## EXECUTIVE SUMMARY — 1eurea2025007 ## Source details **Canonical URL:** [EXECUTIVE SUMMARY — 1eurea2025007](https://www.imf.org/-/media/files/publications/cr/2025/english/1eurea2025007.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2025/english/1eurea2025007.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2025/english/1eurea2025007.pdf.json) --- ### Overview and context - Focus: euro area (EA) banking sector resilience amid successive shocks. - Structural strengths: - "Strong starting capital positions, ample liquidity and a diversified deposit base". - Banks absorbed COVID-19 effects, surge in inflation, and rapid monetary policy tightening. - Aggregate indicators (end-2024 / referenced dates preserved): - CET1 ratio: "15.9 percent in December 2024". - LCR: "158 percent". - RoE: peak "10.1 percent" in June 2024; "declined to 9.5 percent in December 2024". - Banks’ assets: "EUR 38 trillion or two and a half times GDP". - Eurosystem consolidated balance sheet: "EUR 6.4 trillion" (declined from 69 percent to "42 percent of GDP in 2024"). - NPL ratios: "3.5 percent in 2020:Q2" → "2.24 percent in 2023:Q1" → "2.28 percent in 2024:Q4". - SIs unrealized results in 2023: "EUR 73 billion of unrealized losses ... against EUR 165 billion of profits". - Banking system total assets (SIs): "26,837" (in billions of euros) and "15.9 percent" CET1 in 2024. ### Stress testing design and solvency findings - Exercise design: - Top‑down solvency stress test on end‑2024 data over 2025–2027 covering 95 of 109 SIs. - Two severe three‑year adverse macrofinancial scenarios: Scenario A (geopolitical escalation) and Scenario B (deep recession). - Static balance sheet approach; results reported on a transitional basis (CRR3 not considered). - Principal solvency outcomes: - System remains "broadly resilient, but it would face challenges under adverse macrofinancial conditions." - Only "a few banks would breach regulatory capital requirements" under scenarios. - "A much larger share would see prudential buffers eroded and 'a sizable portion of the system would operate with diminished lending capacity'." - Peak impact is "most pronounced during the second year of the stress period" (2026). - Main drivers of capital depletion: - Deterioration in credit quality → elevated provisioning. - Expansion of RWAs as exposures become riskier. - Income compression: "Both net interest income and fee-based revenues decline". - Solvency stress‑test aggregate capital paths (system‑wide, Figure 8): - Baseline: CET1 rises from "15.7 percent in 2024 to 16.6 percent in 2027"; TCR from "20.0 percent to 20.4 percent". - Geopolitical: CET1 falls to "11.4 percent" and TCR to "14.9 percent" by 2027. - Recessionary: CET1 "11.1 percent" and TCR "14.7 percent" by 2027. - Implied cumulative decline: roughly "450–500 basis points versus baseline". - Bank‑level shortfalls and breaches: - About "8-9 banks (out of 95 SIs) could breach their SREP capital requirements." - Aggregate capital shortfall: "0.05 percent of RWA" (geopolitical) and "0.1 percent of RWA" (recessionary). - Between "25-30 percent of banks would dip into their prudential buffers." - Additional capital required to restore buffers: "1.0 percent of RWA (geopolitical)" and "1.1 percent of RWA (recessionary)." ### Heterogeneity by business model and size - G‑SIBs: - Enter stress with "thinner capital buffers". - CET1 slides from "14.3 percent to 9.2–9.4 percent"; TCR to about "13 percent" in 2027. - 2024 RoE: "7.9 percent". - Lenders: - Stronger starting positions; baseline CET1 ~ "17.9 percent". - Recessionary CET1 ~ "11.7 percent" in 2027. - Provisions peak (system role): lenders show highest sensitivity with 2026 provisioning rising to "2.40 percent (geopolitical)" and "2.16 percent (recessionary)". - Investment banks: - "Prove the most resilient"; retain near "20–21 percent" CET1 and TCR above "26 percent" after stress. - NIM falls from "0.78 percent" (2024) to "0.11 percent" under recessionary scenario. - Universal banks: - End severe scenario CET1 "just below 12 percent" and TCR near "14.8 percent". - Baseline NIM ~ "1.92 percent" → "1.50 percent" (recessionary by 2027). - Leverage dynamics: - Aggregate baseline leverage: "5.7 percent in 2024 to 6.8 percent in 2027". - Geopolitical trough: "5.2 percent in 2026"; recessionary ends at "5.0 percent in 2027". - G‑SIB leverage drops "4.8 percent to about 4.0 percent" in adverse paths (close to regulatory minima "3.60-3.85 percent" including P2R and G‑SIB add‑on). ### Household and corporate credit risk modeling - Household vulnerability (HFCS‑based projections to end‑2026): - Baseline: "15 percent of households, holding 17 percent of outstanding debt, could become overburdened (essential payments > 70 percent of income)." - Geopolitical: "over 20 percent of households holding 22 percent of debt could be overburdened." - Recessionary: impact "would be cut by half" due to offsetting lower interest payments. - Sensitivity test (200bps interest rate increase, 10 percent wage income decrease, 5 percentage‑point unemployment shock): "20 percent of households ... holding 25 percent of bank debt" become financially stretched; aggregate consumption could fall by "about 10 percent". - PD and LGD approaches: - PDs calibrated country‑level for forty material geographies; retail models semi‑structural using microdata. - Corporate PDs via firm‑level simulations mapped to Moody’s default frequency matrix. - LGD for secured lending calibrated using LTV, reported LGDs, and property price paths; fixed LGD of "35 percent" assumed for sovereign/financial institution PDs in some estimations. ### Provisioning, market risk, NII, NFCI and other P&L items - System provisioning path (system aggregate, Figure 2): - Baseline: "0.63 percent" (2025); "0.93 percent" (2026); "0.85 percent" (2027). - Geopolitical: "0.81 percent" (2025); peak "2.21 percent" (2026); "1.08 percent" (2027). - Recessionary: "0.77 percent" (2025); peak "2.06 percent" (2026); "1.11 percent" (2027). - Total market risk capital impact: - "EUR 113 billion" (geopolitical). - "EUR 130 billion" (recessionary). - Comparable to "EUR 136 billion" market losses in 2023 EBA adverse scenario. - NIM trajectories (Figure 6): - Baseline: "1.67 percent" (three years). - Geopolitical: falls to "1.55 percent" by 2027. - Recessionary: falls to "1.25 percent" by 2027 (about "40 basis points lower than the baseline"). - NFCI projections (system level): - Baseline NFCI ratio: "0.68 percent". - Geopolitical trough 2026: "50 bps" (system), representing a "0.2 percentage point" drop (~30 percent contraction). - Recessionary trough 2026: "55 bps" (system). - Other P&L and tax/dividend rules: - Other P&L items set to 5‑year average to assets and held constant in stress. - Bank‑specific effective tax rate: historical 5‑year average; no tax when negative PBT. - Dividend payout rate: "60 percent" of total comprehensive income; dividends set to zero when losses or capital breaches. ### Liquidity risks, CFLST, LCR, USD exposure and contingent funding - System liquidity metrics and outcomes: - LCR remained at "158 percent" (headline) and median LCR increased to "192 percent" in some samples; lowest close to "130". - Survival horizons: "exceed two months for most banks under various stress scenarios". - Severe outflow scenario: "a negative CBC in 10 percent of banks within one month". - Cash‑flow exposures and CBC: - Contractual outflows within first four weeks: "50 percent of total funding" (weighted average). - Contractual inflows within first four weeks: "27 percent of total funding". - Cumulated net funding gap first 4 weeks: "31 percent of total funding" ≈ "EUR 7,500 billion". - CBC amounts to "21 percent of total assets", higher than net funding gap ("19 percent of total assets") in unstressed contractual data. - USD liquidity risks: - Several large banks would face USD gaps within first week; gap small: "equivalent to 0.5 percent of total assets". - Unweighted median USD LCR = "145". - Historical USD LCR falls to "0" in several G‑SIBs and some investment banks. - Contingent liquidity amplification: - Forced sales by stressed NBFIs could reduce system‑wide average LCR by "almost 50 percentage points", though "most banks’ LCRs would remain above 100 percent". - Increased secured funding, FX and collateral swaps, and margin calls raise contingent liquidity risks. - CFLST results and reverse CFLST: - No large bank fails mild scenarios within first month (SC1/SC3). - Idiosyncratic scenario (SC2): "10 banks failing within a week" and "almost 80 percent of banks failing within 1‑year horizon" in extreme calibrations. - Reverse CFLST: systemwide CBC becomes negative only after iterative escalation; threshold of 20 percent of failing banks by assets reached after 4th iteration in the stylized reverse test with modest CBC shortfall of "1 percent of total assets". ### Solvency–liquidity interactions and business risk - Joint solvency‑liquidity framework: - Modeled for the seven EA G‑SIBs using a Cont, Kotlicki, and Valderrama (2020) style approach. - Horizons: two‑day and two‑week exercises; includes endogenous liquidity shocks from solvency concerns, CCR, margin/collateral calls, deposit run‑offs, and business‑risk linked client attrition. - Key parameters (Table 5 exact values preserved where numeric): - Unsecuritized rate (interbank): "2.91" - Repo rate (market): "3.00" - Rep rate (CB): "3.15" - Effective repo haircut (bank specific): "5.0-9.7" - Fire sales discount: "50.0" - Fraction of illiquid assets eligible for sale: "5.0" - Downgrade threshold (bank specific): "4.10-4.35" - Insolvency threshold (bank specific): "3.60-3.85" - Amplification and non‑linearity: - Losses from the solvency‑liquidity nexus "could reach up to two hundred percent of the initial market shock." - Amplification is "highly non-linear on the size of the shock" and increases near downgrade triggers. - CCR losses from vulnerable NBFIs could be substantial if derivative positions are liquidated due to missed variation margin calls. - Scenario sets and reverse stress testing: - Reference, "credit sensitive", "business as usual", "narrow collateral", "trapped liquidity", and "business risk" scenarios considered. - Reverse stress tests map failure regions across interest‑rate shocks up to "7 percent" and equity shocks up to "50 percent". - Liquidity available under alternatives: reference liquidity "EUR 900 billion"; narrow framework maximum "EUR 750 billion"; "trapped liquidity" maximum "EUR 425 billion". - Example in reference solvency‑liquidity: equity from "EUR 285.3 to 221.3 billion"; LaR "EUR 903.7 billion"; amount borrowed "EUR 400 billion". ### Network analysis and contagion - Coverage and model: - 72 SIs (out of 109) representing about "90 percent" of EA banking assets (June 2024). - Contagion model based on CoMap (Covi, Gorpe, Kok, 2021); 377 interbank exposures. - Baseline findings: - Top 10 hypothetical default events induce on average "1.3 percent of capital losses"; no additional defaults under baseline calibration. - Two banks have CI and VI above 75th percentile; French banks dominate most contagious group; Italian banks form major share of most vulnerable. - Stylized NBFI and market risk scenarios: - NBFI scenario: most contagious event → "3.1 percent" capital losses; top ten average "1.5 percent". - Market risk + NBFI: top event → "4.4 percent" capital losses and "three additional defaults"; average top ten → "1.9 percent". - Interpretation: - "The risk of contagion through interbank exposures within the EA is low" under baseline, supported by robust capital and liquidity. - However, including NBFI links, CCR, SFTs, and market valuation channels materially raises contagion potential and failure counts. - Policy implication: strengthen monitoring of bank–NBFI linkages and incorporate broader channels (CCR, SFT, liquidity) into network simulations. ### Main policy recommendations (selected, verbatim phrasing preserved) - Solvency Stress Testing: - "Align ad‑hoc stress testing regulatory data collections with the on‑going European data integration activities for statistical, prudential and resolution data ... (¶60)." — Authorities: EBA; Timing: ST. - "Continue to expand the ECB/SSM stress‑testing program to include multiple scenarios ... (¶61)." — Authorities: ECB; Timing: ST. - Liquidity Stress Testing: - "Take a forward-looking view to measure contingent liquidity risks ... by recalibrating outflow parameters to stressed market conditions in internal stress test exercises (¶102)." — Authorities: ECB; Timing: ST. - "Provide further guidance on the quantification of outflows from own credit rating downgrades ... (¶103)." — Authorities: ECB; Timing: ST. - "Harmonize the collection of granular supervisory data ... for those banks with a low or volatile FX LCR ... (¶104)." — Authorities: ECB; Timing: MT. - Solvency‑Liquidity Interactions: - "Continue developing tailored stress tests to account for the endogenous interaction between liquidity shocks and capital erosion ... (¶130)." — Authorities: ECB; Timing: ST. - "Continue enhancing the modeling of counterparty credit risk (CCR) ... (¶130)." — Authorities: ECB; Timing: ST. - "Develop a system-wide stress test covering the entire EU financial system ... (¶130)." — Authorities: ESRB, ECB; Timing: ST. - Overall Risk Assessment: - "Conduct reverse stress testing to different configurations of macrofinancial shocks ... (¶131)." — Authorities: ECB; Timing: ST. - Timing legend: "I (immediate) = within one year; ST (short term) = 1–2 years; MT (medium term) = 3–5 years" ### Monitoring and modeling recommendations - Enhance granular, harmonized supervisory data to reduce remapping across stress cycles and enable rapid, desktop‑based stress tests (embed bespoke templates into standing reporting alongside FINREP, COREP, STE). - Complement LCR with forward‑looking conditional liquidity measures such as "liquidity‑at‑risk". - Improve CCR measurement and include NBFI vulnerabilities in solvency and joint solvency‑liquidity stress tests. - Use multiple adverse scenarios (with varied inflation and interest‑rate trajectories) and reverse stress tests to capture non‑linear amplification and business‑risk feedbacks. *Source: IMF — FSAP chapter on euro area stress testing (content unit: 1eurea2025007).* ### EXECUTIVE SUMMARY __________________________________________________________________________ 8 ### EXECUTIVE SUMMARY ### INTRODUCTION - Framework and context sections: - A. Macrofinancial Context - B. Financial Risks and Vulnerabilities - C. FSAP Stress Testing Strategy - Coverage indicates assessment of macrofinancial drivers, identification of vulnerabilities, and an articulated stress-testing strategy for the Financial Sector Assessment Program (FSAP). ### SOLVENCY STRESS TEST - Structure of the solvency stress test: - A. Key Elements of the Stress Test - B. Scenarios - C. Household Vulnerability Assessment - D. Credit Risk Models - E. Market Risk Approach - F. Net Interest Income - G. Net Fee and Commission Income - H. Other Profit and Loss Items - I. Solvency Stress Test Results - J. Recommendations - Analytical inputs and outputs signposted by figures and boxes: - Projected core macro variables (Figure 1) - Structural models for household and corporate credit risk (Box 1, Box 2) - Credit provisions and provisioning dynamics (Figure 2) - NII portfolio segmentation and passthrough analysis (Figures 3–6) - Profitability and capital metrics tracked: Net Fees and Commissions (Figure 7), Capital Adequacy Ratio (Figure 8), Leverage Ratio (Figure 9), CET1 Shortfall (Figure 10) - Contributions to capital depletion and profitability breakdowns (Figures 11–12) - Explicit treatment of household vulnerability and credit risk modeling, with technical details documented in appendices (Appendix IV: Solvency Stress Test—Technical Aspects; Appendix V: Solvency Stress Test—Detailed Results by Business Model). ### LIQUIDITY STRESS TEST - Structure of the liquidity assessment: - A. Overview and Key Findings - B. Structural Liquidity Risks - C. Cash Flow Liquidity Stress Tests - D. LCR Stress Tests - E. Recommendations - Key analytical elements and visualization: - Excess liquidity and composition of ECB APP portfolio (Figure 13) - Liquidity risk analysis overview and funding structure by maturity (Figures 14–15) - ECB liquidity provision and USD metrics (Figures 16–17) - LCR concentration, links to banks and NBFIs, and contingent liquidity risks (Figures 18–20) - Composition of counterbalancing capacity by credit quality step and cash-flow-based stress test results (Figures 21–22) - Scenario evolutions for CBC and reverse cash flow tests (Figures 23–25) - LCR-based stress test results and sensitivity to stressed haircuts from NBFI liquidity analysis (Figures 26–29) - Table-level documentation of counterbalancing capacity: Table 4 (Composition of the Counterbalancing Capacity) - Technical aspects covered in Appendix VI: Liquidity Stress Test—Technical Aspects. ### SOLVENCY-LIQUIDITY INTERACTIONS AND BUSINESS RISK - Structure: - A. Motivation - B. Modeling Approach - C. Dynamics of Balance Sheets Under Stress - D. Parameterization of Model Inputs - E. A Range of Scenarios - F. Results - G. Recommendations - Modeling and diagnostic outputs: - Transmission channels and risk correlations (Figures 30–31) - Stylized balance sheet for combined analysis (Figure 32) - Joint solvency–liquidity outcomes across horizons and sensitivity cases (Figures 33–37) - Parameters summarized in Table 5 (Solvency-Liquidity Interactions: Model Parameters) - Detailed methodology in Appendix VII: Solvency-Liquidity Interactions—Technical Aspects. ### NETWORK ANALYSIS - Structure: - A. Scope - B. Modeling Framework - C. Results - D. Recommendations - Outputs and mapping: - Distributional outcomes for CET1 capital, net liquidity position, and net exposure (Figures 38–39) - Systemic risk maps and contagion/vulnerability scoring (Figure 40; Table 6: Bank Network Analysis: Contagion and Vulnerability Scores) - Technical specification in Appendix VIII: Network Analysis–Technical Aspects. ### SUPPLEMENTARY ANALYSES, FIGURES, AND TABLES - Macro and sectoral context materials: - Macrofinancial conditions and real estate market visuals (Figures 41–42) - Bank market values and business model comparisons, investment fund sector connectivity, and household risk across scenarios (Figures 43–45) - Tables and indices: - Table 1: Main Recommendations on Stress Testing - Table 2: Structure of the EA Financial System - Table 3: Financial Soundness Indicators for Significant Institutions - Table 7: Main Economic Indicators, 2021–2030 - Comprehensive appendices and technical notes listed (I–VIII), including a Risk Assessment Matrix (RAM) and Stress Testing Matrix (STeM). ### GLOSSARY AND TERMINOLOGY - Extensive glossary of acronyms and terms used throughout the assessment, including but not limited to: - €STR, ABS, AC, AE, AM, APP, AuM, CAR, CBC, CBR, CCB, CCoB, CCP, CCR, CCyB, CET1, CF, CFLST, CI, CM, COREP, CRR, CRE, CVA, EA, EAD, ECB, EDF, EEPE, EQ, EU, FINREP, FRTB, FSAP, FVOCI, FVPL, FX, GC, GDP, GFC, G-SIB, HFCS, HQLA, IF R S, IM, IMF, IR, LaR, LCR, LGD, LSI, MIR, NBFI, NFC, NFCI, NFG, NII, NIM, NLO, NPE, NPL, NSFR, NTI, OCI, OFI, O-SII, OTC, P&L, P2R, PBT, PD, PiT, PVA, QT, RoE, RWA, SCI, SCO, SFT, SI, STA, STE, SREP, SSM, SyRB, TA, TCI, TCR, TM, TN, TTC, VaR, VI, WEO, WM. *Source: https://www.imf.org/-/media/files/publications/cr/2025/english/1eurea2025007.pdf* ### EXECUTIVE SUMMARY ### EXECUTIVE SUMMARY ### Overview and context - Focus: euro area (EA) banking sector resilience amid successive shocks. - Key structural strengths cited: - "Strong starting capital positions, ample liquidity and a diversified deposit base". - Banks absorbed COVID-19 effects, surge in inflation, and rapid monetary policy tightening. - Aggregate indicators (end-2024/2023 references): - CET1 ratio: "15.9 percent in December 2024". - Liquidity Coverage Ratio (LCR): "158 percent". - Return on Equity (RoE): peaked at "10.1 percent" in June 2024; "declined to 9.5 percent in December 2024". - System assets: banks’ assets amount to "EUR 38 trillion or two and a half times GDP". - Eurosystem consolidated balance sheet: "EUR 6.4 trillion" (declined