lines; in some cases banks had to post margins themselves. - Large commercial banks acting as market dealers were constrained by regulatory capital in performing market-stabilizing roles. - Outcome: - Eurosystem measures succeeded in restoring market liquidity and containing volatility though not equally successful at restoring market depth. - Some observers described developments as a “dash for collateral” rather than a “dash for cash” as investors fled to German and French collateral until PEPP eased concerns. ### Cash-flow analysis set-up and sample - Liquidity stress tests performed by the IMF FSAP team are cash-flow based stress tests run on confidential supervisory data provided by the authorities. - Key features of tests: - (i) A one-year time horizon from a common reference date. (This period was divided in 10 ”maturity buckets.”) - (ii) Application of “behavioral” assumptions (roll-off rates) on contractual cash flow entitlements and obligations across asset, liability, and off-balance sheet categories. - (iii) Three different stress scenarios formalized via rolloff and haircut parameters entailing exceptional but plausible demands on bank liquid resources. - Coverage: - Stress tests conducted on all SI banks and on a limited, randomly extracted representative sample of LSIs. - All 17 SI banks supervised by the ECB under the SSM were included (4 holding companies supervised by the ECB but not considered credit institutions were not included); these entities account for about 45 percent of the assets of the German banking sector. - Out of more than 1,300 LSIs, 40 banks were selected randomly from the authorities’ list; sample includes commercial banks, savings banks, cooperative banks, and building societies and real estate banks. - The total assets of the banks included in the 40-bank sample represent about 3 percent of the total assets of the German LSI sector. - Stress tests on dollar-denominated liquidity were run on all 17 SIs and on all LSIs which report dollar-denominated cash flow data (about 40), mostly commercial banks. - Sampling design: - The LSI sample was chosen to balance minimum representation (minimum of five banks per category) and representativeness of bank size within each category, reflecting shares in each quintile of the asset distribution. - Table 4 composition (Total 40 banks; in parentheses total banks in sector by quintile): Total 10 (114) commercial banks, 10 (368) savings banks, 15 (805) cooperative banks, 5 (20) building and mortgage banks = 40 (1,307). ### Cash-flow analysis scenarios and mechanics - Purpose: assess whether banks have sufficient liquidity buffers to withstand net liquidity outflows under unfavorable conditions while maintaining operational capacity without exceptional support. - Method: - Behavioral assumptions (rolloff rates) applied to contractual cash flows by maturity bucket and category; rolloff rates represent share of contractual flow that would produce a net liquidity flow. - Haircuts applied to Counterbalancing Capacity (CBC) assets to reflect adverse market pricing or collateral valuation changes. - Analysis under a “going concern” hypothesis to assess continued business operations without emergency liquidity assistance. - Four scenarios considered (besides a baseline): - Baseline: “normal” cyclic conditions used as benchmark. - Scenario A — “new wave”: - Represents a new unexpected wave of the pandemic with intensified precautionary measures and decline in economic activity. - Dominated by retail funding stress. - Corporate sector intensifies demand for loans (reducing rollover rates on loans and increasing rollover rates on unused committed credit lines) and reduces funding of banks (nonoperational and to lesser extent operational deposits). - Banks experience net outflows of retail deposits as enterprises and households run down liquid savings. - Haircut rates increase somewhat compared to baseline. - Scenario is of mild severity and materializes at shorter horizon. - Scenario B — “risk aversion”: - Increase in risk aversion triggers outflow of funds from some Euro-area sovereign borrowers to which German banks are exposed. - Dominated by market liquidity stress: assets used for liquidity become illiquid, prices collapse, haircut rates increase, reducing banks’ CBC. - Some bank clients exposed to these assets withdraw deposits and banks may be unable to roll off some loans. - Scenario is of high severity. - Scenario C — “persistent inflation”: - Upward shift in inflation expectations leads to continued increases in prices. - Comparable severity to Scenario B but more persistent. - Combined increase in risk aversion with sustained inflation leads to stronger outflows of nonoperational deposits peaking at a one year-horizon and implies stronger haircuts to CBC but weaker outflows from downgrades. - Scenario design pre-dated the war in Ukraine but captures potential repercussions of a protracted war in Ukraine. - Scenario parameter table (Table 5) — selected roll-off rates and haircuts (values preserved exactly as shown): - Outflows (examples): - Securities issued: Scenario A 20-30% | Scenario B 20-75% | Scenario C 25-100% | Baseline 20-30% | Range in comparable FSAPs [not used here] 20-30% - Stable deposits: Scenario A 5-10% | Scenario B 3% | Scenario C 3-5% | Baseline 2% | Range 0-10% - Operational deposits: Scenario A 5-20% | Scenario B 5-20% | Scenario C 5-10% | Baseline 5% | Range 0-50% - Nonoperational deposits: Scenario A 20-30% | Scenario B 20-40% | Scenario C 30-50% | Baseline 10% | Range 15-100% - FX swaps (outflows): Scenario A 75% | Scenario B 50% | Scenario C 100% | Baseline 20% | Range 50-100% - Other derivatives (outflows): Scenario A 50% | Scenario B 25% | Scenario C 75% | Baseline 20% | Range 50-100% - Other outflows: Scenario A 50-100% | Scenario B 20-50% | Scenario C 50-100% | Baseline 20% | Range 50-100% - Inflows (examples): - Loans to retail: Scenario A 5-10% | Scenario B 10-20% | Scenario C 5-20% | Baseline 0% | Range 0-60% - Loans to NFCs: Scenario A 5-10% | Scenario B 10-15% | Scenario C 10-20% | Baseline 0% | Range 0-60% - Loans to OFIs: Scenario A 30-50% | Scenario B 35-50% | Scenario C 25-75% | Baseline 10% | Range 30-100% - FX swaps (inflows): Scenario A 75% | Scenario B 50% | Scenario C 100% | Baseline 20% | Range 25-100% - Other derivatives (inflows): Scenario A 40% | Scenario