from 69 percent to "42 percent of GDP in 2024"). - NPL ratios: declined from "3.5 percent in 2020:Q2 to 2.24 percent in 2023:Q1 but rose to 2.28 percent in 2024:Q4". - SIs unrealized losses/profits in 2023: "EUR 73 billion of unrealized losses ... against EUR 165 billion of profits". - Banking system total assets (SIs): "26,837" (in billions of euros) and "15.9 percent" CET1 in 2024 (see financial soundness table excerpt). ### Stress testing design and solvency findings - Exercise: top-down stress testing using data up to end-2024 under two severe three-year macrofinancial scenarios: - Scenario A: escalation of geopolitical tensions. - Scenario B: deep and widespread recession. - Main solvency outcomes: - "The solvency analysis finds that the EA banking system remains broadly resilient, but it would face challenges under adverse macrofinancial conditions." - Only "a few banks would breach regulatory capital requirements" under the scenarios. - A much larger share would see prudential buffers eroded and "a sizable portion of the system would operate with diminished lending capacity". - Timing of peak impact: "most pronounced during the second year of the stress period". - Primary drivers of capital depletion: - Deterioration in credit quality causing elevated provisioning. - Expansion of risk-weighted assets as exposures become riskier. - Income compression: "Both net interest income and fee-based revenues decline" leading to insufficient earnings to fully offset earlier losses. ### Heterogeneity by business model and size - G-SIBs: - Enter stress with "thinner capital buffers". - "Experience the most significant capital depletion and some approach minimum regulatory thresholds." - Lower 2024 RoE: "7.9 percent". - Mid-sized domestic/diversified lenders: - Benefit from "stronger initial positions" and higher RoE (diversified lenders at "12.2 percent"). - Investment banks and asset managers: - "Prove the most resilient" with higher starting capital and lower credit-sensitive exposures. - Universal banks: - "Broadly follow the average trend" but benefit from diversified income streams moderating the impact. - Larger banks: - "Posted lower capital and liquidity ratios than medium/small sized banks." ### Liquidity risks, USD exposure, and contingent funding - System-wide liquidity metrics: - LCR remained stable at "158 percent". - Survival horizons: "exceed two months for most banks under various stress scenarios". - No large bank moves into negative CBC in mild outflow scenarios within the first month; under severe outflows, "a negative CBC in 10 percent of banks within one month". - USD liquidity risk: - "Some banks have material exposure to U.S. dollar (USD) liquidity risk." - USD cash flow tests: several large banks would face USD gaps within the first week, but gaps are small: "equivalent to 0.5 percent of total assets". - These USD gaps are "small ... compared to the depth of EUR/USD swap markets". - Contingent liquidity risks increased since 2018 due to: - Higher share of secured funding, FX and collateral swaps triggering collateral calls. - Exposure to margin calls on derivatives. - Accounting for forced sales by stressed NBFIs could reduce system-wide average LCR by "almost 50 percentage points", though "most banks’ LCRs would remain above 100 percent" given significant buffers. ### Joint solvency-liquidity interactions and amplification - Joint stress testing for G-SIBs modeled: - Endogenous liquidity shocks triggered by solvency concerns. - Business model sustainability via client attrition and credit downgrades tied to excessive leverage. - Liquidity outflows linked to LCR run-off rates over two-day and two-week horizons, including borrowing costs of mitigating actions. - Key quantitative insights: - Losses from the solvency-liquidity nexus "could reach up to two hundred percent of the initial market shock". - Losses are "highly non-linear on the size of the shock" and increase near a credit downgrade. - CCR losses from vulnerable NBFIs could be substantial if banks liquidate derivative positions due to missed variation margin calls. ### Monitoring, metrics, and model recommendations - Need for granular monitoring: - "Granular monitoring of balance sheet items in relation to their sensitivity to solvency and liquidity risk, particularly when correlations break under stress." - Proposed metric enhancements: - Complement LCR with forward-looking conditional liquidity measures such as "liquidity-at-risk". - Enhance modeling of CCR, including exposures to vulnerable NBFIs. - Reverse stress and sensitivity testing: - Highlights heterogeneity in banks’ sensitivity to market shocks and CCR losses. - Reverse stress testing identifies plausible shock combinations triggering insolvency when liquidity feedback loops are considered. - Sensitivity tests underscore the ECB’s broad collateral framework's stabilizing role while quantifying funding losses from trapped liquidity and franchise erosion. ### Interconnectedness and contagion - Interbank network analysis: - "The risk of contagion through interbank exposures within the EA is low." - Robust capital and liquidity underpin interbank resilience. - Risks from NBFIs and market volatility remain important amplifiers that can produce capital losses and cascading defaults through networks of large exposures. ### Main policy recommendations (selected and verbatim phrasing preserved from Table 1) - Solvency Stress Testing - "Align ad‑hoc stress testing regulatory data collections with the on-going European data integration activities for statistical, prudential and resolution data ... (¶60)." — Authorities: EBA; Timing: ST. - "Continue to expand the ECB/SSM stress‑testing program to include multiple scenarios ... (¶61)." — Authorities: ECB; Timing: ST. - Liquidity Stress Testing - "Take a forward-looking view to measure contingent liquidity risks ... by recalibrating outflow parameters to stressed market conditions in internal stress test exercises (¶102)." — Authorities: ECB; Timing: ST. - "Provide further guidance on the quantification of outflows from own credit rating downgrades ... (¶103)." — Authorities: ECB; Timing: ST. - "Harmonize the collection of granular supervisory data ... for those banks with a low or volatile FX LCR ... (¶104)." — Authorities: ECB; Timing: MT. - Solvency-Liquidity Interactions - "Continue developing tailored stress tests to account for the endogenous interaction between liquidity shocks and capital erosion ... (¶130)." — Authorities: ECB; Timing: ST. - "Continue enhancing the modeling of counterparty credit risk (CCR) ... (¶130)." — Authorities: ECB; Timing: ST. - "Develop a system-wide stress test covering the entire EU financial system ... (¶130)." — Authorities: ESRB, ECB; Timing: ST. - Overall Risk Assessment - "Conduct reverse stress testing to different configurations of macrofinancial shocks ... (¶131)." — Authorities: ECB; Timing: ST. - Timing legend: "*I (immediate) = within one year; ST (short term) = 1–2 years; MT (medium term) = 3–5 years*" *EXECUTIVE SUMMARY — 1eurea2025007.* ### 7. While quantitative tightening is expected to progress in an orderly way, it warrants ### 7. While quantitative tightening is expected to progress in an orderly way, it warrants ### Monitoring liquidity conditions and NBFI–bank interconnectedness - Careful monitoring of liquidity conditions in banks and NBFIs is warranted given the growth of the NBFI sector and its interconnectedness with banks (Figure 44). - Potential widening of sovereign spreads could raise borrowing rates and heighten credit risk. - If fiscal headroom erodes, banks’ exposures to highly indebted sovereigns could act as a transmitter of shocks and adversely affect the availability of credit to the real economy. - Excessive leverage in NBFIs could amplify stress via counterparty credit risk; example: the 2021 failure of Archegos resulted in over USD 10 billion of losses across banks. - Large credit losses can lead to franchise erosion, reduced client activity, funding pressures, and adverse feedback loops. - Recent spike in financial market volatility following US trade tariff announcements, coupled with European banks’ growing reliance on US repo funding, could create rollover risks. ### FSAP Stress Testing Strategy (overview) - A range of stress tests was used to assess resilience: - A fully-fledged solvency stress test against two severe, adverse macrofinancial scenarios. - A wide range of liquidity stress tests to estimate resilience to market-wide and idiosyncratic liquidity risks. - A solvency-liquidity stress test to quantify amplification of equity losses from endogenous liquidity stress, counterparty credit risk, and business risk. - A network analysis to assess amplification losses and cascading defaults through interbank exposures, including under stressed market conditions. - A system-wide stress test to evaluate the ability of NBFIs to satisfy liquidity demands in a market shock scenario and assess knock-on effects on core funding markets and financial institutions; findings were used to stress banks’ LCR ratios and inform counterparty credit risk (CCR) analysis in the solvency-liquidity module. ### Relation to EU-wide stress tests - The 2023 EU-wide stress test found a capital depletion of 459 basis points under the adverse scenario. - At the time of the 2025 FSAP, banks’ resilience was being tested under the 2025 EU-wide stress test, with results expected to be published in August 2025. - The FSAP stress testing exercise includes additional modules (liquidity risk, solvency–liquidity interactions, network effects) beyond the EU-wide exercise. ### Solvency Stress Test — Key Elements - Data and sample: - Conducted on end-2024 data at the highest level of consolidation in the EA. - Performed over 2025–2027. - Covered 95 (out of 109) SIs in the EA COREP and FINREP data; 14 banks excluded due to insufficient regulatory data or specialized business models. - Excluded LSIs; contagion dynamics from correlated failures of small entities are out of scope. - Hurdle rates used to assess resilience: - Minimum capital ratio / SREP capital requirement: Pillar 1 (4.5 percent for CET1) plus Pillar 2 (P2R). - Minimum capital ratio plus the combined buffer requirement (CBR) consisting of Capital Conservation Buffer (CCoB), Countercyclical Capital Buffer (CCyB), G-SII buffer, O-SII buffer, and Systemic Risk Buffer (SyRB). - The 3 percent Basel III minimum leverage ratio. For the solvency-liquidity module, the leverage requirement included P2R and G-SIB add-on. - Methodology: - Static balance sheet approach: exposures and liabilities remain constant at the cut-off date level throughout the horizon. - Results cover all 95 banks in aggregate; banks grouped into four business-model groups: - G-SIBs (7 globally diversified banks) - Lenders (59 banks focused on deposit taking and loan origination) - Investment banks (11 banks with income from fees, commissions, and trading) - Universal banks (18 banks integrating lending with insurance, fee-based services, and trading) - Developmental, promotional, and custodian banks were excluded due to specialized business models. - Transitional treatment: - The FSAP stress testing analysis does not consider implementation of the Capital Requirement Regulation (CRR3); results are presented on a transitional basis and do not account for the new EU banking package effective from January 1, 2025 (restated end-2024 balance sheets and output floor impact). ### Scenarios — Macrofinancial and Market Risk - Macrofinancial scenarios: - Two adverse scenarios: a “geopolitical scenario” and a “recessionary scenario.” - Geopolitical scenario: escalation of conflicts, heightened commodity price volatility, disrupted global production chains, large adverse trade, price, and tariff shocks (“trade wars”); for the EA, GDP growth projected at -3 percent in 2025 (a contraction of 4 percentage points from baseline) relative to the 0.2 percentage-point slowdown forecasted in the April 2025 WEO (due to the April 2 tariffs). - Recessionary scenario: synchronized global slowdown amplified by sovereign debt distress in the EA, widening credit spreads, term premium decompression, and confidence losses; accommodative monetary policy mitigates adverse impact on aggregate demand. - Baseline scenario: WEO projections of the global economy and financial conditions; baseline in line with January 2025 WEO. - Scenarios project GDP growth, unemployment, headline inflation, housing prices, labor costs, and oil price for all 20 EA members and 16 major economies. - Severity relative to historical shocks: - Trough GDP paths at -3 and -4 percent, compared to GFC and Covid-19 peaks at -4.5 and -6 percent respectively. - Cumulative 2-year growth is -9 and -8.3 percent relative to baseline, corresponding to 2.8 and 2.6 standard deviations in the recessionary and geopolitical scenarios, respectively. - Unemployment shock in the recessionary scenario reaches 13.5 percent (more severe than the 12 percent peak after the European sovereign debt crisis). - Housing price drops of -14.5 and -9.5 percent, more severe than historical stress events. - Inflationary shock in the geopolitical scenario at 4.5 percent; deflationary shock in the recession scenario at -0.2 percent. - Market risk scenarios: - Two short-term market stress scenarios calibrated to expected shortfall at 0.1 percent of market factors marginal distributions over holding periods of 20 or 60 days. - Left-column scenario (aligned with geopolitical macro scenario): increase in interest rates (short end of EUR yield curve), commodity prices and credit spreads, and a sharp contraction in equity prices. - Right-column scenario (aligned with recessionary macro scenario): drop in commodity and equity prices, larger increase in credit spread of mid- and high-risk EA sovereigns, short-end EUR yield curve remains muted. ### Household Vulnerability Assessment - Vulnerabilities vary across the EA with higher risk for households with greater leverage, higher share of cost-of-living expenses relative to income, and prevalence of floating-rate mortgages. - Heterogeneity in floating-rate mortgage shares: highest in Finland and Estonia, lowest in France and Germany. - Using 2021 HFCS microdata, simulation results by end-2026: - Under baseline, 15 percent of households, holding 17 percent of outstanding debt, could become overburdened (essential payments > 70 percent of income). - In the geopolitical scenario, over 20 percent of households holding 22 percent of debt could be overburdened. - In the recessionary scenario, the impact would be cut by half due to offsetting effects of lower interest payments and cost-of-living expenses. - Sensitivity test (200bps interest rate increase, 10 percent wage income decrease, 5 percentage-point unemployment shock): - 20 percent of households could be financially stretched on average, holding 25 percent of bank debt. - One out of five consumers could cut non-essential consumption to repay debt and afford basic expenses, leading to a reduction of about 10 percent in aggregate consumption and potential second-round effects exacerbating default risk. - These household results were used to quantify loan loss provisions from retail loans in the solvency stress test. ### Credit Risk Models and PD Estimation - Credit risk covers potential losses from loans and advances (domestic and cross-border) to households, corporates, governments, credit institutions, other financial institutions, and default risk in debt securities. - IFRS measurement: debt securities in the banking book can be measured at amortized cost (AC), fair value through profit and loss (FVPL), and fair value through other comprehensive income (FVOCI). - Modeling approach: - AC securities: credit impairments estimated similarly to loans and advances. - FVPL securities: credit risk embedded in market risk methodology via price changes from risk-free rate movements or credit spread widening. - FVOCI securities: credit risk estimated through both market risk methodology and banking book credit impairment estimation. - Probability of Default (PD) estimation: - PDs for household lending calibrated at the country level covering the forty material geographies of SIs (the twenty EA countries plus: Australia, Brazil, China, Czech Republic, Denmark, Hong Kong, Hungary, India, Japan, Mexico, Norway, Poland, Romania, Russia, Singapore, Sweden, Switzerland, Türkiye, UK, and US). - Other exposures treated as domestic. - Retail credit risk models use a semi-structural approach incorporating macrofinancial conditions, borrower characteristics, and loan terms. - For household loans, country-level models calibrated for the twenty-two European countries covered in the 2021 HFCS; remaining geographies mapped to comparable EA countries based on household balance sheet leverage. - For corporate loans and bonds in AC and FVOCI, PDs derived using firm-level data on a country-aggregate basis for the same set of countries. - PDs for sovereigns and financial institutions (loans and debt securities subject to provisioning) estimated using projected credit spreads defined as the difference between bond yields and the risk-free rate—approximated by Germany’s 10-year government bond yield under the scenarios—assuming a fixed loss given default (LGD) of 35 percent. - PD-path construction: - For each bank and lending segment, a PD path over each scenario was generated as a weighted average of aggregate, country-specific PDs based on distribution of the bank’s lending exposures across countries (COREP C09.02). - The PD path is adjusted to reflect bank- and segment-specific starting point-in-time (PiT) PD, calculated as the weighted average of observed transition probabilities from stage 1 to stage 3, and stage 2 to stage 3, weighted by total balances in each stage. - Adjustment performed using the distance-to-default formula (equation 1) and the Beta-linking methodology to derive IFRS 9-compliant transition matrices for each bank and lending segment. - Equation provided: - 푃퐷푖,푡 = 푁(퐺(푃퐷푖,0)+퐺(푎푃퐷푡)−퐺(푎푃퐷0)) - where PDi,t is the bank specific probability of default and aPDt is the aggregate probability of default at time t, G is inverse of cumulative distribution function of standard normal distribution and N is cumulative distribution function of standard normal distribution. *Source: IMF — FSAP chapter on euro area stress testing (content unit: 1eurea2025007).* ### 26. LGD rates for collateralized lending are calibrated through structural modelling (Gross ### 26. LGD rates for collateralized lending are calibrated through structural modelling (Gross et al., 2020) ### LGD calibration and secured lending - LGD calibration uses reported information on the value of collateral in (loan to value, LTV), starting point reported LGDs, and property price paths. - Bank-specific property price growth paths for secured LGD projections are calculated as weighted averages of the aggregate country-level property price paths, with weights based on the distribution of each bank’s exposures. - The property price growth path estimation uses the same methodology applied for the respective PDs. ### Data coverage and mapping - Firm level data from Capital IQ covering: twenty EA countries, 5 non-EA European countries (Denmark, Norway, Sweden, Switzerland, and UK), Mexico, and US. - Remaining material geographies were mapped against peer countries using data on corporate leverage. - Consequentially, the PD of sovereign lending to Germany, Austria and Netherland is zero. ### Household credit risk: structural model (Box 1) - Microdata source: 2021 HFCS (latest), covering 83,000 households and 200,000 personal files across 22 countries (20 EA countries, Czech Republic, and Hungary). - Households are “aged forward” to project financial position as of end 2024 (starting point of the stress test). - Four-step credit risk model: - First: forecast households’ balance sheets, payments, income and consumption to project ‘vulnerable’ households using Monte Carlo simulations of unemployment shocks at the person level (controlling for employment status and type of labor contract) and account for unemployment benefits for unemployed individuals (at around 10 percent of initial income). Maturing loans are replaced by new loans with the debt to income (DTI) at origination; floating rate mortgages are reset over the life of the loan; and new issuances are repriced at prevailing market rates. - Second: estimate the link between being financially vulnerable and default risk (PD). Default proxied by being on arrears over 90 days (stage 3 loans) or less than 90 days (stage 2 loans). A battery of logistic regressions at the individual household level identifies financial stress indicators and thresholds (in line with IFRS9). - Third: run a horse race across adjusted DSTI thresholds; the best performing indicator is a cost-of-living adjusted debt service to income (DSTI) ratio which includes debt service, essential consumption (food and energy cost) and rents. For most countries, the relative increase in the probability of default is highest when the borrower’s adjusted DSTI ratio exceeds 70 percent of disposable income (“overburdened” household). - Fourth: project the share of banks’ retail loan portfolio with a credit default (stage 3 loans) or credit event (stage 2 loans) by forecasting migration of loans held by overburdened households under each scenario. - Logistic regressions control for: household income tercile, savings ratio, wealth ratio, and personal / household characteristics including age, gender, education, household size, number of people employed in the household, loan to value ratio of the main residence, credit constraints, source of income, and family / public financial assistance. ### Corporate credit risk: structural model (Box 2) - PD generation follows Tressel and Ding (2021) methodology for almost 23,000 listed non‑financial corporations in 27 countries. - Uses 2024 balance‑sheet data as common anchor; country‑specific panel regressions estimate sales growth, return on assets, leverage and interest‑coverage ratio as functions of lags, real‑GDP growth, and a broad financial‑conditions index. - Macro inputs are taken from the IMF’s WEO baseline and the FSAP adverse scenarios; accounting identities transform projected income‑statement items into debt, equity, cash buffers over a three‑year horizon. - Interest expenses evolve with firms’ historical funding mix and scenario‑specific shifts in short‑ and long‑term corporate borrowing rates. - Simulated balance‑sheet indicators are mapped into one‑year forward PDs using a Moody’s matrix that assigns empirical default frequency by interest‑coverage ratio band and debt‑to‑equity band; aggregated PDs are debt‑weighted across firms. - Aggregated PDs are benchmarked to country‑specific default measures via two anchors: rescaling the 2024 country aggregate to match Moody’s‑KMV one‑year expected default frequency (EDF) using mean and median EDFs; the scaling factor is applied to the entire forward path of raw PDs under each macro scenario. ### Credit impairments and P&L treatment - Provisions and credit impairments estimated in accordance with IFRS 9: - Provisions are calculated on a 12-month basis for stage 1 exposures and on a lifetime basis for stage 2 and stage 3 exposures. - PDs for stage 3 are assumed to be 100 percent. - Exposure at Default (EAD) includes on-balance sheet amounts and off-balance sheet items (the latter weighted according to segment characteristics). - Static balance sheet approach: matured loans are assumed reinitialized in the same stage as at maturity. - No write-offs assumed; stage 3 exposures either remain in default or cure and migrate to stage 1 or stage 2 based on the projected transition matrix. - Credit impairments in the P&L correspond directly to the annual change in balance sheet provisions. ### Provisioning results (Figure 2 and narrative) - System aggregate (full sample of 95 banks) provisioning path under scenarios: - Baseline: 0.63 percent of total loans in 2025; 0.93 percent in 2026; 0.85 percent in 2027. - Geopolitical scenario: 0.81 percent in 2025; peak 2.21 percent in 2026; 1.08 percent in 2027. - Recessionary scenario: 0.77 percent in 2025; peak 2.06 percent in 2026; 1.11 percent in 2027. - Business-model heterogeneity: - G‑SIBs and universal banks broadly track the system profile. - Lenders show highest sensitivity: baseline 0.94 percent in 2026 rising to 2.40 percent (geopolitical) and 2.16 percent (recessionary) in 2026. - Investment banks least affected: provisions peak at 1.12 percent (geopolitical) and 1.06 percent (recessionary) in 2026. - Timing: provisions spike in 2026 as exposures migrate into stages 2-3; partial decline in 2027 follows recovery in property prices. - Note: provisioning rates vary significantly across member states. ### Capital requirements for credit risk - Credit risk charges in RWAs simulated separately for internal ratings‑based (IRB) and standardized (STA) portfolios. - IRB: asymptotic single risk factor model for unexpected losses (Basel III) applied across exposure types; RWAs subject to PDs and LGDs, provisions for credit losses, and credit conversion factor of off‑balance sheet items. - Regulatory through‑the‑cycle (TTC) PDs calibrated as weighted average of PiT PDs for each year of the scenarios (weighted by 0.2) and respective TTC PDs of the previous year (weighted by 0.8, starting with reported in COREP TTC PDs). - Regulatory downturn LGD is the maximum between the reported downturn LGD at end 2024 and the estimated PiT LGD. - STA: credit risk charges estimated using density of credit RWA at cut‑off date (end of 2024); NPEs subtracted assuming a risk weight of 100 percent; performing RWA to performing exposures ratio applied to project annual credit RWA; calculated separately for each bank. ### Market risk assessment (partial revaluation approach) - Instruments at fair value (FVOCI and FVPL) revalued using bank‑specific sensitivities (delta, gamma, vega) reported in the ECB’s STE template. - Risk factors: commodity risk (CM), credit spread risk (CR), equity risk (EQ), interest rate risk (IR), and foreign exchange (FX). - FX shock corresponded to first‑year FX depreciation in the adverse macro scenarios; no market shocks applied in baseline. - Market risk incorporated as a one‑off overlay in the first year; model valuation losses are not reversed in subsequent years. - Revaluation formula (aggregates delta, gamma, vega across risk factors): - ∆V = ∑(Delta_j^BB + Delta_j^TB) · ∆ε_j + 0.5 · Gamma_j · (∆ε_j)^2 + Vega_j · ∆σ_j, for j ∈ {CM, CR, EQ, IR, FX} - A floor applied to market risk losses: maximum between - i. the bank’s own fund requirements based on FRTB as reported in COREP (C 91.00); and - ii. 8 percent of the bank’s RWAs for market risk. - Concern: inclusion of instruments categorized at AC in banking book deltas can overestimate market risk because banks may report fair value deltas for instruments categorized as AC if they regularly calculate fair value. - Sensitivity adjustment for banking book deltas for IR and CR: - Deltã_j^BB = (1 − AC Debt Securities / (AC Debt Securities + FVOCI Debt Securities)) · Delta_j^BB, for j = IR, CR - The un‑adjusted Delta_j^BB may overestimate market risk; the adjusted Deltã_j^BB can underestimate it if banks exclude debt instruments reported under AC. FSAP uses un‑adjusted Delta_j^BB for main results to be conservative. - Total market risk capital impact across all banks: - EUR 113 billion in the geopolitical scenario. - EUR 130 billion in the recessionary scenario. - Comparable magnitude to the 2023 EBA EU-wide stress test where total market losses amounted to EUR 136 billion in the EBA adverse scenario. ### Net Interest Income (NII) approach and scope - Scope: all interest‑earning assets and interest‑bearing liabilities except derivatives; included instruments classified at AC and fair value (FVOCI and FVPL); only on‑balance sheet items covered. - Conservative assumption: NII in adverse scenarios not allowed to exceed the level at the cut‑off date for each bank. - NPEs assumed to generate no interest income. - Granular portfolio segmentation along two dimensions: - By portfolio instrument and counterparty sector (e.g., debt securities/wholesale funding; deposits: HH sight deposits; loans and advances: HH mortgages; etc.). - By country of counterparty: (i) all EA countries; (ii) EU non‑EA countries with country‑specific adverse scenario (Czech Republic, Denmark, Hungary, Poland and Sweden); (iii) non‑EU countries based on materiality (Brazil, Japan, Mexico, Norway, Switzerland, United Kingdom, and United States). All other exposures and liabilities treated as domestic. - Repricing gap methodology requires: - (i) repricing structure at cut‑off date (“repricing ladder”); - (ii) time series for interest rates for newly originated exposures and liabilities (“interest rates for new business”). - Interest rates for new business projected using satellite models on ECB MIR data; base rates projected using a simple structural model. - NII calculation (static balance sheet, repricing gap): - NII_b,t = NII_b,t=0 + ∑ Interest Income Delta_b,t,j − ∑ Interest Expense Delta_b,t,j - NII_b,t=0 is the NII during the last year prior to the cut‑off date and represents a “base NII”. *Source: IMF staff calculations and FSAP analysis as presented in the supplied content.* ### 42. Figures 4 and 5 display the estimated cumulative passthroughs from ∆풊 ### 1eurea2025007 - 42. Figures 4 and 5 display the estimated cumulative passthroughs from ∆풊 ### Loan and deposit new-business rate passthroughs - Figures 4 and 5 display estimated cumulative passthroughs from ∆풊풕,풄푺푻 to ∆풊풕,풋풏풃 over horizons of one and two years (Y1 and Y2). - NFC loans: - A 1 percentage point increase in the short-term rate leads to a median increase of about 0.6 percentage points in new business rates in the first year and a cumulative median increase of about 0.8 percentage points in the second year. - Mortgages (new business rates): - Median passthrough is 0.39 after one year and 0.51 after two years. - There is considerable dispersion across the 20 euro area countries. - Non-mortgage household credit (new business rates): - Medians of 0.32 after one year and 0.38 after two years. - NFC loans (new business rates): - Medians of 0.63 after one year and 0.80 after two years (considerably higher than for households). - Sight deposits: - Lowest passthroughs observed; cumulative median passthrough in the second year is below 0.05 percent for both households and corporates. - Implication: banks with a sizeable deposit franchise stand to benefit from higher interest rates. - Methodological note: - Projected new business rates were transformed into “deltas” relative to the average interest rate for the outstanding stock at T0 of the corresponding segment. - These outstanding-stock rates were obtained from MIR. - Bank-level MIR data unavailable; assumed the same projected “deltas” for new business rates applied equally to all banks. - Visualization notes: - Box-and-whisker charts show the 10th, 25th, 50th, 75th and 90th percentiles across the 20 EA countries. ### Net Interest Margin (NIM) projections and heterogeneity - Aggregate NIM trajectories (Figure 6): - Baseline: NIM stays at 1.67 percent over the three years. - Geopolitical scenario: NIM falls slightly to 1.55 percent by 2027. - Recessionary scenario: system-wide NIM falls to 1.25 percent by 2027, about 40 basis points lower than the baseline. - Business-model heterogeneity: - G-SIBs: - Start at 1.37 percent; drop to 1.11 percent in the recessionary case by 2027. - Lenders: - Start at 2.09 percent; drop to 1.43 in the recessionary scenario by 2027. - Investment banks: - NIM contracts from 0.78 percent in 2024 to 0.11 percent under the recessionary scenario. - Universal banks: - Margins ease from 1.92 percent to 1.50 percent (recessionary scenario). - Overall message: - Rising funding costs and weaker asset yields in adverse environments compress banks’ core intermediation income, with magnitudes differing by balance-sheet structure and funding mix. ### Net Fee and Commission Income (NFCI) - Projection approach: - Annual-frequency econometric panel regression with bank-specific fixed effects; regressors included lagged NFCI ratio, real GDP growth, stock market returns, CPI inflation, residential housing price inflation, first difference of 1-month EURIBOR, first difference of 10-year sovereign yield, EUR/USD FX depreciation, and growth in US stock prices. - Up to one annual lag allowed; LASSO methodology used for variable selection. - NFCI projected at the aggregate level; model estimated separately for large banks and for small and medium sized banks. - Scenario impacts and magnitudes: - Baseline: NFCI ratio assumed to remain constant at the cut-off date level (system-wide 0.68 percent). - Geopolitical scenario: - NFCI ratio drops about 0.2 percentage points at the trough. - System-wide path: 68 bps (baseline) → 60 bps in 2025 → 50 bps in 2026 (trough) → 58 bps in 2027. - At 2026 trough: G-SIBs 65 bps → 48 bps; lenders 64 bps → 46 bps; investment banks 74 bps → 55 bps; universal banks 73 bps → 55 bps. - A 0.2 percentage point drop (from 0.68 percent) represents a contraction of about 30 percent in NFCI. - Recessionary scenario: - Milder in first year: 65 bps in 2025. - Compresses fee income to 55 bps in 2026; rebounds to 61 bps in 2027. - Client revenues (NTI) handling: - Baseline: annual income from client revenues set equal to the 5-year average of NTI-to-assets ratio. - Adverse scenarios: client revenues set to 0 in year 1; 20 percent haircut relative to baseline in remaining years. - Rationale: capture near-zero client revenues in stress and partial reduction subsequently; acknowledges NTI includes revaluation of fair value instruments and averaging mitigates bias. - Key takeaway: - NFCI is highly sensitive to macro-financial stress; fee intensity in adverse scenarios remains well below baseline even after partial recovery by 2027. ### Other P&L items, taxes and dividends - Other P&L items: - Projected by taking a 5-year average relative to total assets (mostly non-interest expenses such as wages and branch operating costs). - These items held at the same level in stress scenarios as in the baseline. - Tax treatment: - Historical effective-rate approach: each bank’s average effective tax rate computed as income tax expense divided by profit before tax (PBT) over the past five profitable fiscal years. - Rate applied only to periods with positive PBT; no tax expense recognized when a bank records a loss the entire year. - If a loss year is followed by a profit year, current year PBT adjusted to offset accumulated loss carry-forward before calculating income tax. - Dividends: - Payout rate applied to total comprehensive income equal to 60 percent. - If the bank incurs losses, dividends set to zero. - Dividends further adjusted to preserve capital: when banks’ capital positions breach minimum regulatory requirements (Pillar I and II), dividends adjusted to zero. ### Solvency stress-test outcomes and capital metrics - System-wide capital outcomes (Figure 8): - Baseline: CET1 rises from 15.7 percent in 2024 to 16.6 percent in 2027; total capital ratio (TCR) edges up from 20.0 percent to 20.4 percent. - Adverse scenarios: - Geopolitical: CET1 falls to 11.4 percent and TCR to 14.9 percent by 2027. - Recessionary: CET1 11.1 percent and TCR 14.7 percent by 2027. - Implied cumulative decline roughly 450–500 basis points versus baseline. - Timing: - Capital ratios bottom in 2026 when credit losses and RWA inflation peak; modest uptick in 2027 from retained earnings and slight RWA contraction. - Business-model heterogeneity: - G-SIBs: - CET1 slides from 14.3 percent to 9.2–9.4 percent; TCR to about 13 percent in 2027—only modestly above regulatory minima. - Start with thinnest buffers and experience steepest depletion. - Lenders: - Shed around 500 bp of CET1 (baseline 17.9 percent versus recessionary 11.7 percent in 2027); stronger starting capital gives higher resilience. - Universal banks: - End severe scenario with CET1 just below 12 percent and TCR near 14.8 percent. - Investment banks: - Remain best-capitalized, retaining near 20–21 percent CET1 and TCR above 26 percent after stress. - Caveat: - Stress test excludes CCR losses which could be high for banks with large derivative-market exposures. - Leverage ratio dynamics (Figure 9): - Aggregate baseline: strengthens from 5.7 percent in 2024 to 6.8 percent in 2027. - Geopolitical shock: - System-wide leverage slips to 5.7 percent in 2025, troughs at 5.2 percent in 2026 (-130 bps versus baseline), recovers marginally to 5.3 percent in 2027. - Recessionary shock: - Ends at 5.0 percent in 2027 (-180 bps versus baseline). - G-SIBs: - Drop from 4.8 percent to about 4.0 percent in both adverse paths; close to regulatory minimum (ranging between 3.60-3.85 percent, including P2R and G-SIB add-on). - Lenders: - Fall from 7.5 percent (baseline) to roughly 5.9 percent (geopolitical) and 5.6 percent (recessionary) at 2026 trough. - Investment banks: - Bottom near 7.3 percent at stress trough before rebounding. - Universal banks: - Slip from 7.1 percent (2026 baseline) to 5.9 percent (recessionary) by 2027. - Timing: - 2026 dip coincides with peak credit-loss charges and leverage-exposure inflation; muted 2027 recovery leaves leverage constraints tighter under adverse conditions. - SREP and prudential buffer impacts (Figures 10 and related text): - Bank-level breaches and aggregate shortfall: - About 8-9 banks (out of 95 SIs) could breach their SREP capital requirements. - Aggregate capital shortfall: 0.05 percent of RWA in the geopolitical scenario and 0.1 percent of RWA in the recessionary scenario. - Between 25-30 percent of banks would dip into their prudential buffers. - Additional capital required to restore buffers: 1.0 percent of RWA (geopolitical) and 1.1 percent of RWA (recessionary). - Group-specific shortfalls: - G-SIBs: - Prudential buffer shortfall equivalent to 1.49 percent of RWA (geopolitical) and 1.43 percent of RWA (recessionary). - Lenders: - SREP capital shortfall 0.24 percent (geopolitical) and 0.52 percent (recessionary) of RWA. - Prudential buffer shortfall 0.91 percent (geopolitical) and 1.44 percent (recessionary) of RWA. - Investment banks: - Do not breach SREP in geopolitical; breach by 0.06 percent of RWA in recessionary. - Prudential buffer deficit modest (up to 0.20 percent of RWA). - Universal banks: - Meet SREP in both scenarios but would need 0.60 percent (geopolitical) to 0.63 percent (recessionary) of RWA to rebuild combined buffers. - Interpretation: - Recessionary shock more severe on bank buffers than the geopolitical shock, increasing buffer shortfalls by roughly 10–60 bp across groups. - Robust NII in a geopolitical scenario offsets some loan losses; deterioration in both NII and borrower creditworthiness amplifies income loss in a severe recession. *Source: IMF staff calculations (content unit: 1eurea2025007).