B 20% | Scenario C 50% | Baseline 20% | Range 50-100% - Maturing portfolio: Scenario A 50% | Scenario B 50% | Scenario C 40% | Baseline 20% | Range 50-100% - Other inflows: Scenario A 15% | Scenario B 10% | Scenario C 20% | Baseline 10% | Range 20-100% - Counterbalancing capacity (haircuts and usable fractions): - Level 1: Scenario A 5% | Scenario B 5% | Scenario C 10% | Baseline 2-5% | Range 5-30% - Level 2a: Scenario A 10% | Scenario B 10-25% | Scenario C 30% | Baseline 5-10% | Range 11-30% - Level 2b: Scenario A 10-20% | Scenario B 20-40% | Scenario C 30-50% | Baseline 5-10% | Range 20-40% - Other tradable assets: Scenario A 20-50% | Scenario B 25-65% | Scenario C 75-100% | Baseline 10-25% | Range 27-70% - Nontradable assets: Scenario A 40% | Scenario B 50% | Scenario C 90% | Baseline 25% | Range 75% - Commitments and contingencies: - Inflows from undrawn committed facilities: Scenario A 50% | Scenario B 40% | Scenario C 30% | Baseline 20-50% | Range 0-50% - Outflows from committed facilities: Scenario A 10-15% | Scenario B 5-7% | Scenario C 8-10% | Baseline 20% | Range 0-55% - Outflows from downgrades: Scenario A 20% | Scenario B 50% | Scenario C 75% | Baseline 0% | Range 20-100% - Sources for scenario assumptions: Fund staff estimates and assumptions. *Source: Excerpt from the IMF FSAP chapter on Germany (pages 55–64).* ### 62. Reflecting confidentiality as well as methodological considerations inherent in the use ### 62. Reflecting confidentiality as well as methodological considerations inherent in the use ### Stress‑test indicators and presentation - Results are presented only in aggregate form. - Analysis focuses on the following indicators for each maturity bucket: - (i) Cumulative net outflows (in percent of assets); - (ii) Remaining counterbalancing capacity (in percent of the stock of CBC outstanding at the reference date); - (iii) Cumulative liquidity shortfall (defined as the additional liquidity that would be needed, at the beginning of the period, to prevent the bank becoming illiquid at that maturity bucket), in percent of assets; - (iv) Number of banks (if any) that become illiquid, completely exhausting their CBC buffer. ### Contractual cash‑flow structure: LSIs versus SIs - LSIs: - Main contractual outflows: retail deposits, followed by NFC deposits, primarily at short maturity. - Main inflows: short-term claims on the central bank and claims on retail and non-financial sector clients. - Contractual outflows amount to about 40, 57 and 68 percent of total assets after 1, 6 and 12 months, respectively. - Corresponding contractual inflows amount to 11, 19 and 26 percent of total assets. - Net contractual funding gap: about 40 percent of total assets at the one-year horizon. - SIs: - Largest contractual outflow category: FX swaps (35 percent of total assets), matched by corresponding inflows. - Outflows from maturing level 1 tradable assets: 12 percent of total assets, concentrated at maturities up to 1 month. - Main contractual inflows: central bank liquid claims (13 percent of total assets), level 1 tradable assets (14 percent), and claims on non-financial corporations (6). - Net contractual funding gap: on average about 14 percent of assets at the 12 months horizon. ### Cash‑flow based stress‑test results for LSIs - Resilience up to 3 months: - Under Scenario A: 4 LSI banks out of 40 become illiquid within three months. - Under Scenario B: 5 LSI banks out of 40 become illiquid within three months. - Under Scenario C: 8 LSI banks out of 40 become illiquid within three months. - At 12 months: - Under Scenario C: 12 banks out of 40 become illiquid. - Liquidity shortfall magnitudes: - Cumulative liquidity shortfall in the worst-case (Scenario C at 12 months) amounts to less than 1 percent of the total assets of the banks included in the sample. - Main drivers of shortfalls (bank-specific): inadequate initial CBC, loss of CBC due to haircuts, significant outflows from non-operational NFCs, other credit institutions, or financial customer deposits. - Heterogeneity: - Affected banks are primarily small and mid-sized LSIs; affected bank types are heterogeneous. ### Counterbalancing capacity (CBC) — composition and coverage - Average holdings: - Banks in the sample hold on average about 20 percent of their assets in liquid form usable as CBC. - Composition (LSI sample): - Cash, reserves at the central bank and level 1 tradable assets account together for about two thirds of the CBC. - Composition (SI sample): - Central bank reserves account for 60 percent of CBC; together with level 1 and 2 assets constitute more than ¾ of CBC buffers. - Credit quality: - Most of the level 1 and level 2 tradable assets are of highest credit quality (CQ1). - In banks with dollar‑denominated CBC, level 1 and level 2 USD assets are almost exclusively of highest credit quality. - Coverage versus contractual gap: - The average CBC amount (about 20 percent of assets) exceeds the net average contractual funding gap (excluding outflows of retail deposits) even up to the one‑year horizon. ### Usage of CBC under scenarios and time horizons - LSIs (average, Scenario C): - Use about one-half of initial CBC to cover net outflows in the first month. - By the end of the year, three‑fourths of initial buffers would be used under Scenario C; about one‑half under Scenario A. - Declines in asset-weighted CBC: - Declines by 5 percentage points in the first week under Scenario A. - Declines by 9-10 percentage points in the first week under Scenarios B and C. - At 12 months widens to 10 percentage points under Scenario A and about 14 points under Scenario C. - Distributional effects: - Under baseline at one-month horizon the interquartile range of remaining CBC is about 12 percent of assets; it narrows to 7 percent in Scenario C. - Interpretation: banks subject to larger outflow risks tend to hold higher CBC buffers. ### Backstops, network arrangements, and limitations - Public and cooperative pillar networks: - “Primary” banks centralize liquidity management at regional (savings banks) or central (credit cooperatives) level, re-depositing funds at Landesbanken and DZ Bank. - Banks can draw on these claims and on additional support from other primary banks in their network in case of stress. - Such extraordinary support is not included in the stress tests, but mitigates idiosyncratic liquidity