* ### 57. In the baseline, healthy NII and NFCI generation lift system-wide capital with ### 1eurea2025007 - 57. In the baseline, healthy NII and NFCI generation lift system-wide capital with ### Capital dynamics and scenario contributions - Baseline: healthy NII and NFCI generation lift system‑wide capital with moderate market‑risk gains adding a further boost (Figure 11). These benefits are partly offset by credit‑loss charges, operating costs, taxes, and dividends, while gradual growth in RWA trims the ratio at the end. The overall effect is a mild strengthening of capital buffers over the three‑year period. - Geopolitical shock: capital depletion is driven by credit risk. Revenues soften, market‑risk results deteriorate, and decisive erosion comes from higher credit‑risk provisions and a stress‑driven increase in RWAs. CET 1 is pulled several points below the baseline. - Recessionary shock: follows the same pattern as the geopolitical shock but with slightly deeper revenue compression, more moderate credit‑losses, and a milder surge in RWAs. The final CET 1 ratio ends up marginally below that in the geopolitical scenario, marking the weakest capital position of the three paths. - Figure 11 note: Figure 11 is a waterfall chart that tracks how successive incomes and expenses items (along with change in RWA) move the system‑wide CET 1 ratio from its 2024 starting point to its 2027 end‑point under each scenario. - Baseline RWA growth driver: The increase in credit RWAs under the baseline is mainly driven by the treatment of defaulted exposures as not being written off over the horizon, causing defaulted exposures to accumulate and increase in proportion relative to total exposures. Downturn LGD remains stable throughout the baseline while the TTC LGD applied to performing exposures decreases due to favorable conditions, producing relatively higher capital charges for defaulted exposures and a modest increase in risk weight density. ### Profitability over the three‑year horizon - Baseline profitability path: - Peak return on assets of about 0.51 percent of total assets in 2025. - Return on assets around 0.28 percent by 2027, as credit losses remain contained. - Geopolitical scenario profitability path: - Losses of roughly 0.08 percent of assets in 2025. - Losses deepen to about 0.48 percent in 2026. - Profits turn modestly positive again at about 0.10 percent in 2027 as market conditions stabilize and provisioning eases. - Recessionary scenario profitability path: - Loss of roughly 0.13 percent in 2025. - Loss widens to about 0.55 percent in 2026 amid surging credit impairments and compressed margins. - Earnings remain marginally negative around 0.08 percent in 2027, reflecting prolonged drag from a broad‑based downturn. ### Recommendations on stress‑testing and reporting - Embed bespoke templates now deployed for each EU‑wide stress test into the standing supervisory reporting framework alongside FINREP, COREP, and the STE. - Fold ad‑hoc data collections (sensitivities, repricing ladders, granular market‑risk sheets) into the regular schedule to create a single, coherent taxonomy and set of aggregation rules, reducing costly remapping across stress‑test cycles. - Benefits of harmonization: - A continuously updated, stress‑test‑ready dataset enabling more frequent and shorter‑notice scenario runs and desktop‑based stress tests. - Reduced reconciliation and reporting errors and enhanced comparability across institutions and jurisdictions. - Direct routing of harmonized data into the capital‑impact engine so shocks to credit, market, and income variables immediately refresh CET1 trajectories for timely solvency analysis. - Elimination of duplication, lighter reporting burden, faster and more consistent processes, and greater transparency for banks and supervisors. - Stress‑testing design recommendation: - Move beyond one adverse scenario and deploy multiple scenarios, each with its own inflation and interest‑rate trajectory (examples described: stagflation with sharply higher policy rates; rapid disinflation with aggressive monetary easing and a flattening yield curve; sovereign‑spread shock with prolonged low—but volatile—policy rates). - Use multiple scenarios to reveal different projections for NII, funding costs, trading‑book valuations, and stage‑migration dynamics and to expose hidden concentrations (e.g., reliance on sight deposits, large fixed‑rate mortgage books, structurally short option positions). - Consider reverse stress tests to address scenario uncertainty. ### Liquidity stress test — overview and key findings - ECB monetary context: - Since the 2018 EA FSAP, supervisory liquidity risk assessment enhanced and the ECB started quantitative tightening. - Excess reserves declined from a peak in November 2022 (EUR 4.8 trillion) to EUR 2.8 trillion in April 2025 (Hartung et al., 2025). Excess liquidity remained well above the levels observed during the EA FSAP 2018. - ECB operational framework (March 2024): - Standard refinancing operations play a central role with banks able to borrow liquidity at a rate of 15 basis points above the deposit rate, with full allotment against a broad set of collateral. This change aims to reduce market stigma and allow banks to reduce precautionary liquidity buffers above regulatory minimum (100 percent ratios for both the LCR and the NSFR). - Key change: the spread between the rate on MROs and DFR reduced to 15 basis points from the previous spread of 50 basis points (as from 18 September 2024). - Asset allocation and term premia: - ECB asset purchase program reduced term premia for assets included in the program; example: 10‑year term premia for sovereign securities were reduced by 95 basis points on average (Esser et al. 2019). - Asset encumbrance and buffer quality: - Aggregate EA banks’ overnight contractual liquidity gap increased by 4 percentage points to 31 percent of total assets in 2020-2024. - Decrease in the share of stable deposits and an increase in SFTs (repos, collateral swaps), with secured funding from CCPs significant for some banks. - Level of AE is relatively low across all business models: half of banks have an AE ratio of less than 15 percent of total assets. - Banks have encumbered non‑tradeable loan portfolios (almost EUR 0.5 trillion) with the ECB. - Overall, highly rated sovereign bonds constitute almost half of the liquidity buffers. ### Liquidity stress testing approach and results - Analysis coverage and data: - Liquidity risk analysis used data as of end December 2024. - Bank sample ranged from a minimum of 80 banks to a maximum of 102 depending on the test; some segmented tests used 30‑40 banks. - Structural analysis covered LCR (all currencies and USD), funding concentration, AE, collateral swaps and NSFR. - FSAP team did not stress NSFR ratios under alternative scenarios from those prescribed by Basel and focused on cash flow-based stress tests and LCR. - Cash flow‑based liquidity stress tests (CFLST): - Conducted over a wide range of scenarios of increasing severity and central bank support assumptions, covering horizons between overnight and 1‑year using supervisory contractual cash flows by maturity bucket (Annex VI. Table 1 referenced). - Key resilience metrics and stress outcomes: - All banks in the sample comfortably meet LCR and NSFR requirements, with an average LCR close to 190 percent and NSFR close to 140 percent (Figure 15). - Median LCR increased to 192 percent while the lowest was close to 130. - Monthly intra‑quarter LCR volatility is low. - Survival horizons exceed two months for most banks under various stress scenarios, but USD outflow challenges remain. - No large bank moves into negative CBC in mild outflow scenarios within the first month. In a severe outflow scenario including large deposit outflows, a negative CBC occurs in 10 percent of banks within one month. - U.S. dollar cash flow test results: several large banks would face a USD liquidity gap within the first week of stress, but the gap is small, equivalent to 0.5 percent of total assets. - Contingent liquidity risks and amplification channels: - Increased share of secured funding and greater use of collateral swaps may trigger collateral calls. - Exposure to margin calls on derivatives can be substantial and immediate, amplifying liquidity pressures, especially if multiple calls occur across counterparties. - Forced asset sales by stressed NBFIs in a market shock could depress asset prices, increasing valuation haircuts on collateral and reducing effective value of liquid assets; system‑wide average LCR could fall by nearly 50 percentage points yet would remain above the regulatory minimum of 100 percent. - FX funding risks: - USD is the most important foreign currency exposure for EA banks. Very low USD LCRs exist for some banks; investment banks have the lowest USD LCR in the sample, followed by G‑SIBs. - No regulatory requirement to maintain 100 percent LCR by significant currency. - Banks rely on FX swap and spot market functioning and the backstop of ECB/Federal Reserve Board swap lines; ECB conducts weekly 7‑day USD tenders with full allotment but applies a 12 percent exchange rate margin, making auctions more expensive than market swaps and typically resulting in low amounts obtained (e.g., below USD 100 million). ### Structural liquidity risks — LCR and funding structure - LCR levels and behavior: - LCR ratios are significantly above regulatory requirements and have not been affected by the decline in long‑term funding from the ECB. - Median LCR ratio: 192 percent; lowest close to 130. - Banks often target LCRs of 120-130 percent for signaling purposes; intraday liquidity needs could lower regulatory LCR ratios by 2-5 percentage points. - Unweighted median share of HQLA across business groups: between 20 to 30 percent of total assets. - Investment banks have highest HQLA buffers and highest LCR volatility. - In the last five years, through several shocks, the ECB provided up to EUR 3.5 billion of short‑term liquidity to banks (Figure 16). - Funding concentration and short‑term funding: - G‑SIBs have historically recorded the lowest LCR buffers due to larger shares of wholesale funding and optimized liquidity risk management. - Lenders (domestically oriented banks) have higher LCRs because of lower reliance on short‑term wholesale market funding (especially SFTs). - Overnight funding (including non‑maturing items) is close to 32 percent of total funding; some G‑SIBs and large investment banks have an even higher share of short‑term unsecured funding from financial corporations, increasing vulnerability to sudden market‑wide liquidity events. *Source: IMF staff calculations.* ### 75. While the FSAP team did not obtain granular supervisory data on USD funding of ### 1eurea2025007 - 75. ### USD funding, FX swap markets, and USD LCR - ECB analysis highlights USD funding concentration, refinancing risks and counterparty risks in FX swap and repo markets, with majority of trades provided by a few large banks and very short-term (one day) maturity of FX swap markets. - Banks rely on short-term wholesale USD funding to channel USD to EA NBFIs; EA banks receive USD funding from their US affiliates and further lend to NBFIs. In case of USD cash surpluses, these are sold in FX swap markets. - Non-centrally cleared repos to NBFIs involve counterparty risk. - FSAP findings on USD liquidity: - USD LCR is very important to about a third of banks (typically large G-SIBs). - Unweighted median USD LCR = 145. - USD LCR has historically fallen to 0 in several G-SIBs and some investment banks. - Investment banks with very low USD LCR are subsidiaries of US banks and thus have access to Federal Reserve Board or parent bank USD funding facilities (assuming cross-border USD flows are not affected by regulatory ringfencing). ### NSFR, Asset Encumbrance (AE), and funding resilience - Net Stable Funding Ratio (NSFR): - All banks have met the 100 percent NSFR requirement since its inception in 2021. - The ratio has been quite stable, with an unweighted average close to 140 percent through the sample. - Banks tend to target a higher ratio by issuing more bonds with longer than 12-month maturity than needed to meet minimum NSFR requirements. - Asset Encumbrance (AE): - Average AE ratio declined by 6 percentage points since 2021, reaching almost 15 percent of total assets. - Some banks across groups have much higher AE ratios reflecting dominant business models of issuing ABS, covered bonds or dependence on ECB lending facilities. - High AE may shift risk to unsecured creditors and hinder ability to obtain additional market or central bank liquidity (central banks require unencumbered collateral). - Overall, except for a few banks, relatively low levels of AE highlight additional liquidity generation ability in times of systemic and/or idiosyncratic stress. - FSAP cash flow stress tests do not consider additional liquidity generating capacity from pledging eligible claims or own issuances to the ECB, although this may be available under certain conditions. ### Collateral Swaps (CSs) and recent developments - Collateral swaps can redistribute liquidity within the financial system and, if maturity >30 days, increase LCR (if borrowing HQLA) and if >6 months, affect NSFR. - Risks: - Banks relying on constant rollover of CSs could face refinancing issues in market stress (non-HQLA collateral not accepted or haircuts widen). - Market participation and size: - Only 23 banks in the sample are active participants in the CS market. - Banks in the sample are typically net liquidity borrowers (collateral upgrade swaps dominate). - Net amount borrowed increased from 0.02 percent of total assets (end 2022) to 0.2 percent of total assets (end 2024). - Most CSs are in EUR; USD share is very small. - Aggregate implication: - Increased use of CSs may reflect liquidity risk transfer to NBFIs, including insurance companies and pension funds. - FSAP did not have counterparty level data; NBFIs may be CS counterparties and these transactions may not provide resilient liquidity in stress and may increase AE when banks enter collateral upgrade transactions. ### Funding concentration and CCP exposures - Funding concentration varies by business model: - Investment banks have the most concentrated funding sources (share of 10 largest funding providers in total funding), mainly due to being subsidiaries of large foreign-owned groups (mostly USA, UK and in one case CH). - G-SIBs have the lowest concentration. - Median values of funding from the 10 largest funding providers are modest (around 10 percent for lenders and universal banks), with outliers reaching 15 or 25 percent. - Investment banks are most dependent on wholesale, less stable and concentrated sources of funding; they tend to have higher LCR and share of HQLA in total assets. - CCP funding: - Six CCP groups provide almost 3 percent of funding (mostly via repos and deposits) to 29 banks in the sample. - Majority of exposures are repo transactions (94 percent of total exposure to CCPs) with the remainder unsecured. - Some cases where these exposures constitute up to 6 percent of total funding for a bank. ### Contingent liquidity risks and rating downgrade effects - G-SIBs, universal banks and investment banks have higher contingent liquidity risk exposure due to links with other banks and NBFIs and reliance on SFT funding. - SFT and margin risks: - G-SIBs and investment banks rely on a significant share of SFT funding and may face significant margin calls. - LCR data submissions suggest banks estimate contingent liquidity risks from derivatives and other sources are relatively low and do not exceed 2-4 percent of HQLA for G-SIBs and investment banks; however, this may not be conservatively estimated given historical look back approaches. - Own credit rating downgrade risk: - Downgrades may trigger loss of access to unsecured funding, reduction of credit limits, outflows from deposits provided by other banks and NBFIs. - FSAP findings reveal significant heterogeneity: some banks report 0 outflows, some very low in EUR nominal terms, and a minority have substantial outflows when measured in terms of HQLA. - Investment and universal banks report very low values despite interconnections and wholesale funding dependence; uncertainty remains whether foreign parents would provide additional funding under stress. - Recommendation emphasized: - Continue ensuring banks use a range of hypothetical stress scenarios in internal liquidity stress tests, including outflows from market valuation changes on derivatives. ### Cash Flow Liquidity Stress Tests (CFLST) — key results and metrics - Objectives: - CFLST quantifies liquidity risk and risk bearing capacity of EA banks under baseline and multiple stress scenarios, estimating magnitude of potential liquidity needs of individual banks and the system (sample of 95 banks). - CFLST reveals under which circumstances banks would need additional liquidity support due to contractual liquidity gaps and absence of sufficient CBC. - CFLST does not consider redistribution of liquidity within the banking system (migration of deposits among banks). - Contractual liquidity exposure (2024:Q4): - Contractual outflows within the first four weeks amount to 50 percent of total funding (weighted average; excluding open maturity and overnight retail deposits (18 percent of total funding) and open maturity and overnight corporate deposits (32 percent of total funding)). - Contractual inflows amount to about 27 percent of total funding (excluding inflows from central bank deposits (26 percent of total funding)). - Cumulated net funding gap over the first 4 weeks reaches about 31 percent of total funding or EUR 7,500 billion. - Compared to 2020, banks have become more exposed to liquidity risk from changes in value of derivatives and SFTs while inflows from central banks have declined. - Main drivers of net outflows: - (i) Outflows from various types of deposits = 26 percent of total funding. - (ii) SFT (mainly repo outflows) = 9.3 percent of total funding. SFT outflows net of reverse repos would be zero for the same 30-day time horizon; difference arises from maturity mismatch between repo and reverse repo transactions, with overnight inflows 1 percent less than reverse repos. - Counterbalancing Capacity (CBC): - CBC in the first month fully covers the aggregate net cash flow gap. - Cash and withdrawable central bank reserves dominate CBC composition; central bank deposits and 0 percent risk-weight securities account for about two thirds of CBC. - Other non-HQLA items are roughly 30 percent of CBC. - CBC amounts to 21 percent of total assets, higher than the net funding gap (19 percent of total assets over the first four weeks) in the unstressed reported contractual data (excluding retail and operational deposits). - Distribution of CBC is uneven; some banks face shortfalls under stress. - Credit quality and asset composition: - Most securities included in CBC have low to very low credit risk (majority of unencumbered assets fall into CQS1 and Level 1 tradeable assets). - Banks active in FX funding markets accumulated large amounts of U.S. sovereign debt securities and U.S. government-sponsored enterprise issued papers, which fall under CQS1. - Sovereign exposure concentration and maturities: - Five countries (Italy, Spain, Germany, United States, and France) cover roughly 65 percent of sovereign exposures in the sample. - Average remaining maturity of these sovereign securities is close to two years. - CBC composition change (table summary strings as in source): - 20202024Change - Coins and bank notes - 21-1 - Withdrawable central bank reserves - 4935-14 - Level 1 tradable assets - 29323 - Level 2A tradable assets - 220 - Level 2B tradable assets - 220 - Other tradable assets - 1110-1 - Non tradable assets eligible for central - 363 - Own issuances eligible for central banks - 01111 - Undrawn committed facilities received - 31-1 - Total CBC in terms of TA - 2120-1 *Source: IMF staff summary of the FSAP analysis as presented in the supplied content.* ### 93. Most banks are able to absorb liquidity shocks simulated in the two systemic risk ### 1eurea2025007 - 93. Most banks are able to absorb liquidity shocks simulated in the two systemic risk ### Cash flow–based liquidity stress test (CFLST) results - Most banks are able to absorb liquidity shocks simulated in the two systemic risk scenarios (Appendix VI. Table 1). - No large bank fails the mild scenarios (SC 1 and SC 3) within the first month of the stress (Figure 22). - Severe outflows would lead to negative CBC for 10 percent of the banks in the sample within a month. - Under the idiosyncratic risk scenario (SC 2): - The impact of the shock leads to 10 banks failing within a week. - Almost 80 percent of banks failing within 1-year horizon. - Stressed outflows simulated above a 30-day horizon are illustrative only, as no bank could survive for long with massive deposit outflows. - G-SIBs and universal banks suffer the largest net outflows. - Large banks, including G-SIBs and investment banks, have substantial liquidity buffers that help absorb the shocks. - Banks which fail have: - A lower share of CBC. - A higher share of short-term credit and liquidity facilities extended to clients. ### Impact of higher haircuts and collateral composition (SC 3) - In the severe systemic risk scenario (SC 3), applying higher haircuts to lower quality HQLA assets reveals little impact on banks within the first 6 months of stress because banks have enough HQLA CBC at the beginning of the stress even though financial markets are assumed to be not functioning in this scenario. - While the impact on the composition of CBC collateral is visible after several months of stress, the maximum increase in the funding gap is 0.2 percent of total assets. - Once CBC composition shifts towards a lower share of central bank reserves and very HQLA, more banks may face negative CBC at an earlier stage of liquidity stress. ### USD CFLST findings - The impact of CFLST in USD is higher than that in total currencies. - Several large banks would face USD liquidity gaps within the first week of stress, but the gap is small (0.5 percent of total assets). - Most affected banks have very low USD LCR and substantial committed facilities and loan inflows in USD. ### Business-model heterogeneity across scenarios - Scenario 1 (systemic risk): Investment banks are relatively less affected. - Investment banks receive more inflows (including contractual ones) and can scale down their balance sheet during the 30-day stress period. - They do not rely on retail deposits and loans and receive inflows from SFTs, lending to other financial institutions, deposits from central banks, and other short-term facilities; their CBC increases within the 30-day horizon while other banks see declining CBC. - Scenario 2 (idiosyncratic risk): G-SIBs and universal banks are most affected. - Most impact comes from wholesale funding and the outflow of non-operational deposits. - These banks have shorter funding structures to optimize maturity transformation and return on equity. - Lenders (mostly retail banks) are less affected. - Results by business models are similar to those obtained by ECB in 2019 (classification slightly different). ### Reverse CFLST - Reverse CFLST results reveal that the majority of banks can withstand very severe outflows. - Methodology: - Uses severe idiosyncratic scenario 2 (full CBC) as starting point and assumes maximum severity reaches twice the magnitude of the initial shock (except when the shock is 0 or 100 percent). - Pass/fail threshold: size of banks with negative CBC reaches 20 percent of total assets. - Number of discrete non-linear convex steps for increase in stress parameters: 20. - Findings: - Banks that first breach negative CBC limits have large contingent outflows and low CBC to total assets ratio. - Banks individually can withstand significant outflows before the 20 percent total assets of banks with negative CBC is reached. - Only after the 13th iteration the systemwide CBC becomes negative. - The threshold of 20 percent of failing banks assets in terms of total assets is reached after 4th iteration due to some large banks with negative CBC, yet CBC shortfall of liquidity is modest at 1 percent of total assets. ### LCR stress tests and sensitivity analysis - LCR stress tests used outflow parameters from cash flow severe idiosyncratic stress scenario 2 to test falling below targeted minimum (120 percent) and minimum (100 percent) ratios; full HQLA assumed, with haircuts based on the geopolitical solvency scenario (no additional haircuts). - Results: - Impact on aggregate LCR would be significant: the median LCR declines by 80 percentage points with many banks falling below the minimum ratio and some falling close to 0. - G-SIBs and universal banks are most affected; investment banks and lenders maintain a median LCR close to 100. - Stressed haircuts from the NBFI system-wide stress test were applied to HQLA in the LCR using two horizons: 2 days and two-week shocks. - LCR sensitivity analysis used aggregated (by HQLA category) haircuts with two assumptions: i) average per HQLA category; and ii) maximum per HQLA category. - Applying LCR-based parameters to inflows and outflows and shocking just HQLA leads to a modest decline in LCR with little difference between average and maximum haircuts. - Larger relative impact on G-SIBs and universal banks. - Overall impact: - Multiple banks would fall below the 120 percent LCR target, and some banks would fall below the 100 percent LCR requirement. - Adding the impact of forced sales by stressed NBFIs on valuation haircuts to the LCR scenario, the aggregate LCR could decline by almost 50 percentage points. - Little difference between results using 2 day and two-week haircuts. ### Recommendations on liquidity risk analysis - ECB and SSM could enhance liquidity risk analysis by taking a forward-looking view to measure contingent liquidity risks from market valuation changes in SFTs, derivatives, and other market exposures. - Key liquidity risks to monitor include contingent outflows and unexpected flows from interlinkages between banks and NBFIs. - Under stressed market conditions, outflows may exceed those calibrated in the LCR regulation, which is backward-looking. - Ongoing efforts to quantify CCR and liquidity risks under stressed market conditions should continue, involving: - Use of counterparty-level data. - Estimation of amplification effects as they propagate through the network of exposures (as outlined by Barbieri et al. (2025)). - Internal liquidity risk analysis by SSM using updated stress parameters could guide banks on quantifying outflows from own credit rating downgrades by defining types of outflows and minimum outflow rates to address shortcomings observed in the 2019 ECB Sensitivity Analysis of Liquidity Risk. - SSM supervisors in liquidity profile analysis assess reduction of secured and unsecured wholesale funding, correlation between funding markets and diversification, contingent off-balance sheet exposures, FX convertibility and access to foreign exchange markets, estimates of future balance-sheet growth, and implicit rollover/extension of liquidity support. - When assessing liquidity and funding risks as part of SREP, own rating downgrade outflows are typically based on collateral outflows or additional margin calls which may be low in banks with a small trading book or limited covered bond issuance; however, a rating downgrade may affect wholesale funding including outflows from operational deposits. - Recalibrate outflow rates using recent events in the USA and Switzerland to harmonize assessment of own credit rating downgrade impact in SSM SREP and ECB systemwide liquidity risk analysis. - FX funding risks: - Better measured by harmonizing granular reporting templates by counterparty, collateral and concentration. - LCR (Corep) collects FX funding data with some additional Finrep data, but counterparty-level data collection is not harmonized beyond LCR FX funding concentration templates. - Harmonize collection of granular supervisory data by counterparty, rating, type of transaction, and collateral using a proportional approach. - Incorporating this in Implementing Technical Standards (ITS) reporting would be time-consuming; integrating it into the STE could be more efficient. ### Solvency–liquidity interactions and business risk - Motivation: - March 2023 banking turmoil questioned timeliness of regulatory liquidity ratios in detecting early signs of stress; BIS (2024) noted liquidity supervision might need to evolve considering banks’ business models and high-frequency data. - Need to consider banks’ ability to monetize liquid assets and interactions with concentration risks and capital position. - Monitoring liquidity risks at a shorter horizon than the 30-day LCR may be important; a shorter horizon, 2-days or 2-weeks, may be appropriate given potential margin and collateral calls amid acute market shocks. - The failure of Credit Suisse highlighted business model sustainability as a trigger for massive outflows; business risk played a key role in the failure. - SNB (2023) reports a decrease in capital market business of investment banking of Credit Suisse of over 70 percent in 2022, while transaction-based income in wealth management dropped by over 20 percent. This contributed to negative income in 2023 (CHF 7.2 billion). - Credit Suisse incurred a downgrade in November 2022 that increased borrowing costs and limited access to short-term funding markets. - In March 2021, several global banks incurred large losses due to Archegos; Credit Suisse suffered the biggest loss, amounting to over USD 5 billion. - No capital requirements associated to business risk in Pillar 1 capital ratios; SSM conducts business model assessments in SREP but stress testing practices do not include business risk when assessing bank resilience. - Modeling approach: - FSAP conducted a joint stress testing framework for solvency and liquidity, taking a forward-looking view of liquidity risk, business risk and CCR. - Liquidity and solvency are strongly interrelated; many liquidity shocks arise endogenously from balance sheet positions or concerns over business model sustainability. - The framework integrates interaction and amplification mechanisms using a two-week and a two-day scenario respectively, testing bank resilience to combined solvency risk and endogenous liquidity risk in a market shock scenario, accounting for defaults from vulnerable counterparties in derivatives transactions. - The model links market shocks with LCR and NBFI CCR for an overarching and consistent view. - The analysis is conducted on the seven EA G-SIBs using supervisory data as of end-2024. *Sources: ECB and IMF staff calculations.* ### 109. Figure 30 illustrates the interplay between solvency and liquidity risk, based on the ### 1eurea2025007 - 109. Figure 30 illustrates the interplay between solvency and liquidity risk, based on the ### Overview - Figure 30 and accompanying text illustrate the interplay between solvency and liquidity risk using the Cont, Kotlicki, and Valderrama (2020) framework. - The stress scenario involves shocks to interest rates, credit spreads, and equity prices across markets, producing fair value losses, counterparty credit risk (CCR) losses, margin and collateral calls, deposit run-off, and potential downgrades that feed back into funding and liquidity pressures. ### Solvency–Liquidity Transmission Channels (mechanics) - Solvency is hit first through fair value losses (purple dotted area). - Defaults by vulnerable derivative counterparties trigger CCR losses (purple lined area), further eroding profitability. - Market shocks reduce the value of derivatives and collateral, prompting margin and collateral calls. - Liquidity pressure intensifies as (non-maturing) deposits run off (at the run-off rates prescribed in the LCR), adding to contractual net cash outflows (systematic factor). - Initial capital loss raises leverage, increasing downgrade risk; a downgrade can cut off unsecured funding, weaken investor confidence and trigger further deposit outflows (idiosyncratic factor), amplifying stress. ### Modeling approach and balance sheet representation - The modeling approach requires a granular balance sheet (Figure 32): assets broken down by vulnerability to market shocks and liquidity risk, off-balance-sheet items (e.g., AuM), income by business line (capital-based, AM, WM), and liabilities split into contractual flows, behavioral flows (non-maturing deposits), and contingent flows (derivatives, SFTs, downgrade-linked). - Market shocks include shifts in benchmark interest rates in EUR and USD across tenors, credit spreads on sovereign/corporate bonds in Europe and the US, and equity prices (major indices in Europe, US, Asia, and Latin America). - Changes in market valuation are computed using a partial revaluation approach (see paragraph 33 in source). ### Counterparty Credit Risk (CCR) quantification across horizons - Two-week stress exercise: CCR losses driven by default of the three most vulnerable counterparties among the top 20 derivatives exposures (net of credit risk mitigation), excluding qualifying CCPs, central banks, central governments, and international financial institutions. - Two-day scenario: CCR linked to endogenously determined NBFI counterparty defaults projected in the system-wide liquidity analysis. - Appendix VII describes technical aspects of the model. ### Liquidity flows, LaR, and mitigating actions - Obligations coming due at the end of the horizon include: - Unconditional flows: maturing liabilities and contractual outflows; expected outflows for non-maturing liabilities based on LCR calibration. - Contingent flows linked to margin and collateral calls: for margined derivatives and SFT transactions, flows linked to negative changes in fair value. - Contingent flows triggered by own credit downgrade: rating downgrade generates run-off of credit sensitive deposits (e.g., high value deposits). - Liquidity reserves increase with scheduled contractual inflows. - Liquidity at risk (LaR) measures net outflows corresponding to the stress scenario considered; LaR is conditional on the scenario, including evolution of liquidity balances and maturing liabilities. - Mitigating actions a bank can employ quickly (but at cost): - Unsecuritized borrowing in the interbank market at overnight market rate conditional on not being downgraded. - Repo borrowing with central bank or market counterparty, requiring unencumbered eligible assets. - Liquidation of assets (“fire sales”) — a fraction of illiquid assets sold in the short term at a market discount. - Mitigating actions raise liquidity but are costly (borrowing costs or forced asset sales), further eroding capital (pink lined area in Figure 30). A weak capital position may prevent access to short-term funding markets; shocks to fair value and haircuts limit monetizable liquidity. ### Parameterization and key numeric inputs - Model inputs require: granular balance sheet data, risk parameters (market risk sensitivities, model risk adjustments, CCR exposures), income breakdown by source, and liquidity data (CBC by instrument, asset encumbrance, contractual cash inflows/outflows by type/time bucket). - Other inputs calibrated to financial conditions and regulatory environment prevailing at end-2024. Specific parameter values reported in Table 5: - Unsecuritized rate (interbank): 2.91 - Repo rate (market): 3.00 - Rep rate (CB): 3.15 - Effective repo haircut (bank specific): 5.0-9.7 - Fire sales discount: 50.0 - Fraction of illiquid assets eligible for sale: 5.0 - Downgrade threshold (bank specific): 4.10-4.35 - Insolvency threshold (bank specific): 3.60-3.85 - Additional calibrations and mappings: - Repo haircut mapping follows Eurosystem monetary policy framework (ECB, 2022). - Fraction of illiquid assets and fire sale discount follow Cont et al. (2020). - Downgrade threshold = 50-basis points buffer over bank-specific minimum regulatory leverage requirements (“distance to default buffer”). - Share of credit sensitive outflows calibrated by applying LCR run-off rates to non-maturing deposits and contractual outflows maturing within scenario horizon. - Insolvency trigger: breach of regulatory leverage requirement including Pillar 1, Pillar 2R, and G-SIB add-on. - Illiquidity failure: maturing liabilities and liquidity demands exceed bank’s capacity to raise liquidity. ### Scenario set used in two-week exercise (to address model uncertainty) - Reference scenario: geopolitical market shock from solvency stress test (Annex III. Table 5) and parameterization described above. - “Credit sensitive” scenario: - Run-off rate for “deposits subject to higher outflows” increases to 20 percent (from the 13 and 18 percent rates prescribed in CRR and assumed in reference). - For operational deposits not covered by deposit guarantee scheme, run-off increases to 40 percent from 25 percent in CRR. - Counterparties trigger initial margin (IM) calls by 20 percent of the bank IM collateral posted in margined transactions. - “Business as usual” scenario: maturing loans rolled over during stress horizon (reflecting observed behavior in stressed banks). - Narrow collateral scenario: eligible credit claims and eligible own issuances become ineligible. - “Trapped liquidity” scenario: collateral in foreign currency becomes unavailable following a downgrade (proxy for intra-group transfer frictions). - Business risk scenario: client attrition in IB and AM when leverage exceeds downgrade trigger, reducing fee income from capital-based and AM by 30 percent. ### CCR alternative methods in two-day scenario - “No CCR” scenario: market shock triggers portfolio valuation losses but excludes CCR losses. - “CCR measured as outright defaults” scenario: three most vulnerable counterparties among 20 largest exposures default (losses net of credit risk mitigation), excluding qualifying CCPs, central banks, central governments, and international financial institutions. - “CCR estimated from NBFI risks” scenario: bank-specific default rate on margined transactions estimated using transaction-level data on derivatives and SFTs; average default rate applied to each bank’s stressed EEPE. - “CCR estimated from NBFI defaults and Market Volatility” scenario: combines CCR from NBFI defaults with widening of credit spreads via raising borrowing rates by 2 percentage points. ### Reverse stress testing and uncertainty analysis - Reverse stress testing: generate range of adverse scenarios by varying severity of shocks to risk factors; identify critical amplitudes that can lead to insolvency or illiquidity. Valuation of balance sheet components under each scenario assumed linear scaling of risk factor impacts. - Representative shock ranges used in reverse stress tests: interest rate shocks up to 7 percent (y axis) and equity price shocks up to 50 percent (x axis). - Reverse stress test outcome regions (Figures 33–37): - Light grey: bank solvent, liquid, no borrowing action. - Dark grey: bank solvent, liquid, with borrowing action. - Light orange: bank liquid, insolvent due to borrowing costs. - Dark orange: bank liquid, insolvent due to size of shock. ### Results and key insights - Speed of solvency–liquidity spiral increases with correlation of shocks; G-SIBs stand out given interconnections via derivatives, SFTs, and exposures to institutional, credit-sensitive counterparties including NBFIs. - Diversification of income sources (AM/WM/trading) that cushions shocks in normal times can introduce correlations across balance sheet funding structures and off-balance-sheet items in severe market scenarios, creating correlation risk and potential “death spiral.” - Correlation risk across profits, equity, leverage, and liquidity is not included in Basel capital/liquidity requirements (though expected to be assessed as part of the SREP and may result in Pillar 2 measures). - Amplification effects from the solvency–liquidity nexus are highly non-linear and can exceed 100 percent of the initial market shock, with largest impact around the downgrade trigger. - Reverse stress tests suggest intermediate shocks may trigger insolvency through interaction of solvency and liquidity rather than a single channel alone. *Source: IMF staff elaboration (content unit 1eurea2025007 - paragraphs 109–124 and Table 5).