strain risk. - Limitations: - A more integrated stress test for cooperative and savings/Landesbank networks could not be implemented due to data limitations and confidentiality constraints. ### SI sample stress‑test results - Illiquidity outcomes: - No SI bank would become illiquid before six months under any scenario. - Within one year: - Scenario B: two SI banks would exhaust liquidity buffers. - Scenario C: three SI banks would exhaust liquidity buffers. - Scenario A: none would exhaust liquidity buffers. - Shortfall magnitudes for affected SIs: - Liquidity shortfall would amount to about 0.1 percent of sample assets and 1-2 percent of equity. - For the affected SI banks, it amounts to an average of 1.5 percent of assets (footnote clarification). - Average CBC usage (SIs): - Under Scenario A: use less than one-fourth of initial CBC in 12 months. - Under Scenario B: use more than one-half of initial CBC in 12 months. - Under Scenario C: use about 60 percent of initial CBC in 12 months. - Conclusion: SIs have stronger buffers relative to expected outflows and more limited capacity to mobilize extraordinary support given larger absolute flows. ### US dollar liquidity risk (selected banks) - Scope: - Most SIs and about 40 LSIs report the US dollar as significant currency. - Dollar‑denominated flows: - Originate primarily from FX swaps and, among SIs, from other derivative contracts; level 1 tradable assets and, among LSIs, bank exposures to NFCs also contribute. - Insufficient USD CBC for some banks: - Some banks hold USD CBC insufficient to cover potential net outflows even within a few days under stress; a few small banks hold no USD CBC. - Stress‑test quantified USD shortfalls (using same rollover rates and haircuts as other currencies and abstracting from converting euro CBC): - Within 5 days, under Scenario C: - US dollar liquidity shortfall could reach 0.3 percent of assets for affected SIs. - US dollar liquidity shortfall could reach 1 percent of assets for affected LSIs. - After one month these values would rise to about 1½ percent of assets. - Mitigants not included in tests: - High depth/ liquidity of euro‑dollar currency market enabling rapid conversion of euro CBC to USD (but may fail in severe stress). - Parent‑group support for foreign-owned banks. - ECB–Federal Reserve euro‑dollar swap line arrangements. - These arrangements are extraordinary and not contemplated in the stress tests. ### Supplementary findings on USD LCR and NSFR (Figure 18 highlights) - About ¾ of reporting banks have a USD LCR below 80 percent. - Most SIs and about one half of LSIs have a USD NSFR ratio below 100 percent. - Among LSIs, USD‑denominated CBC does not cover cumulative dollar‑denominated net outflows even from day 1; among SIs, in aggregate, initial USD‑denominated CBC exceeds net cumulative dollar‑denominated outflows over the entire horizon. - Note: Values for the LSI sample as of Sep. 2021 and for the SI sample Dec. 2021. Less than 40 out of about 1,300 LSIs have significant USD transactions that require reporting. Most SIs report USD liquidity flows. There is no formal requirement to maintain 100 percent LCR or NSFR for foreign currencies. ### Empirical links between supervisory ratios (LCR, NSFR) and cash‑flow stress‑test outcomes (Box 7) - Methodology: - Tests include median regressions of pass/fail dummies on LCR and NSFR, OLS regressions of remaining CBC on LCR/NSFR, and median comparisons of passing vs failing banks across horizons and scenarios for 40 LSI banks. - Key empirical findings: - Banks remaining liquid at horizons of 3 months or longer have a median LCR about 20 percentage points lower than other banks in Scenario A and about 40-50 percentage points lower in Scenarios B and C (estimates not significant at conventional levels). - For NSFR, median differences between liquid and illiquid banks are smaller but almost always significant: estimated median difference up to 30 percentage points in Scenario A and about 15 percentage points in Scenarios B and C, across horizons. - Banks with higher LCR or NSFR generally used a lower share of their initial CBC in all three scenarios, especially at longer horizons. Example: under Scenario C, a bank with an NSFR of 160 is estimated to have used after 12 months 5 percent less of its initial CBC than a bank with an NSFR of 120. - Banks with an LCR above the sample median were less likely to become illiquid; the median LCR of banks that became illiquid was significantly lower than the median LCR of other banks. - Relevance of NSFR for passing the test increases with horizon. - Interpretation: strong correlation exists between supervisory LCR and NSFR and cash‑flow based liquidity stress‑test performance. *Source: Extracted content from the provided IMF chapter/section.* ### Box 7. Germany: LSI Cash-flow Based Liquidity Stress Results and LCR and NSFR Ratios ### Box 7. Germany: LSI Cash-flow Based Liquidity Stress Results and LCR and NSFR Ratios ### Conclusions: Liquidity buffers and stress-test outcomes - Weighted average liquidity coverage ratio (LCR) stands at about 160 percent. - Vast majority of banks have LCR and NSFR ratios well in excess of the regulatory minimum of 100 percent. - Average counter balancing capacity (CBC) stands at about 20 percent of assets and is mostly composed of highest quality assets. - Cash-flow based stress-test findings: - Under the least favorable scenario, only eight LSI banks in a randomly selected sample of 40 appear at risk of becoming illiquid; no SI bank would become illiquid before the conventional three-month horizon. - Within 12 months, 12 LSI banks out of 40 and three SI banks out of 17 could become illiquid. - Shortfalls in this extreme case would be manageable: on average less than 1 percent of assets in the LSI sample and less than 0.1 percent of assets among SIs. - Suggested resilience improvements for selected banks: - Lengthen the tenors of deposits from non-financial corporations. - Adjust the composition of their CBC and increase it. - Currency-specific liquidity risk: - A few banks have thin US dollar liquidity buffers and could be exposed to inadequate buffers to cover potential net outflows in a severe stress scenario. - Some banks may rely on central bank swap lines if euro-dollar market access is limited or costly. *Source: IMF staff, Box 7 — Germany: LSI Cash-flow Based Liquidity Stress Results and LCR and NSFR Ratios.* ### Interconnectedness: structure and cross-border exposures - Domestic interbank market: - Germany’s financial system is interconnected; a relatively small number of banks account for a large share of interconnections. - Interbank market appears segmented among Significant Institutions (SIs) and among Less Significant Institutions (LSIs). - Cooperative and savings banks centralize liquidity management at network level; cooperative banks redeposit liquidity as interbank claims on DZ Bank; savings banks engage in similar arrangements with Landesbank. - Deposits benefit from sector-specific insurance schemes (IPS) that reduce likelihood of rapid withdrawals in stress. - Intersectoral linkages: - Significant linkages exist from monetary financial institutions (MFIs) to households and among MFIs and non-bank financial institutions. - Significant institutions, savings banks, and credit cooperatives’ lending account for over 90 percent of the total exposure of the domestic banking sector (the 21 SIs account for close to half of the total exposures). - Household deposits account for the bulk of banks’ liabilities; non-financial corporates are largely funded by banks. - Domestic holdings of domestic government: - MFIs increased their position on domestic government from 8.53 percent of total assets in 2015 to 10.46 percent in 2021 (increase mainly attributable to Eurosystem’s asset purchase programs where Deutsche Bundesbank increased its exposure). - Exposures of German MFIs excluding Deutsche Bundesbank to domestic government decreased from 7.5 percent of their total assets in 2015 to 4.3 percent in 2021. - Cross-border exposures: - German banks’ total foreign exposure (as percent of total assets) is moderate compared to peer countries, accounting for approximately a fifth of total assets. - Top 20 countries of exposure are predominantly in Europe and North America, and to a lesser extent Asia. - German banks’ exposures are higher against U.S. NBFIs (52.5 billion USD) and U.K. banks (82.6 billion USD) relative to peers. - France, Netherlands, and Italy: more than half of their banks’ claims on Germany are for Germany’s non-bank private sector. *Source: IMF staff, Box 7 — Germany: LSI Cash-flow Based Liquidity Stress Results and LCR and NSFR Ratios.* ### Bank contagion analysis: methodology, data, and key findings - Approach: - Network model of contagion based on Espinosa-Vega and Sole (2010) and Covi, Gorpe, and Kok (2021). - Simulations of cascades of bank failures following hypothetical failures of individual banks due to credit or funding shocks occurring through domestic interbank interlinkages. - Each bank in the sample of 1309 banks is assumed to fail one at a time; model quantifies ensuing defaults under various assumptions on Loss Given Default (LGD) and haircuts on fire sales. - Contagion Index aggregates total default costs scaled by CET1 capital of recipient banks. - Data: - Network constructed from German credit registry by the Bundesbank as of 2021:Q3. - Interbank exposure data matched with balance sheet information for each bank. - Balance sheet info: LSIs from supervisory files as of 2021:Q3; SIs from EBA Transparency exercise as of June 2021. - For LSIs, marketable assets approximated as residual between total assets and loans; for SIs, marketable assets are Level 1 Financial Assets and cash balances as published by the EBA. - Key findings: - Primary directions of contagion: - About 51 percent of contagion losses are caused by contagion from SIs to LSIs. - 36.06 percent of contagion losses are caused by contagion from LSIs to SIs. - Only 9.66 percent + 3.37 percent = 13.03 percent of losses are caused by contagion among SIs or among LSIs. - Concentration: - A few large banks account for most contagion risks; most contagion losses are accounted for by the top 10 most contagious banks. - Amplification: - Losses from the 1st round tend to dominate; as shock severity increases, 2nd and other rounds account for a higher share of total losses, especially when a small bank is the trigger. - Amplification effects remain very small when the 10 most contagious banks are the trigger banks. - Impact on capital: - Both SIs and LSIs are impacted by contagion losses as a share of their own capital. - Credit cooperatives, savings banks, and Landesbanken are significantly exposed to contagion risks; a very large share of contagion losses (44.7 percent) goes from Credit Cooperatives to Credit Cooperatives, from Landesbanken to Savings Banks (21.3 percent), and from Savings Banks to Landesbanken (15.2 percent). - Sample sizes (Table 6): - Total lending banks in interbank interconnectedness sample: 1,297; total borrowing banks: 1,277. - Breakdown examples: Credit Cooperatives — 778 lending / 783 borrowing; Savings Banks — 367 lending / 366 borrowing; SIs — 21 lending / 21 borrowing. *Source: IMF staff, Box 7 — Germany: LSI Cash-flow Based Liquidity Stress Results and LCR and NSFR Ratios.* ### Special Topic — Corporate risk analysis: pandemic impact and sensitivity analysis - Methodology: - Tools from Tressel and Ding (2021) applied to analyze impact on the nonfinancial corporate (NFC) sector and policy responses. - Sensitivity analysis uses market analysts’ 12-month forward forecasts of firm sales (IBES dataset) to calibrate sales shock x% industry by industry for 263 listed firms. - Shock parameter computed as percent change in analysts’ sales forecasts between January 2020 (pre-pandemic) and June 2020 (after initial shock), aggregated at industry level using each firm’s 2019 total assets as weight; applied to end-2019 sales of each firm. - Findings on sales shock: - By mid-2020, analysts’ forecasts implied 2020 overall sales for the sample of 263 listed firms would be 17 percent lower than predicted at the start of the year. - Sectoral differentiation (selected examples from Text Table 1): - Agriculture (2 firms): shock to sales 0.1 percent; predicted change in 2020 sales -315 percent (table formatting indicates sector-level numbers). - Air transport (2): 2.2 percent; predicted change in 2020 sales -53 to -62 percent. - Amusement and Recreation (6): 0.1 percent; predicted change in 2020 sales -55 to -61 percent. - Business services (46): 2.5 percent; predicted change in 2020 sales -74 percent. - Communication (10): 7.1 percent; predicted change in 2020 sales -28 percent. - Manufacturing (145): 60.2 percent; predicted change in 2020 sales -19 to -10 percent. - Total (263 listed companies): 100.0 percent; shock to sales -17 percent; predicted change in 2020 sales -8 percent. - Interpretation: - German NFC sector’s pre-pandemic liquidity and solvency buffers, firms’ adjustments, and timely, expansive, and flexible government policy responses contributed to resilience amid the pandemic. - Scenario analysis (not detailed here) indicates baseline recovery should allow most firms to return to profitability, contain debt-at-risk, and rebuild buffers; an adverse scenario with contraction in 2022-24 and rising inflation yields less benign but still broadly favorable solvency and liquidity trends. *Source: IMF staff, Box 7 — Germany: LSI Cash-flow Based Liquidity Stress Results and LCR and NSFR Ratios.