* ### Appendix III. Table 5 which are scaled linearly with moves to the plotted factors. ### Appendix III. Table 5 which are scaled linearly with moves to the plotted factors. ### Scenario findings and failure-region shifts - Combining a “credit sensitive” scenario with “business as usual” dynamics expands failure regions: the failure region moves “south-west” under the “credit sensitive” scenario and shifts further toward the origin when combined with “business as usual”. - Example thresholds (reference vs combined scenarios): - Interest shock: 4 percent (reference) → 3.5 percent (credit sensitive + business as usual) - Equity shock: + 50 percent (reference) → 45 percent (credit sensitive + business as usual) - Failure-region axes and shock ranges used across charts: - Benchmark market rate shocks range between 0 and 7 percent (EUR 1 year OIS; y axis). - Reference equity index (EQ-Europe) shocks range between 0 and 50 percent (x axis). ### Liquidity generation under alternative assumptions - Reference scenario liquidity available: EUR 900 billion. - Narrow Eurosystem collateral framework: maximum liquidity generation reaches EUR 750 billion. - “Trapped liquidity” (FX collateral cannot be monetized under stress): maximum liquidity generation EUR 425 billion. - In the reference solvency-liquidity example: - Equity drops from EUR 285.3 to 221.3 billion under the market risk scenario. - Liquidity at risk stands at EUR 903.7 billion. - Amount of liquidity borrowed reaches EUR 400 billion. ### Business risk, credit-sensitive flows, and amplification - “Business risk” effects on solvency resemble “credit sensitive” flows: liquidity shocks and profitability shocks can have comparable impacts on solvency risk. - Policy implication: projecting P&L items in solvency stress tests while ignoring liquidity pressures can underestimate banks’ resilience. - Specific scenario parameter changes: - “Credit sensitive” scenario adjustments: - Run-off rate of category 1 and 2 deposits increases to 20 percent (from 13, and 18, percent in the regulatory framework, respectively). - Operational deposits not covered by DGS are subject to a 40 percent (rather than 25 percent) run-off rate. - Initial margin (IM) on collateral posted in margined transactions increases by 20 percent. - “Business-as-usual” scenario: maturing loans are rolled over. - “Business risk” scenario: client attrition reduces fee and commission income by 30 percent, conditional on a credit downgrade. ### Counterparty credit risk (CCR), NBFIs, and market volatility (two-day horizon) - Without CCR losses, even severe market and liquidity shocks are unlikely to cause insolvency in the two-day scenario. - CCR inclusion materially raises default risk: - CCR defined as outright defaults of the three most vulnerable counterparties among the largest twenty exposures increases solvency risk notably. - CCR defined as NBFIs failing to meet margin calls (TN “Systemic Risk Analysis – NBFIs”) expands failure regions significantly, especially under high interest rates and market volatility. - Amplification measures: - Liquidity costs could amplify the initial market shock up to 150 percent (observed in the “No CCR” vs amplification analysis). - Adding wider credit spreads (two percentage points) from heightened market volatility significantly increases credit risk and amplification effects for moderate shocks. - CCR loss offsets caveat: CCR losses may be partially offset by the initial margin posted by NBFIs, provided collateral is excluded from EEPE exposure. ### Recommendations (policy guidance distilled from Appendix III) - Joint solvency-liquidity stress tests for G-SIBs are important to capture amplification risks from market shocks, confidence loss, and CCR. - Under severe but plausible shocks, failure risk rises due to higher borrowing costs from endogenous liquidity stress and franchise-driven business risk. - Forward-looking liquidity risk assessment and robust CCR measurement are key to assessing bank resilience: - Enhance LCR ratios with a conditional, granular “liquidity-at-risk” metric—based on balance sheet composition and behavioral flows—to detect vulnerabilities before outflows occur. - Include CCR losses, especially from vulnerable NBFIs, to obtain a more realistic view of resilience. - Authorities could build on recent CCR stress testing efforts to quantify endogenously NBFI-driven CCR losses under a common scenario; this requires a system-wide perspective to identify intersectoral funding and liquidity dependencies at entity and transaction level. - Reverse stress testing recommendations: - Calibrate shock severity to identify critical thresholds that could lead to insolvency. - Incorporate uncertainty around key parameters and behavioral responses to address model risk and make stress-testing outputs more policy-relevant amid cyclical volatility and structural change. ### Network analysis: scope, data, and scenarios - Scope and dataset: - Coverage: 72 SIs (out of 109) at the highest level of consolidation across 17 countries. - Representing about 90 percent of EA banking system assets as of June 2024. - Interbank network composition: 61 lenders and 50 counterparties with a total of 377 exposures (edges). - Data sources: ECB’s COREP and FINREP supervisory data, including large exposure data reporting. - Modeling framework: - Contagion model based on the CoMap framework (Covi, Gorpe, and Kok, 2021), using simulation to quantify knock-on effects from a hypothetical default through the interbank network. - Key baseline calibration parameters include loss given default, funding shortfall rate, net liquidity position, pool of assets available for fire sales, discount rate, and hurdle rate (details in Appendix VIII). - Alternative scenarios for network contagion: - Baseline (“Reference Scenario”): parameters calibrated broadly in line with Covi, Gorpe, and Kok (2021). - NBFI Risk Scenario: - CET1 capital is depleted to absorb losses from the defaults of the banks’ top five NBFI credit exposures (insurance undertakings and pension funding excluded). - Produces a stylized scenario with lower initial bank capital than baseline. - Market Risk Scenario (combined with NBFI risks): - In addition to lower initial capital as in the NBFI risk scenario, apply: - Haircut of 10 percent on HQLA. - Haircut of 50 percent on credit risk mitigation instruments. - Results in lower liquidity surplus (net liquidity positions), higher net exposures, and higher loss given default than baseline. - Distributions illustrated (figures referenced in Appendix III): - Figure 38: Distribution of CET1 capital across baseline and NBFI risk scenario. - Figure 39: Distribution of net liquidity positions and net exposures across baseline and market risk scenario combined with NBFI risks. *Source: IMF staff calculations.* ### 138. Bank network analysis facilitates the assessment of each bank’s potential to propagate ### 1eurea2025007 - 138. Bank network analysis facilitates the assessment of each bank’s potential to propagate ### Methodology and Indices - Interbank analysis uses a series of hypothetical defaults where the default of each entity is triggered iteratively to obtain: - the number of defaults caused by the “trigger event”; and - the losses incurred by each entity in the network in each simulation. - Following Covi, Gorpe, and Kok (2021), two indices are constructed from simulation losses to rank banks: - Contagion index (CI): system-wide losses induced by bank i in percent of total capital in the system (excluding bank i). - Vulnerability index (VI): average loss experienced by bank i across all simulations in percent of its own capital. - Results are visualized in systemic risk maps (scatter plots) combining CI and VI; median and 75th percentile lines are used to identify banks that are highly systemic and highly vulnerable (north-east quadrant). ### Baseline Analysis Findings - Simulation scope: 72 trigger events. - Top 10 hypothetical default events (baseline, Table 6 panel A): - These top ten events induce, on average, 1.3 percent of capital losses to the euro area banking system. - Contagion transmitted entirely through credit losses in the baseline. - Exogenous defaults do not lead to any additional defaults in the network under baseline calibration. - Comparison to Covi, Gorpe, and Kok (2021): - Their top 10 events induced on average 2.5 percent of capital losses (2017Q3 data), comparable to the most contagious event in the current analysis. - Bank-level distributions and concentrations: - Two banks identified with CI and VI scores above the 75th percentile. - Six banks with high CI scores. - Three entities with high VI scores. - G-SIBs: potential to induce high system-wide losses but do not appear vulnerable to shocks from within the euro area banking system. - Highly vulnerable banks tend to be classified as universal or lenders. - French banks dominate the group of the most contagious banks; Italian entities form the major share among the most vulnerable banks. ### NBFI Risk Scenario (stylized) - Assumption: outright default of the five largest NBFIs (exploratory/stylized scenario). - Key outcomes (Table 6 panel B): - Three events among the top 10 hypothetical defaults (events ranked 1, 2, and 6) result in one additional default in the network, respectively. - The most contagious event leads to one additional default and 3.1 percent of capital losses. - The top ten exogenous default events, on average, induce 1.5 percent of capital losses to the system. - Losses transmitted entirely through the credit channel. - Caveat: simulations underestimate capital losses because initial conditions were calibrated to lower banks’ capital surplus, resulting in lower total capital in the network. - Disaggregated results by business model and domicile are broadly comparable to baseline findings. ### Market Risk Scenario (combined with NBFI risks) - Scenario description: NBFI risks combined with increased market volatility. - Key outcomes (Table 6 panel C): - Exogenous defaults of seven banks (events ranked 1-8, except 4) could induce eleven additional defaults in total. - On average, these events trigger 1.9 percent of capital losses to the euro area banking system. - The most contagious event under this scenario leads to three additional defaults and 4.4 percent of capital losses. - Channel composition: losses due to credit risk account for the entirety of the losses. - Defaults are entirely triggered due to insolvency rather than illiquidity (contrast with Covi, Gorpe, and Kok (2021) and Aikman et al. (2018)); this may reflect ample excess liquidity resulting from Eurosystem measures during and after the COVID-19 pandemic. - Disaggregated results remain robust. ### Key Quantitative Points - Number of trigger events simulated: 72 - Baseline: top 10 events → average 1.3 percent of capital losses; no additional defaults. - Covi, Gorpe, and Kok (2021) (2017Q3): top 10 events → average 2.5 percent of capital losses. - NBFI scenario: most contagious event → 3.1 percent of capital losses; top ten average → 1.5 percent. - Market risk + NBFI scenario: seven exogenous defaults could induce eleven additional defaults; average losses → 1.9 percent; worst event → 4.4 percent and three additional defaults. - Panel A (Table 6): top contagion index values include 2.5, 2.2, 1.8, 1.8, 1.2, 1.0, 0.8, 0.6, 0.6, 0.5 (with corresponding VI and total defaults as reported). - Panel B (Table 6): top contagion index values include 3.1, 2.9, 2.0, 2.0, 1.3, 1.2, 0.9, 0.7, 0.6, 0.6. - Panel C (Table 6): top contagion index values include 4.4, 3.4, 2.3, 2.2, 1.7, 1.3, 1.3, 1.1, 0.7, 0.6. ### Recommendations and Policy Implications - Current assessment: the risk of contagion through interbank exposures within the euro area is currently low; robust capital and liquidity positions underpin interbank resilience. - In the baseline simulation, a bank’s failure will not trigger cascading defaults through the interbank network. - The average size of losses of the top 10 events in baseline = 1.3 percent of aggregate capital. - Key risks and vulnerabilities to monitor: - NBFIs and market volatility are significant amplifiers of bank stress: a failure of the most systemic bank could trigger 4.4 percent capital losses and three cascading defaults. - The interbank analysis likely understates contagion risk because it excludes: - banks’ exposures to NBFIs (except stylized scenario), - counterparty credit risk (CCR) from derivatives and SFTs, - liquidity-related borrowing costs, and - capital impacts from market volatility. - The finding that contagion stems from insolvency—not illiquidity—should be viewed cautiously because ECB balance sheet unwinding may reduce excess liquidity. - Recommended actions: - Strengthen monitoring of bank–NBFI linkages using granular data. - Run network simulations that incorporate broader exposures and channels (including CCR, SFTs, liquidity dynamics, and market-to-market capital impacts), as shown by Barbieri et al. (2025). *Source: IMF staff analysis in "Euro Area" chapter (Bank network analysis), based on ECB and IMF staff calculations.* ### Appendix I. Risk Assessment Matrix (RAM) ### Appendix I. Risk Assessment Matrix (RAM) ### Global Risks — key findings and policy responses - Trade policy and investment shocks - Likelihood of Risk: High - Expected Impact of Risk: High - Expected impact description: Higher trade barriers or sanctions reduce external trade, disrupt FDI and supply chains, and trigger further U.S. dollar appreciation, tighter financial conditions, and higher inflation. Weaker export growth, combined with higher uncertainty and weaker consumer and business confidence, weighs on the corporate sector and result in lower investment and a slower recovery in private consumption, ultimately undermining productivity and lowering potential output. - Policy responses: - Continue advocating for a stable, rules-based global trading system and pursuing constructive engagement. - Ensure consistency with WTO principles in the use of targeted instruments (e.g., safeguard procedures and anti-dumping, anti-subsidy, and anti-coercion measures). - Diversify global partnerships and advance new free trade agreements. - Deepen single market and avoid industrial policy that creates distortions or provokes retaliation. - Deepening Geoeconomic Fragmentation - Likelihood of Risk: High - Expected Impact of Risk: High - Expected impact description: Persistent conflicts, inward-oriented policies, protectionism, weaker international cooperation, labor mobility curbs, and fracturing technological and payments systems lead to higher input costs, hinder green transition, and lower trade and potential growth. Trade barriers and supply disruptions lead to shortages in crucial inputs, higher inflation and production bottlenecks that reduce economic activity and decrease confidence. - Policy responses: - Diversify energy production and secure supply chains to avoid shortages of critical raw materials. - Diversify global partnerships and advance new free trade agreements. - Continue advocating for a stable, rules-based global trading system and pursuing de-escalation and constructive engagement. - Ensure consistency with WTO principles in the use of targeted instruments (e.g., safeguard procedures and anti-dumping, anti-subsidy, and anti-coercion measures). - Tighter financial conditions and systemic instability - Likelihood of Risk: Medium - Expected Impact of Risk: Medium - Expected impact description: Higher-for-longer interest rates and term premia amid looser financial regulation, rising investments in cryptocurrencies, and higher trade barriers trigger asset repricing, market dislocations, weak bank and NBFI distress, and further U.S. dollar appreciation, which widens global imbalances and worsens debt affordability. Higher funding costs and a shift in risk sentiment lead to bond repricing and financial tightening, reducing credit growth. Insolvencies increase, resulting in deterioration of bank balance sheets and profitability. Rates staying high for longer will also lead to housing market corrections. Sovereign spreads increase, straining fiscal sustainability in high-debt countries. - Policy responses: - Enhance liquidity support to financial institutions and markets to avoid contagion and prevent liquidity shortages morph into insolvencies. - Ensure strong coordination between the ECB and the national authorities on financial stability risks. - Use countercyclical financial policy to support viable financial institutions. - Regional Conflict - Likelihood of Risk: Medium - Expected Impact of Risk: Medium - Expected impact description: Intensification of conflicts (e.g., in the Middle East, Ukraine, Sahel, and East Africa) or terrorism disrupt trade in energy and food, tourism, supply chains, remittances, FDI and financial flows, payment systems, and increase refugee flows. Increased uncertainty weakens consumer and business confidence, reducing consumption and investment. Spikes in energy prices and supply disruption reduce competitiveness and the purchasing power of households. - Policy responses: - Accelerate the energy transition. - Provide targeted support to vulnerable households to mitigate the impact if risks materialize. - Commodity Price Volatility - Likelihood of Risk: Medium - Expected Impact of Risk: Medium - Expected impact description: Supply and demand volatility (due to conflicts, trade restrictions, OPEC+ decisions, AE energy policies, or green transition) increases commodity price volatility, external and fiscal pressures, social discontent, and economic instability. Higher commodity import prices lead to higher energy prices that fuel inflationary pressures. Export competitiveness of European firms is adversely affected which in turn slows down activity. High energy prices have an adverse impact on households, leading to lower domestic demand. - Policy responses: - Maintain monetary policy flexibility. - Allow automatic stabilizers to operate and provide fiscal support to vulnerable households. - Safeguard energy security by accelerating the green transition and electricity market integration. - Provide targeted support to vulnerable households to mitigate the impact of higher energy prices. - Cyberthreats - Likelihood of Risk: Medium - Expected Impact of Risk: Medium - Expected impact description: Cyberattacks on physical or digital infrastructure (including digital currency and crypto assets), technical failures, or misuse of AI technologies trigger financial and economic instability. Depending on the country level of digitalization and exposure to digital infrastructure, cyberattacks disrupt the financial system as well as the real economy. - Policy responses: - Advance crisis preparedness to cyberattacks. - Further strengthen coordination at the European/international level. - Strengthen the operational resilience of the financial system. - Climate Change - Likelihood of Risk: Medium - Expected Impact of Risk: Medium - Expected impact description: Extreme climate events driven by rising temperatures cause loss of life, damage to infrastructure, supply disruptions, lower growth, and financial instability. Productivity declines or shortages lead to price increases. EU members may receive migrants from economies facing severe climate disruptions. - Policy responses: - Build fiscal space that can be used in response to large climate shocks. - Enhance the EU budget to invest efficiently to mitigate climate risks and flexibly respond to extreme climate events. - Accelerate green transition. - Global growth acceleration - Likelihood of Risk: Low - Expected Impact of Risk: Medium - Expected impact description: Easing of conflicts, positive supply-side surprises (e.g., oil production shocks), productivity gains from AI, or structural reforms raise global demand and trade. Higher export growth, combined with stronger consumer and business confidence, supports the corporate sector and results in higher investment, lower unemployment, and a faster recovery in private consumption. Higher growth leads to an improvement in public debt sustainability in some high-debt countries. - Policy responses: - Allow automatic stabilizers to operate and accelerate fiscal consolidation to rebuild buffer. - Promote high quality public investment in infrastructure, and advance structural reforms. - Diversify global partnerships and advance new free trade agreements. ### Euro Area Domestic Risks — key findings and policy responses - Disorderly energy transition - Likelihood of Risk: Medium - Expected Impact of Risk: Medium - Expected impact description: A disorderly shift to net-zero emissions (e.g., owing to shortages in critical metals) and climate policy uncertainty cause supply disruptions, stranded assets, market volatility, and subdued investment and growth. Higher energy prices lead to higher inflation and decreased real incomes. Increased climate policy uncertainty lowers investments in green technology. - Policy responses: - Provide temporary, targeted fiscal policy support to households and businesses severely affected by energy transition. - Promote public investment and accelerate structural reforms to improve energy efficiency and facilitate labor reallocation with active labor market policies. - Higher defense spending - Likelihood of Risk: Medium - Expected Impact of Risk: Medium - Expected impact description: New NATO commitments or a lower-than-expected efficiency of additional defense spending could result in higher than anticipated defense spending. Higher