* ### 83. A sensitivity analysis proposed in Tressel and Ding (2021) illustrates the potentially ### 1deuea2022009 - 83. A sensitivity analysis proposed in Tressel and Ding (2021) illustrates the potentially large impact of the expected decline in sales on the NFC sector’s ICR, liquidity and equity. ### Sensitivity analysis methodology (Tressel and Ding (2021)) - Impact on individual firms’ EBIT and ICR assessed using: - EBIT_post_shock = EBIT_pre_shock – x% × [Net sales – y% × Costs of goods sold] - EBIT_post_shock: earnings before interest and taxes after the shock - EBIT_pre_shock: earnings before interest and taxes before the shock - x%: share of sales lost as a result of the shock - y%: share of production costs adjusted for each unit of sales lost (reflects firm decisions and policies) - ICR_post_shock = EBIT_post_shock / Interest expenses - Impact on end-of-period cash balance: - Cash balance_post_shock = Initial cash balance + EBIT_post_shock – [z% × STD/TD × Interest payments + (1– STD/TD) × Interest payments] + (z – 100)% × Short-term debt + (Depreciation + Amortization) – CAPEX - Initial cash balance adjusted for working capital commitments - STD/TD: share of short-term debt and long-term debt maturing during the year in total debt - z%: issuance of new debt during the year as a share of maturing short-term and long-term debt - z >100%: short-term debt increases by (z – 100)% - z =100%: maturing debt rolled over - z <100%: new debt issued is smaller than maturing stock - Interest payments: total interest payment on the initial debt of the firm - Stress scenario assumption: firms do not initiate new fixed capital investments and set Depreciation + Amortization = CAPEX - Initial stock of cash defined as: cash and equivalents + short-term investments + receivables – (accrued payables + accrued payrolls and other short-term liabilities) - Four behavioral responses for y (production cost offset of lost sales): - (i) y = 0% - (ii) y = 50% - (iii) y = 75% - (iv) y = 100% - Sensitivity analysis assumption: all short-term debt and long-term debt maturing during the year is rolled over (z =100%) ### Aggregated sensitivity analysis findings (ICR, cash balances, equity) - Pre-shock baseline (end-2019): - 23 percent of listed firms had an ICR below 1; these firms accounted for 4 percent of total debt. - ICR impact under sales shock scenarios: - If firms fully offset decline in sales by cutting production costs (y = 100%): - Share of firms at risk (ICR<1) would have risen from 23 percent pre-pandemic to 60 percent. - Debt-at-risk would have risen to 41 percent of total NFC debt. - If firms unable to cut production costs (y = 0%): - Share of firms at risk (ICR<1) would have risen to 85 percent. - Debt-at-risk would have risen to 83 percent of total NFC debt. - Median ICR: declined from 7.2 in 2019 to 5.7 in 2020 (sample medians reported). - Cash balance impact: - Pre-shock: 26 percent of firms had cash balance<0. - If firms fully offset decline in sales (y = 100%): - 38 percent of firms would have negative cash balances in absence of new borrowing. - If firms unable to cut production costs (y = 0%): - 54 percent of firms would have negative cash balances in absence of new borrowing. - Equity impact: - If firms fully offset decline in sales (y = 100%): - More than 5 percent of firms could have ended up with equity below zero. - With no offsetting cost reductions (y = 0%): - Sales shock would have caused more than 12 percent of listed firms to end up with equity below zero. - Observed 2019–2020 outcome for listed firms with available data: - Decline in sales contained to 8 percent for the 263 listed firms with data for both 2019 and 2020. - Debt-at-risk rose to 16 percent of total debt. - Median ICR in sample declined from 7.2 (294 firms, 2019) to 5.7 (320 firms, 2020). - Firms in highly affected sectors saw the 25th percentile of the ICR turn negative. ### Dynamic scenario-based liquidity and solvency stress test - Approach: - Imposes macroeconomic scenarios on end-2020 firm-level data. - Core methodology: firm-level OLS and Probit panel regressions linking firm indicators to past firm-level structural and cyclical characteristics, industry fixed effects, and macro-financial conditions (covering 2003-2019). - Regressions estimated largely country-by-country to reflect Germany-specific relationships. - Baseline scenario: - Envisages further recovery from the pandemic in 2022-23 and growth averaging about 1.4 percent in the subsequent three years (Figure 23). - In baseline: ICRs improve over time; share of debt in firms with ICR<1 and share of debt in firms with cash<0 fall over time. - Adverse scenario: - Envisages a renewed economic slowdown in 2022 and a significant contraction in 2023-24 amid rising inflation. - ICRs deteriorate; share of debt in firms with ICR<1 starts to rise again in 2023. - Share of debt in firms with cash<0 remains more elevated than in baseline. - Probability of default rises in 2023, albeit from a low base. - Notes on scenarios: - Numbers for 2021–2023 are model-based estimates. - Scenarios incorporate broad macroeconomic and monetary policy interventions implicitly (interest rate cuts, unconventional monetary policies, fiscal measures, social safety net packages) that support activity and contain corporate borrowing costs. - Scenarios do not incorporate the impact of specific liquidity and solvency support policies implemented by the German authorities on individual firms. ### Key