defense spending supports growth but raises concerns about public sector debt sustainability and raises interest rates. - Policy responses: - Limit the use of the national escape of the EU fiscal rules clause to the initial phase of scaling up defense investment expenditures. - Assess the consequence of increased defense spending on debt sustainability. - Closely monitor efficiency of additional defense spending. - Populism and Polarization - Likelihood of Risk: Medium - Expected Impact of Risk: Medium - Expected impact description: Real income loss, spillovers from conflicts, dissatisfaction with migration, and worsening inequality ignite populism, polarization, and resistance to reforms. Delayed and suboptimal policies weaken confidence and raise uncertainty, lowering growth and leading to market repricing. Delayed fiscal adjustment weakens fiscal sustainability and increases sovereign risks. - Policy responses: - Increase growth and productivity, and ensure benefits are shared widely. - Ensure that increased defense spending and fiscal consolidation do not undermine targeted social spending or exacerbate inequality. - Provide temporary support to vulnerable households if needed. - Realization of Financial Sector Vulnerabilities - Likelihood of Risk: Low - Expected Impact of Risk: High - Expected impact description: A shift in market perception undermines the ability to roll over and service debt, re-igniting financial fragmentation and adversely affecting the banking system. NBFIs could amplify risk propagation in the banking sector and system-wide spillovers from investment fund distress. Higher funding costs and a shift in risk sentiment lead to bond repricing and financial tightening, reducing credit growth. Insolvencies increase, resulting in deterioration of bank balance sheets and profitability. - Policy responses: - Enhance liquidity support to financial institutions and markets to avoid contagion and prevent liquidity shortages morph into insolvencies. - Ensure strong coordination between the ECB and the national authorities on financial stability risks. - Use countercyclical financial policy to support viable financial institutions. - Rely on bank resolution systems to address unsound banks. - Enhance system-wide monitoring and improving data sharing. - Shifting sentiment on countries with high public debt - Likelihood of Risk: Low - Expected Impact of Risk: High - Expected impact description: Policy slippages with weak growth outturns in some high-debt euro area countries, along with weak trust in the Governance Framework, could raise concerns over debt sustainability in high debt countries. Sharp increases in funding costs strain high-debt countries’ ability to service their debt resulting in adverse real-financial feedback loops and financial fragmentation that weighs on economic activity and impairs monetary policy transmission. - Policy responses: - Activate EU support lines for high-debt countries under stress. - Make use of the transmission protection instrument (TPI) if higher spreads are not based on fundamentals. - Enhance liquidity support to financial institutions and markets with strong coordination between the ECB and the national authorities on financial stability risks. --- ### A. Banking Sector: Solvency Stress Test — Top-down by IMF ### 1. Institutional Perimeter - Institutions included: 95 SIs (out of 109 SIs), of which 7 are G-SIBs. - Market share: About 99 percent of the banking sector assets. - Data and baseline date: - Data vintage: 2024:Q4. - Supervisory data: Bank balance sheet and supervisory statistics (including FINREP and COREP), information on IRRBB, short-term exercise (STE), provided by the ECB. Expected Default Frequency sourced from Moody’s. - Household analysis relies on household survey microdata from the 2021 (latest) HFCS survey, covering 83,000 households across 22 countries (EA, CZ, and HU) and 200,000 personal files. Montecarlo simulations of unemployment shocks at the person level. Projections of households’ balance sheets, consumption, and debt repayments, allowing for new issuances of maturing loans. - Market and publicly available data, such as information from ECB statistical data warehouse on funding and lending rates for new business by type of asset and funding portfolios, complemented with commercial databases such as Capital IQ. Corporate sector analysis uses data from Orbis. - Scope of consolidation: banking activities of the consolidated banking group for banks having their headquarters in the euro area. - Coverage of sovereign and non-sovereign securities exposures: debt securities measured through fair value (FVPL and FVOCI) and amortized cost (AC) account. ### 2. Channels of Risk Propagation and Methodology - Methodology: - FSAP team satellite models and methodologies. - For internally modelled exposures (IRB), projection of PiT and TTC PDs, PiT and DT LGDs, EAD, and RWA. For SA exposures, Projection of new flows of defaulted exposures and RWA based on risk weights for performing and nonperforming loans separately. Provisioning for IRB and SA modeled using IFRS 9 transition matrix approach. - Static balance-sheet approach, allowing the re-issuance of maturing loans at current market rates. - Traded risk impact from the revaluation of instruments at fair value (FVPL and FVOCI, including hedging instruments) will be assessed using bank-specific sensitivities reported in COREP/Short-Term Exercise to market risk factors. The analysis will use one-off market stress scenarios that have a similar narrative to the macro scenarios but correspond to a shorter time horizon. Risk factors include interest rate, commodity, equity, FX, and credit spread. - Satellite models for macrofinancial linkages: - Models for credit losses, funding costs, lending rates. - Within EA, for household and corporate, analysis of PD using micro-data at individual household (HFCS) and corporate (Datastream and Capital IQ). Outside of EA, expected default frequency used as proxies for corporate PDs, while a panel model used for household PDs. - LGD shocks for collateralized exposures linked to paths for real estate prices in the scenario using a smoothing factor to account for the TTC regulatory approach. - Interest income projected at geography-portfolio segment level using a structural approach applying interest rate shocks on new business and repricing of floating rate instruments. Funding costs projected at portfolio level using funding structure by product and maturity bucket. ### 3. Tail Shocks, Scenarios, and Horizon - Stress test horizon: 2025 – 2027 (three years). - Scenario set: - Baseline scenario drawn from the January 2025 WEO macroeconomic projections. - Adverse scenario 1: A geopolitical scenario featuring an escalation of geopolitical conflicts. - Adverse scenario 2: A recessionary scenario showing a synchronized global slowdown amplified by sovereign debt distress in EA. - The two adverse scenarios rely on GFM, a structural macro econometric model of the world economy, disaggregated into 40 national economies, documented in Vitek (2015). - Second-round effects and Sensitivity analysis: - Household “consumption at risk,” defined as the consumption of “economically vulnerable households” (for which the sum of debt service and consumption exceeds gross income) as a share of aggregate consumption. The elasticity of unemployment to changes in consumption will be used to test second-round effects on default risk. - Solvency and liquidity risk interactions testing business risk will be assessed in April 2025 for the G-SIBs. ### 4. Risks Covered and Buffers - Risks covered include credit (on loans and debt securities), market (valuation impact of financial instruments with respect to market risk factors such as interest rates, foreign exchange, credit spread, equity prices) and interest rate risk. ### Behavioral Adjustment Assumptions - Static balance sheet approach: size of portfolios (gross of NPLs) remains constant throughout the stress testing horizon (with no write-offs allowed). - In projecting RWAs, standardized and IRB portfolios are differentiated. For standardized portfolios, RWAs change due to shift in composition of performing and non-performing exposures, and a deterioration in creditworthiness is modeled as a credit rating downgrade linked to the initial rating of the exposure and the projected rise in loan losses. For IRB portfolios, through-the-cycle-PDs, downturn LGDs and EAD for each asset class/industry are used to project risk weights. - Interest income from nonperforming loans is not accrued. - Dividends are paid out by banks that remain profitable and adequately capitalized. The tax rate and the dividend rate are both set at 30 percent. ### Regulatory and Market-Based Standards and Parameters - Two hurdle rates considered: - “Minimum capital hurdle”: regulatory minimum Pillar 1 capital requirements (4.5 percent for CET1 ratio) plus Pillar 2 requirements (P2R). - “Breaching buffers hurdle”: includes the SREP capital requirements and capital buffers (CCoB, max (G-SII, O-SII), and SyRB). The CCyB is assumed to be zero in the scenarios. - Leverage ratio during the stress test horizon assessed against the 3 percent Basel III minimum requirement. ### Reporting Form for Results - Output presentation includes: - Aggregate capital path for each scenario by groups of banks, categorized by business model. - Aggregate capital shortfall relative to RWAs. - Number of banks and percent of banking assets in the system which fall below the hurdle rates. - Outputs also include information on the impact of different result drivers, including profit components. --- ### B. Banking Sector: Liquidity Stress Test — Top-Down by FSAP Team ### 1. Institutional Perimeter and Data - Institutions included: 95 SIs (out of 109 SIs), of which 7 are G-SIBs. - Market share: About 99 percent of the banking sector assets. - Data and horizon: - Data vintage: 2024: Q2 updated to 2024: Q4 in April/May 2025. - Data: Supervisory data from ITS files (FINREP, COREP). - Scope of consolidation: Consolidated group basis. Perimeter of the banking group (CRD V). Insurance activities are excluded; banking associates are included. ### 2. Channels, Methodology, and Stress Horizons - Methodology: - Structural analysis: evolution of LCR, NSFR, Asset Encumbrance, Funding concentration and Collateral Swaps. - Dynamic analysis: - (i) LCR-stress tests, using more severe scenarios than regulatory ones. Breakdown by significant currency, where available. - (ii) Cashflow-based stress test. Breakdown by significant currency, where available. - (iii) Reverse stress test to imply under which outflows banks would not meet regulatory requirements (LCR) or become illiquid (negative CBC). - Stress test horizon: - 30 days for LCR-based tests, and 1-day through 1-year for cashflow analysis. *Source: Appendix I. Risk Assessment Matrix (RAM) and associated stress testing appendices from the provided IMF content unit.* ### 3. Type of ### 1eurea2025007 - 3. Type of ### Liquidity Stress Test: Scenario Analysis and Framework - Domain: Top-Down by FSAP Team. - Purpose: Cash flow liquidity stress tests using various stress scenarios with varying intensity of adverse liquidity conditions. - Main risks analyzed: - (i) idiosyncratic risk due to reputational risks/decline in CET1 capital; - (ii) market upheaval and tightening of market liquidity conditions (linked to solvency adverse scenario, where possible), deposit run-offs, outflows from top funding sources. - Scenario types noted: - Cashflow scenarios with different adverse liquidity intensities. - Reverse stress tests to identify outflows that would breach regulatory limits or cause negative CBC. ### Liquidity Buffers and Behavioral Adjustments - Behavioral adjustments: - Different amounts of CBC using assumptions about ECB monetary policy (collateral) normalization. - Liquidity from the central bank (except for the lender of the last resort measures) considered under different assumptions about what type of collateral is included into CB eligible CBC. - Buffers: - Capacity of banks to generate liquidity from inflows and from assets under stress (i.e., counter-balancing capacity). ### Regulatory Standards and Reporting (Liquidity) - Regulatory/accounting and market-based standards: - For the LCR, the hurdle rate is set at 100 percent at the aggregate currency level (per Basel III and domestic regulation). - For cashflow analysis, outcomes of interest are the Net Liquidity Position and the survival period. - Reporting format / Output presentation: - Outputs include: - (1) Average LCR, Net Liquidity Position and survival period, - (2) Number of institutions with LCR below regulatory limits, - (3) Reverse stress tests. ### Solvency–Liquidity Interactions: Scope and Methodology - Domain: Top-down by FSAP Team. - Institutional perimeter: - Institutions included: 7 (all) G-SIBs. - Market share: 44 percent of the banking sector assets. - Data vintage: 2024: Q4. - Data: Supervisory data from ITS files (FINREP, COREP). - Scope of consolidation: Consolidated group basis. Perimeter of the banking group (CRD V). - Insurance activities are excluded; banking associates are included. - Methodology: - Based on Cont, R., Kotlicki, A., and Valderrama, L., (2020), “Liquidity at Risk: Joint stress testing of solvency and liquidity”. - Dynamic analysis includes mitigating actions to fend off liquidity pressures. - Sensitivity tests include reverse stress test to imply under which outflows banks would not meet regulatory requirements (LCR) or become illiquid (negative CBC). - Stress test horizon: 2-week and 2-days horizon. ### Solvency–Liquidity: Scenarios, Buffers, and Reporting - Approach / Scenario analysis: - “Reference scenario” (market shock calibrated for the solvency stress test); - “credit sensitive scenario” due to own credit downgrade; - “business as usual scenario” for signaling effects; - “narrow collateral scenario” whereby eligible credit claims and own issuances become ineligible for CB liquidity; - “trapped liquidity scenario” whereby FX collateral becomes ineligible; - “business risk scenario” showing credit attrition prompted by credit risk concerns. - Behavioral adjustments to mitigate liquidity shortfalls: - (i) unsecuritized borrowing; - (ii) repo borrowing; - (iii) liquidation of assets. - Buffers: - CET1 over bank specific regulatory leverage ratio (solvency), and counterbalancing capacity (liquidity). - Regulatory standards: - Bank specific regulatory leverage ratio (Pillar 1+Pillar 2R+G-SIB add-on). - Reporting of results / Output presentation: - Aggregate results include: - (1) Regions of failure due to illiquidity/insolvency; - (2) Solvency-liquidity diagrams for selected G-SIBs. ### Network Analysis: Design and Outputs - Domain: Top-down by FSAP Team. - Institutional perimeter: - Institutions included: 72 SIs (out of 109 SIs), of which 7 are G-SIBs. - Market share: Around 90 percent of the banking sector assets. - Data vintage: 2024: Q4. - Data: Supervisory data from ITS files (FINREP, COREP). - Scope of consolidation: Consolidated group basis. Perimeter of the banking group (CRD V). - Insurance activities are excluded; banking associates are included. - Methodology: - Based on Covi, G., Gorpe, M. Z., & Kok, C. (2021). “CoMap: Mapping Contagion in the Euro Area Banking Sector”. - Stress test horizon: 1-month horizon. - Type of analyses / Scenarios: - “Baseline scenario” (parameters calibrated as in Covi et al, 2021). - “NBFI risk scenario” including the default of the to five NBFI credit exposures (excluding insurance firms). - “Market risk scenario” including a 10 percent haircut in HQLA and 50 percent efficiency of hedges. - “Combined scenario” including the NBFI and Market risk scenarios. - Buffers and behavioral assumptions: - Behavioral adjustments: approach simulates cascading defaults and counterparty credit losses. No mitigating actions are assumed. - Buffers: CET1 over bank specific regulatory leverage ratio (solvency), and counterbalancing capacity (liquidity). - Regulatory standards: - Bank specific minimum CET1 ratio (Pillar 1+Pillar 2R) (solvency), and liquidity reserves (HQLA) minimum net outflows (liquidity). - Reporting format / Output presentation: - Outputs include: - (1) Number of triggered defaults; - (2) Capital depleted due to triggered defaults; - (3) Contagion Index; - (4) Vulnerability Index. - Results are presented by banks’ business models. ### Annex III: Macrofinancial Scenario Paths (Selected Euro Area Aggregates) - Annex III provides variable paths across macrofinancial scenarios for multiple countries and aggregates. Selected Euro area aggregate paths: - GDP Growth Paths (Percent) — Euro area: 0.75 2024; 1.03 2025; 1.37 2026; 1.32 2027. - Unemployment Paths (Percent) — Euro area: 6.48 2024; 6.51 2025; 6.41 2026; 6.31 2027. - Headline Inflation Paths (Percent) — Euro area: 2.34 2024; 2.11 2025; 2.02 2026; 1.97 2027. - House Price Index Paths (Rebased to 100 in 2024) — Euro area: 100 2024; 102.1 2025; 104.2 2026; 106.2 2027. - Market Risk Scenarios (Annex III Table 5): - Note indicates risk factor mappings (CM: commodities; CR: credit spreads; EQ: equity; IR: interest rates) and a list of delta shocks and volatility shocks for multiple instruments and tenors as reported in the source (examples include energy 56.3 -58.8; industrial metals 24.8 -28.1; USD-1M 155.8 19.1 155.8 19.1; EUR-1M 148.7 14.8 14.8). The table presents delta shock and volatility shock parameters by risk factor for the Geopolitical - market scenario and Recessionary - market scenario. ### Appendix IV: Solvency Stress Test — Technical Aspects - Credit Risk: - All financial institutions in the solvency stress test have adopted IFRS9 standards and credit impairments are calibrated in accordance with this accounting framework. - Scenario-based transition matrices are estimated using Beta-linking (Gross et al., 2020), where an aggregate PD is projected and adapted to stage 1 and stage 2 exposures according to the most recently observed transition matrices. - Starting transition matrices for every bank and lending segment are constructed with data from FINREP (F04.04.1 for balance stocks and F12.02 for flows between stages). - LGD for unsecured lending is calibrated using the Frye and Jacobs (2012) method. - Starting loan to values for collateralized lending and starting LGDs for all lending segments are reported in the STE templates for credit risk. - Both secured and unsecured LGDs have been calibrated at the bank-specific portfolio level. - Under both adverse scenarios, the substantial decline in property values materially affects the LGD of secured lending and thus the credit risk charges associated with mortgage lending. - Net Interest Income (NII): - Data sources used to obtain inputs: - i. The NII in the year before the cut-off date, 푁퐼퐼 푏,푡=0, obtained from FINREP (F16.01). - ii. Exposures and liabilities for each segment—by portfolio type and by country of counterparty—obtained from FINREP (F20.04 and F20.06, respectively). - iii. The repricing ladder for each segment obtained by combining sources: STE templates for interest rate risk in the banking book (IRRBB) and template J05.00 from the supervisory IRRBB module, complemented with COREP C66.01 when other sources were not available. - iv. Projections for the new business rate deltas, 푑푟 푡,푗 푛푏, were obtained from satellite models estimated using the ECB’s data set “MFI Interest Rates Statistics” (MIR). *Source: IMF staff calculations.