assessments and institutional context - Germany’s NFC sector strengths: - Sustained strong export performance; competitiveness and resilience of NFC sector. - Authorities’ strong policymaking capacity during the pandemic expected to help contain financial stability implications under adverse scenarios. - Insolvency and restructuring regime: - Ranked among the most effective and efficient in a 2018 OECD Economics Department Working Paper. - A 2022 IMF Departmental Paper favorably assesses Germany’s crisis preparedness of its insolvency and restructuring regime. ### Special topic: Real estate market (RRE and CRE) - Drivers of demand and supply: - Historically low interest rates and stable income growth supported housing demand. - Immigration, urbanization, work-from-home arrangements, changing preferences toward more space, and growing share of single-person households increased demand. - Supply constraints: construction sector labor shortages and capacity constraints (capacity utilization reported at 82 percent at end-2021), material shortages, limited land availability in larger cities, and time-consuming construction approval processes. - Housing backlog and activity (2009–2020): - Construction sector employment increased by almost 40 percent during 2009-20. - Housing backlog grew by 240 percent (from 232 thousand in 2009 to 779 thousand in 2020). - Dwelling completions: 306 thousand dwellings in 2020. - Building permits issued: 369 thousand in 2020. - Housing backlog estimated at about 779 thousand dwellings at end-2020. - CRE and RRE investment dynamics: - CRE and RRE remained attractive relative to bond market yields. - Net initial yield for CRE in 2021: 2.9 percent in the 7 largest cities and 4.9 percent in 127 towns and cities. - 10-year government bond yield: -0.2 percent in 2021. - Tail risks and price-at-risk assessments (as of 2021Q3 and related horizons): - Probability of negative real price growth one year ahead increased to: - 2.2 percent for RRE (from 0.7 percent at end-2019) - 66 percent for CRE (from 24 percent at end-2019) - Under severe adverse scenarios (5th percentile): - 5 percent chance that RRE could fall by 14 percent in real and cumulative terms over the medium-term. - 5 percent chance that CRE could fall by 30 percent in real and cumulative terms over the medium-term. - Calibration: - RRE adverse scenario: simultaneous 2 standard deviations shock to leverage (change in household debt-to-GDP, interest payments-to-disposable income), affordability measure (house price-to-gdp per capita ratio (misalignment)), and financial conditions index. - CRE adverse scenario: 2 standard deviations shock to employment. - Valuation and overvaluation indicators (end-2021 / 2021Q3 / end-2020 references): - Price-to-rent and price-to-income ratios suggest deviations from long-run averages: - 21 percent deviation for price-to-rent as of end-2021. - 37 percent deviation for price-to-income as of end-2021. - Econometric country-level model (accounting for real interest rates) suggests RRE overvaluation of about 10-15 percent as of 2021Q3. - City-level panel data (127 cities) indicates larger overvaluation estimates at end-2020: - For the largest seven cities, house price overvaluation at end-2020 estimated at about 20-35 percent for Berlin, Hamburg, Stuttgart, Frankfurt, Cologne, and Dusseldorf, and about 50 percent for Munich. - Bundesbank estimates (based on several methodologies) suggest overall RRE overvaluation of 20-35 percent, with overvaluation in cities in the range of 15-40 percent in 2021. *Sources: Datastream, Capital IQ and IMF staff estimates.* ### 96. RRE price overvaluations suggest potential pockets of vulnerabilities in bank ### RRE price overvaluations suggest potential pockets of vulnerabilities in bank exposures to real estate ### Main findings - RRE price overvaluations point to potential pockets of vulnerabilities in bank exposures to real estate. - Most loans for house purchases in Germany are granted by banks; about five percent of outstanding mortgage loans is held by insurance companies. - On aggregate, CRE and RRE risks on banks’ balance sheets have not materialized through 2021Q3; the Bundesbank’s Bank Lending Survey does not point to any reported relaxation of lending standards at end-2021. - Authorities could underestimate risks due to a lack of comprehensive borrower-based information (e.g., LTV, DSTI). - Immediate risks may stem from sub-sectors—e.g., non-food retail and potentially hotel businesses. - Potential risks include COVID outlook uncertainties and their impact on economic growth and rising interest rates in conjunction with higher inflationary pressures. ### Quantitative indicators and exposures - Outstanding bank loans for house purchases at 2021Q3: exceeded EUR 1.6 trillion (46 percent of GDP and about half of total loans to domestic enterprises and households). - Growth of housing loans to domestic enterprises and resident individuals: 7.1 percent year-on-year in 2021Q3 (the highest growth in about two decades). - Banks’ exposures to CRE sectors (construction, housing corporations, and other real estate activities): about EUR 602 billion (6 percent of assets and 19 percent of total loans to enterprises and households). - Lending to CRE sectors growth in 2021Q3: 6.4 percent y/y, after peaking at 8.3 percent y/y in early 2019. - Exposure of Germany’s life insurers to real estate (direct and indirect investments): about 3 percent of assets. - Data collection improvement: Current data gaps on lending standards regarding housing loans are expected to be closed through regular data collection by Deutsche Bundesbank starting in 2023Q1. ### Property Price-at-Risk (PaR) results (IMF staff estimates) - Residential real estate (RRE) PaR: - Analysis points to increased tail risks in 2021Q3 compared to pre-pandemic at end-2019. - Under an adverse scenario, RRE prices could fall by some 14 percent (in cumulative terms) over 3 years. - Commercial real estate (CRE) PaR: - Models suggest increased tail risks in 2021Q3 compared to pre-pandemic at end-2019. - Under an adverse scenario, CRE prices could fall by 30 percent (in cumulative terms) over 3 years. - PaR methodology notes: - RRE and CRE conditional one-year-ahead probability distributions of price growth are based on a parametric, t-skew density fitted over quantile regression estimates for 2021:Q3 and 2019:Q4; figures are annualized growth rates. - The RRE adverse scenario is calibrated as a simultaneous 2 standard deviations shock to leverage (change in household debt-to-GDP, interest payments-to-disposable income), affordability measure (house price-to-gdp per capita ratio (misalignment)), and financial conditions index. - The CRE adverse scenario is calibrated as a simultaneous 2 standard deviations shock to employment. - Two- and three-year-ahead PaR estimates are compounded growth rates. ### Banking sector soundness indicators (September 2021 snapshot, selected) - Commercial Banks: - Tier 1 Capital Ratio to RWA: 17.4 - Total Capital Ratio (CAR) to RWA: 20.1 - Liquid assets to total assets: 39.5 - Liquid assets to short term liabilities: 160.9 - ROE: 4.4 - ROA: 0.4 - Interest margin to gross income: 34.9 - Noninterest expenses to gross income: 71.4 - NPL to gross loans: 1.8 - NPL net of provisions to capital: 8.3 - Provisions to NPLs: 37.9 - All banks (aggregate): - Tier 1 Capital Ratio to RWA: 16.8 - Total Capital Ratio (CAR) to RWA: 18.8 - Liquid assets to total assets: 26.1 - Liquid assets to short term liabilities: 170.9 - ROE: 3.9 - ROA: 0.4 - Interest margin to gross income: 40.7 - Noninterest expenses to gross income: 65.6 - NPL to gross loans: 1.4 - NPL net of provisions to capital: 7.0 - Provisions to NPLs: 35.4 ### Banking sector stress testing framework (solvency test, key features) - Institutional perimeter: - 16 SIs under IFRS9 accounting and 1,293 LSIs. - Market share covered: 87 percent of banking system assets. - Total assets for SIs (16 banks using IFRS): €7,603.186 billion. - Total assets for LSIs (1,295 banks): €3,200 billion. - Effective date: Q4 of 2021 for SIs and Q3 of 2021 for LSIs. - Scope of consolidation: consolidated. - Channels and modeling: - Credit risk parameter (PD, LGD, EAD) projections by geographical breakdown (5 jurisdictions) and product (6 asset classes: retail unsecured, retail secured, corporate institutions, and central banks and central governments). - SIs’ credit risk modeling relied on IFRS9 modeling and transition matrices; LSI credit risk modeling relied on traditional approaches using domestic GAAP. - SIs used PDs and LGDs reported in COREP templates and considered exposures by jurisdiction (Germany, the U.S., the U.K., France, Italy, and Spain). - LSIs benchmarked starting PDs and LGDs from EBA risk dashboard for the German banking system. - Nonfinancial corporates’ credit risk elasticity parameters derived from empirical models using German Credit Registry supervisory data with 2021 as benchmark. - Household default risk modeled via a micro-macro model and ECB Household Finance and Consumption Survey under full recourse mortgages assumption (households default only if unemployed and fully deplete financial savings). - Other channels: - Interest rates and Net Interest Income (NII): empirical regression model covering 2006-2019, linked NII to spread between short-term rate and long-term government yield; assumed a 100 percent pass-through from policy rates/short-term market rates to funding costs of banks (very conservative). - Stress test assumed static balance sheet. - Adverse scenario included a 10 percent loss in fee and commission income in 2022-23 (for SIs adverse scenario). - Sovereign credit risk and market risk: Merton model with shocks to sovereign exposures to Italy and Spain (and exposures to Russia) and an additional shock of 25 percent to real estate valuations. - For SIs: evolution of IFRS9 transition matrices based on beta-linked models. - Stress test horizon and scenarios: - Horizon: 5 years (2022-2026). - Baseline from the revised Spring WEO. - Adverse scenario severity benchmarked on a 3 standard deviation shock to real GDP growth relative to baseline over 2022-2023 with closing of output gap at end of simulation horizon. - Macro-financial simulations realized based on MCM GFM macro-financial DSGE model. - Adverse scenario characterized by V-shape path for real GDP growth, tightening of global financial conditions, renewed Covid infections and lockdown measures, global supply chain disruptions, rise of commodity prices, de-anchoring of inflation expectations, and a trade-off for monetary policy between unemployment and inflation. - Sensitivity analysis includes shocks to exposures to Russia. - Data sources: - Supervisory data provided by the ECB for SIs and by the Bundesbank for LSIs (FINREP, COREP; STE files and EBA 2021 stress test submissions for SIs). - Other sources: EBA Transparency Exercises, Fitch, Moody’s KMV, Bloomberg, Haver Analytics, IMF Global Assumptions (GAS), and IMF WEO. ### Key risks, data gaps, and vulnerabilities - Data gaps: - Lack of comprehensive borrower-based information (e.g., LTV, DSTI) may lead authorities to underestimate risks. - Representative hard data on lending standards are not available on a regular basis; the latest special survey on lending standards was conducted by BaFin and Bundesbank in 2019, and by the ECB in 2019. - Vulnerable sub-sectors: - Non-food retail and potentially hotel businesses identified as potential immediate risks. - Macroeconomic risk drivers: - COVID outlook uncertainties, impact on economic growth, rising interest rates, and higher inflationary pressures could pose risks to real estate valuations and bank exposures. ### Policy implications and recommendations (implicit in source text) - Improve borrower-based data collection (e.g., LTV, DSTI) to better assess household and mortgage vulnerabilities. - Monitor sub-sectors with elevated near-term risk (non-food retail, hotels) for loan performance deterioration. - Continue regular collection and publication of lending standards data (Deutsche Bundesbank data collection starting 2023Q1 noted as addressing gaps). - Incorporate adverse PaR scenarios (RRE: cumulative fall of some 14 percent over 3 years; CRE: cumulative fall of 30 percent over 3 years) into supervisory stress testing and contingency planning. - Maintain vigilance on pass-through of rising policy and market rates to bank funding costs and NII given the conservative assumption of 100 percent pass-through used in the stress test. *Source: IMF staff estimates and calculations, and narrative from the IMF Financial Sector Assessment content unit.* ### 4. Risks and ### 4. Risks and Buffers ### A. Banking Sector Solvency Test - Domain: Framework - Behavioral Adjustments - Static balance sheet assumption for LSIs and for SIs. - Base simulations assume no write-offs, while robustness simulations consider write-offs aligned with bank-asset class level write-off rates observed in recent data. - Portfolio composition unchanged over time. - Regulatory and Market-Based Standards and Parameters - Calibration of Risk Parameters - TTC and Initial PiT PDs and LGDs obtained from supervisory files for SIs, or estimated at the asset class level from the EBA risk-dashboard 2021:Q3 for LSIs. - Dynamic from model estimated PDs in line with the scenario considered (WEO baseline, adverse scenarios). - Regulatory/Accounting and Market-Based Standards - Regulatory capital ratios for IRB and STA portfolios, and IFSR9 or national GAAP accounting standards. - For SIs and LSIs, the hurdle rate includes the minimum CET1 ratio, the conservation buffer and the CCyB of 0.75 percent starting Q1 of 2023. A systemic risk buffer of 0.5 percent of RWAs is also added to the hurdle rate of SIs. - Reporting Format for Results - Aggregate results and contributions to evolution of capital ratios. ### B. Liquidity Banking Sector Stress Testing Matrix (STeM) - Domain: Framework — Top-down by FSAP team 1. Institutional Perimeter - Institutions Included - 17 SIs, and 40 randomly selected German LSIs of four different groups: commercial, savings, cooperative, and building societies and mortgage banks. The composition of the LSI sample was chosen to balance the need to include a sufficient number of banks of each group and to reflect, within this constraint, the different number of entities belonging to each group and quintile of the banking sector’s asset distribution. - Excludes branches of non-German banks. - Market Share - 17 SIs (out of 21), account for about [45] percent of banking sector assets. - 40 LSI (out of >1300 LSIs), randomly chosen to be representative of bank type and size distribution. The sample accounts for about 3 percent of LSIs (in terms of banks and assets). - Data and Baseline Date - ECB/SSM and Bundesbank: Liquidity Coverage Ratio and the Net Stable Funding Ratio and Cash flow table from the COREP data repository. - Data as of September 2021 for LSIs and December 2021 for SIs. - Scope of financial consolidation: consolidated at national bank level. 2. Channels of Risk Propagation - Methodology - The cash-flow stress test analyzes the net cash balance, accounting for available unencumbered assets, contractual cash inflows and outflows, and behavioral flows. - For the cash-flow analysis, relevant second-round effects could be considered, including margin calls for existing collateral positions, non-emergency liquidity provision by the central bank, additional asset haircuts due to fire sales, additional repo haircuts due to limited collateral supply, and wholesale funding market freezes because of banks’ solvency and liquidity concerns. - The test was repeated for US dollar liquidity for the relevant reporting banks. - The analysis is complemented with LCR and NSFR statistics. - Stress Test Horizon - For the cash-flow analysis, the horizon of stress events varies by scenario and can extend up to a period of 1 year). 3. Tail Shocks - Scenario Analysis - Baseline and three scenarios are considered, with varying intensity of adverse liquidity conditions and reflecting different liquidity risks (funding and market liquidity). The three stress scenarios were formulated in terms of roll-on/roll-off rates and haircuts to CBC and were designed to capture risks from: - (a) a new wave of covid cases, characterized by net outflows of retail deposits of households drawing down their savings (peaking up at the one-month horizon) and increased use of credit lines. - (b) a significant increase in risk aversion, with higher haircuts on counterbalancing capacity assets due to financial market stress and some outflows of wholesale funds; outflows related to rating downgrades and some deposit outflows peaking at the 2-month horizon. - (c) combined significant increase in risk aversion with higher and more sustained inflation with correspondingly stronger outflows of nonoperational deposits peaking in a one year-horizon and stronger haircuts but weaker outflows from downgrades. - Sensitivity Analysis - A range of alternative scenarios was applied to the entire set of LSIs. 4. Risks and Buffers - Risks/Factors Assessed (how each element is derived, assumptions) - Funding liquidity risk is reflected in funding and asset roll-off rates, the latter providing cash inflows related to non-renewal of maturing assets. - Market liquidity risk is reflected in asset haircuts, which could be influenced by market movements, potential fire sales and collateral supply considerations. - Behavioral Adjustments - Liquidity from the central bank’s emergency lending assistance (ELA) is not considered. - The cash-flow analysis may consider some behavioral assumptions about a counterparty’s ability or willingness to transact based on banks’ solvency and liquidity conditions. 5. Regulatory and Market-Based Standards and Parameters - Calibration of Risk Parameters - The cash-flow analysis may incorporate relevant second-round effects. - Stress funding run-off rates, asset roll-over rates, and asset haircuts are calibrated based on empirical evidence and relevant international experiences. - Regulatory/Accounting and Market-Based Standards - LCR per Basel III; the hurdle at 100 percent (at the aggregate currency level). - Net cash balance for the cash-flow analysis; to pass, a non-negative net cash balance is required, where the balance reflects net funding outflows and counterbalancing capacity. - NSFR per Basel III; limit of 100 percent applied. 6. Reporting Format for Results - Output Presentation - Changes in the system-wide liquidity position, including important drivers for cash outflows, cash inflows and counterbalancing capacity. - Distribution of banks’ liquidity positions. - Number of institutions with LCR/NSFR below 100 percent and/or negative net cash balance. - Amount of liquidity shortfalls (scaled). 7. Infrastructure - Infrastructure developed by IMF staff with FINREP/COREP data input. *Source: 4. Risks and Buffers (chapter content).* --- _Source: https://www.imf.org/-/media/files/publications/cr/2022/english/1deuea2022009.pdf_