* ### 4. New business rates were projected by estimating a range of passthrough regressions ### 4. New business rates were projected by estimating a range of passthrough regressions ### Method for projecting new business rates by segment - Data source and coverage: - The publicly available MIR dataset reports new business rates by portfolio and by country of counterparty, covering: mortgages, non-mortgage household credit, and NFC loans on the asset side; and household sight, household term, NFC sight, and NFC term deposits on the liability side. - For remaining portfolios, the FSAP assumed a unit passthrough from market rates (where market rates were either the 1-month EURIBOR or the sovereign yield). - MIR covers all EU countries of counterparty; for non-EU countries, alternative data sources were used (e.g., country Central Bank statistics, Haver, etc.). - Passthrough regressions: - Regressions estimated independently for each segment j at quarterly frequency using the specification: - ∆i_{t,j}^{nb} = α_j + β_{0,j}^{ST} ∆i_{t,c}^{ST} + β_{1,j}^{ST} ∆i_{t-1,c}^{ST} + β_{1,j}^{LT} ∆i_{t,c}^{LT} + β_{1}^{LT} ∆i_{t-1,c}^{LT} + γ_j GDPgrowth_{t,c} - Notation: ∆ denotes quarterly changes; i_{t,j}^{nb} is the new business rate for segment j; i_{t,c}^{ST} is a short-term rate for country c (EURIBOR for EA countries, short-term sovereign yield for non-EA countries); i_{t,c}^{LT} is the long-term sovereign yield; GDPgrowth_t is quarter-on-quarter real GDP growth. ### Balance-sheet and repricing-ladder inputs - Required inputs at bank-segment level: - A repricing ladder at T0: exposures in each repricing bucket [k,k+1] denoted E_0[k,k+1], with share θ_0[k,k+1] = E_0[k,k+1] / E_0. - Any exposure with time-to-repricing larger than three years can be allocated to the [3;4] year bucket (exposures will not reprice within the three-year stress-testing window). - Scenario-specific projections for the interest rate on new business, denoted r_t^{nb}. ### Model calculations — four steps - Step 1: Simulate exposures originated/repriced in each bucket and period - Law of motion for exposures across buckets: - E_t[k-1,k] = E_{t-1}[k,k+1] + I_t[k-1,k] - I_t[k-1,k] are newly issued/repriced loans in bucket [k-1,k] during year-t. - Key assumption: shares of exposures across buckets are constant over time: - θ_t[k,k+1] = θ_0[k,k+1] for all t,k - This assumption is consistent with the static balance sheet used throughout the stress test. - Step 2: Simulate the average interest rate for each bucket and period - Define r_{t-1}[k,k+1] as average interest rate of exposures that at end of year (t−1) were in bucket [k,k+1]. - Recursive calculation (exposure-weighted average): - i_t[k-1,k] = ρ_t r_{t-1}[k,k+1] + (1−ρ_t) r_t^{nb} where ρ_t = E_{t-1}[k,k+1] / E_t[k-1,k] - Initial condition: r_0[k-1,k] = IIR_0 (initial interest income rate of the portfolio at T0). - Step 3: Calculate interest income - Case without NPEs: - II_t = ∑_{k=0}^3 r_{t-1}[k,k+1] E_0[k,k+1] + (1−ω)(r_t^{nb} − r_{t-1}[0,1]) E_0[0,1] - ω = avg days to repricing / 365 - First term: base rate determined in year (t−1); second term: effect of year t interest-rate shock on interest-sensitive assets. - Incorporating NPLs (assume NPE ratio same across buckets): - ĨI_t = (1 − NPEr_t) ∙ II_t - Interest income adjusted by exposure-weighted average share of performing exposures. - Step 4: Rewrite model in terms of interest-rate deltas relative to IIR_0 - Define dr as interest-rate delta relative to IIR_0 (e.g., dr_t^{nb} = r_t^{nb} − IIR_0). - In no-shock case (r_t^{nb} = IIR_0 for all t), interest income would be ĪI_t = IIR_0 ∙ E_0. - Difference form: - II_t − ĪI_t = ∑_{k=0}^3 dr_{t-1}[k,k+1] E_0[k,k+1] + (1−ω)(dr_t^{nb} − dr_{t-1}[0,1]) E_0[0,1] - Interest Expense Delta defined analogously by replacing exposures with liabilities. ### Net fee and commission income — econometric findings - Main statistically significant drivers across specifications: - GDP growth: positive and significant. - Change in EURIBOR (D.EURIBOR): negative and significant. - Stock price growth: significant with a positive coefficient only for large banks. - Interpretation: - Higher real GDP and stock price growth indicate better real economy and financial markets, expanding fee and commission income. - Stock prices more significant for larger banks due to business models with larger shares of NFCI from securities, corporate finance, asset management. - Negative EURIBOR coefficient may reflect higher rates leading to lower bank business volumes and reduced fee income from payment services. - Selected regression coefficients (all variables expressed in percentages; significance levels: * p<0.10, ** p<0.05, *** p<0.01): - NFCIR (t-1): 0.707*** (columns reported for both large and small/medium banks, range includes 0.623*** to 0.717***). - RGDP growth: 0.00891*** (all regressors), 0.00862*** (Lasso-selected), 0.00880*** (Arellano-Bond) for large banks; 0.00700***, 0.00612**, 0.00584*** for small/medium-sized banks. - Stocks growth: 0.00189** to 0.00222*** for large banks; small/medium-sized banks show insignificant or much smaller coefficients (e.g., 0.000349 to 0.000523). - D.EURIBOR: -0.0270*** to -0.0295*** for large banks; -0.0162** to -0.0184** for small/medium-sized banks. - R-squared reported: 0.609, 0.609, 0.608 (large banks regressions); 0.605 (Arellano-Bond in small/medium table). - Sample sizes N: 595, 595, 568 (large banks); 942, 942, 867 (small/medium-sized banks). ### Solvency stress test — key results by business model - General structure: - Appendix V details drivers of capital depletion and profitability by business model; results shown for three scenarios: baseline, geopolitical shock, and recessionary shock. - G‑SIBs (Globally Systemic Banks) - Baseline: - Strong fee generation and favorable market-risk income boost capital early, offsetting credit-loss charges, operating expenses, and shareholder distributions. - CET 1 ratio moves from mid‑14 percent starting point to just above 15 percent. - Geopolitical and recessionary shocks: - Revenue lines weaken; market-risk results less supportive; dominant drag from sharply higher credit-risk charges. - CET 1 ratio pushed down into the 9 percent range. - Elevated credit losses and RWA growth under stress are primary vulnerabilities; revenue shortfalls secondary. - Profitability: - Baseline returns: about 0.42 percent of assets in 2025; roughly 0.24 percent in 2026; 0.23 percent in 2027. - Geopolitical: losses about 0.20 percent in 2025 deepening to roughly 0.62 percent in 2026; near break‑even in 2027. - Recessionary: losses roughly 0.17 percent in 2025 widening to about 0.63 percent in 2026; around 0.09 percent (slightly negative) in 2027. - Lenders (traditional deposit‑funded banks) - Baseline: - Starting mid‑16 percent CET1, strong core revenues push ratio sharply higher early; ratio finishes slightly above starting point. - Geopolitical: - Revenue lines hold reasonably well, but larger credit‑losses and higher operating costs erode gains; CET 1 falls into the 12–13 percent range by end-horizon (about four percentage points below baseline). - Recessionary: - Deeper revenue hit, higher credit‑risk charges, stronger RWA inflation pull CET 1 down to roughly 11.5 percent (lowest of three scenarios). - Profitability: - Baseline returns: roughly 0.56 percent in 2025 to about 0.32 percent in 2027. - Geopolitical: loss of about 0.24 percent in 2025 deepening to roughly 0.52 percent in 2026; profit around 0.18 percent in 2027. - Recessionary: losses roughly 0.43 percent in 2025 and about 0.63 percent in 2026; slightly negative 0.10 percent in 2027. - Investment banks - Baseline: - Begin with a healthy CET 1 buffer; strong fee income and trading revenues produce front-loaded capital gains; cohort ends horizon comfortably above starting CET 1. - Geopolitical: - Core revenues resilient but market‑risk income and fee generation soften; credit‑risk charges rise; CET 1 pulled down into the mid‑21.1 percent range by end of horizon (below baseline but above regulatory minima). - Recessionary: - Deeper revenue compression, heavier credit‑risk losses, stronger RWA surge bring year-end CET 1 to the low‑20.5 percent range; investment banks remain best‑capitalized cohort. - Profitability: - Baseline returns: roughly 0.31 percent of assets in 2025 tapering to about 0.27 percent by 2027. - Geopolitical: sharp setback to 0.38 percent loss in 2025; break‑even in 2026; modest profit roughly 0.14 percent in 2027. - Recessionary: losses deepen to about 0.41 percent in 2025 and persist in 2026; marginally positive by 2027. ### Key analytical takeaways - Modeling approach: - The projection of new business rates combines segment-level passthrough regressions with a static balance-sheet repricing-ladder simulation, recursive interest-rate averaging, and explicit adjustment for NPEs. - Drivers of income under stress: - Interest‑rate shocks affect interest income through repricing dynamics and the timing fraction ω. - Credit‑risk charges and stress‑related RWA inflation are the dominant drivers of capital depletion in adverse scenarios, particularly for G‑SIBs and Lenders. - Fee and commission income: - Sensitive to real GDP growth and EURIBOR changes; stock-price growth matters more for large banks. - Cohort resilience: - Investment banks display the largest buffers in stress scenarios (CET 1 remaining in low‑20s in adverse paths), while G‑SIBs and Lenders can see CET 1 fall to around 9 percent and roughly 11.5 percent respectively under severe stress. *Source: IMF staff calculations and model description in the chapter "New business rates were projected by estimating a range of passthrough regressions".* ### 10. In the baseline, Universal banks benefit from solid fee income and trading gains, which ### 1eurea2025007 - 10. In the baseline, Universal banks benefit from solid fee income and trading gains, which ### Capital dynamics for Universal banks - Baseline scenario: - Solid fee income and trading gains lift capital materially in the first years (Annex V. Figure 7). - Normal credit‑loss charges, operating costs, and shareholder distributions erode part of the gain. - Measured increase in RWAs trims capital further. - Cohort finishes the period a little above its starting level, indicating the franchise generates enough earnings in benign conditions to build capital modestly. - Baseline outcome: CET 1 slightly above the mid‑15 percent starting level by 2027. - Geopolitical scenario: - NII proves resilient, but fees and trading revenues soften and credit‑risk charges rise markedly. - Higher expenses further reduce capital. - Decisive hit from stress‑related RWA inflation drags the year‑end CET 1 ratio down to just under 12 percent, about 4 percentage points below the baseline finish. - Recessionary scenario: - Deeper fall‑off in revenues, heavier credit‑losses, and an even larger surge in RWAs. - Most severe erosion: CET 1 ratio in the high‑11 percent range at the horizon. - Outcome remains above minimum requirements but underscores sensitivity to broad‑based macro downturns despite diversification. *Source: IMF staff calculations.* ### Profitability across scenarios (Universal banks) - Baseline: - Diversified revenue mix (steady NII, solid fee and trading) keeps earnings strong. - Profits peak at about 0.65 percent of assets in 2025, dip to roughly 0.32 percent in 2026, and rebound to around 0.35 percent in 2027. - Geopolitical: - Revenues soften but profitability is not eliminated. - Returns fall to roughly 0.27 percent in 2025, swing to a moderate loss of about 0.33 percent in 2026 amid valuation hits and higher credit‑risk charges, then recover to a modest profit of roughly 0.20 percent in 2027. - Recessionary: - Most punitive scenario. - Profits compress to about 0.20 percent in 2025 and turn sharply negative (‑0.46 percent) in 2026 as credit impairments surge and margins compress. - Earnings remain slightly negative in 2027 (‑0.06 percent), showing profitability can be impaired under severe macrofinancial conditions. *Source: IMF staff calculations.* ### Liquidity Stress Test — methodological overview (Cash‑flow based liquidity stress tests) - Purpose and indicators: - Transform reported cash‑flow data into stressed cash‑flows and security flow data using scenario‑dependent stress factors. - Focus on liquidity risk exposure (net‑funding gap, NFG, and cumulated net‑funding gap, CNFG) and liquidity risk bearing capacity (CBC and cumulative CBC). - Analysis builds on COREP C66 template with data cut‑off point of December 2024. - Notable changes between 2020 and 2024 in the overnight time bucket: - Overnight CB inflows: 7.8 percent of total funding in 2020 → 0.5 percent in 2024. - SFT funding: negligible in 2020 → 5 percent of overnight flows in 2024. - Share of stable retail overnight deposits declined by 2 percentage points to 10 percent in 2024. - Share of FX swaps increased to 2 percent from 0.3 percent in 2020. - Liquidity vulnerability shifts: - Banks more susceptible to contingent liquidity risks from changes in net value of repos and derivatives. - Net overnight contractual gap widened from 27 percent in 2020 to 31 percent in 2024. - Adverse scenario impact on the value of collateral posted for derivative transactions could not be fully estimated due to data gaps; repo/reverse repo collateral (up to three months) was considered with haircuts from the market risk scenario. ### CFLST scenarios, calibration, and parameters - Scenario design and uncertainty handling: - FSAP team designed several scenarios of increasing severity linked with macro scenarios and reverse stress tests across multiple horizons; reverse tests capped outflow rates at double original rates with condition that 20 percent of banks (by assets) need to fail. - Annex V. Table 1 (key CFLST scenario parameters, Percent) — selected entries preserved verbatim: - SC1 Mild; SC2 Idiosyncratic; SC3 Severe (outflows and CBC haircuts table in source). - Example outflow rates (row headings preserved): Stable retail deposits 5 8 5; Other retail deposits 10 40 10; Operational deposits 25 49 15; Non-operational deposits from credit institutions 100 100 50; FX-swaps maturing 50 100 50. - Example CBC haircuts (level and percentages preserved): Level 1 central bank 100 100 100; Level 1 (CQS 1) 98 98 90; Level 2A corporate bonds (CQS1) 85 80 80; Level 2B ABS (CQS1) 75 75 60; Other tradable assets 75 75 60; Non tradable assets eligible for central banks 62 62 50; Own issuances eligible for central banks 62 62 50. - Time horizons and scenario narratives: - Three scenarios range between one‑day to 1‑year horizon; same cash flow rates through all time buckets. - (i) Mild: Macro shock due to general risk aversion (affects CBC); some drawdown of liquidity facilities. - (ii) Moderate: Macro shock due to sovereign distress (affects CBC and outflows from individual banks); severe drawdown form liquidity facilities. - (iii) Severe: Idiosyncratic risk leads to severe outflows (bank specific). - Inflow and contingent assumptions: - Key inflows from loans to non‑financial corporates, retail loans set at 0 percent in all scenarios. - Other inflows from secured transactions (SFTs) set to 100 percent. - Collateral monetization reduced due to declining market prices (higher margin requirements); outflows from derivatives and other contingent items left as calibrated in bank submissions. - CBC approaches (three): - (i) Unrestricted — Full CBC without additional haircuts when unencumbered eligible collateral exists. - (ii) Restricted to liquid and marketable CBC — non‑marketable components (18 percent of CBC in the sample) disregarded. - (iii) Most restricted — Liquid HQLA CBC with haircuts and market price effects from solvency stress test for liquid private market assets; 28 percent of CBC in the sample disregarded. - Calibration sources: - Parameters calibrated using March 2023 banking turmoil, ECB/SSM 2019 liquidity stress test, and EA 2018 FSAP. - Annex VI. Table 2 (Deposit Outflow Rates in Liquidity Crisis Episodes, Percent) — selected entries preserved verbatim: - Silicon Valley Bank (2023) Observed deposit outflow rate 85; Period 2 days; Daily 43. - Northern Rock (2007) Observed deposit outflow rate 20; Period 4 days; Daily 5. - Wamu (2008) Observed deposit outflow rate 10; Period 10 days; Daily 1. - First Republic Bank (2023) Observed deposit outflow rate 57; Period 90 days; Daily 0.6. - LCR assumption rows: Retail stable 5 30 days 0.2; Retail less stable 10 30 days 0.3; Operational 25 30 days 0.8; Non‑Financial Corporate 40 30 days 1.3; Average (net) outflows for the euro area SIs 4.2 30 days 0.1; IMF ST daily (ex SVB and CS) 1.5. ### Solvency‑Liquidity interactions — methodological highlights - Valuation and CCR: - Changes in market valuation computed with partial revaluation using sensitivities (delta, gamma, vega) from STE market risk module. - CCR losses in two‑day scenario linked to NBFI counterparty default rates estimated in system‑wide liquidity analysis; in two‑week exercise CCR losses driven by default of the three most vulnerable counterparties among top 20 derivatives exposures (excluding qualifying CCPs, central banks, central governments, and international institutions). - Liquidity flows composition at t=2: - Unconditional flows: maturing liabilities (S0) and scheduled contractual outflows (SCO). - Contingent flows: margin and collateral calls for derivatives and SFTs. - Contingent flows from own credit downgrade leading to deposit run‑off. - Resulting conditional increase in cash outflows and scheduled contractual inflows mapped in formulae (equations (2) and (3) in source). - Liquidity at Risk (LaR) and mitigation: - LaR measures net outflows for the stress scenario (equation (4) in source). - Liquidity shortfall allows mitigating actions with a pecking order: unsecuritized interbank borrowing, repo borrowing, liquidation of assets (fire sales). - Mitigating actions increase cash but reduce equity; raising liquidity amplifies the initial market shock on bank capital (Loss amplification formula in source). - Mapping templates: - Marketable assets (Asset M) and non‑margin assets (Asset N) identified from FINREP and COREP templates. - Illiquid assets include encumbered and non‑marketable assets from asset encumbrance template (F32.01). - Initial equity losses arise from fair value shocks, PVA, and CCR losses using top 20 derivative exposures and stressed EEPE. - FINREP (F22.01) used for fee and commission income breakdown; FCI shock calibrated to Credit Suisse’s 2022 FCI loss for geopolitical market scenario narrative. - Contractual cash flows from COREP maturity ladder (C66.01.a) and outflows mapped to LCR categories (C73.00.a) for 2‑week and 2‑day horizons. ### Network analysis—brief overview - Contagion Mapping Model (CoMap): - Framework based on Covi, Gorpe, and Kok (2021) to simulate knock‑on effects from a hypothetical default of an EA bank across interbank exposures. - Captures propagation via credit channel (credit risk) and funding channel (loss of funding). - Simulates hypothetical defaults of each bank in the interbank network to assess contagion channels and system resilience. *Source: IMF staff calculations.* ### 2. The credit channel captures the impact of hypothetical default of a bank on its ### 2. The credit channel captures the impact of hypothetical default of a bank on its ### Credit channel: direct counterparty losses - When a bank in the network defaults, other banks with direct exposures to the defaulting entity face potential losses. - For a subset (풴) of banks defaulting, bank 풊’s credit losses are: - (1) 퐿푂푆푆푖푐푟푒푑푖푡 = 훴j∈풴 훴푘 휆푖푗푘 푥푖푗푘 - where 휆푖푗푘 are exposure-specific loss-given default rates corresponding to claim type k on bank j, and 푥푖푗푘 are the obligations. ### Funding channel: funding withdrawal and liquidity shortfalls - Funding shortfalls arise when defaulting banks withdraw funding from other banks in the network. - Funding shortfall for bank 풊 is: - (2) 푇퐹푆푖 = 훴j∈풴 훴푘 휌푖푘 푥푗푖푘 - where 휌푖푘 is a bank-specific funding shortfall rate applied to funding 푥푗푖푘 received from bank j. - Bank 풊 can pledge HQLA in excess of net liquidity outflows, 훾푖, to the central bank to absorb shortfalls; remaining liquidity shortage is: - (3) 푚푎푥{0, 훴j∈풴 훴푘 휌푖푘 푥푖푗푘 − 훾푖} ### Fire-sale deleveraging and funding losses - If the remaining liquidity shortage is strictly positive, bank 풊 may sell unencumbered marketable non-central bank eligible assets at a discount 훿푖 to meet liquidity needs. - Given a limited pool of such assets, 휃푖, potential losses from fire sales are: - (4) 퐿푂푆푆푖푓푢푛푑푖푛푔 = 훿푖 · 푚푖푛{ 1 1−훿푖 · 푚푎푥{0, 훴j∈풴 훴푘 휌푖푘 푥푖푗푘 − 훾푖}, 휃푖 } ### Default condition: insolvency and illiquidity - Insolvency (default due to insufficient capital): - (5) (푐푖,푡 − 푐푖,푚푖푛 < 퐿푂푆푆푖푐푟푒푑푖푡 + 퐿푂푆푆푖푓푢푛푑푖푛푔 - where 푐푖,푡 is the bank’s capital position at time t and 푐푖,푚푖푛 is minimum capital requirements. - Illiquidity (default due to insufficient liquid assets): - (6) 휃푖 < 1 1−훿푖 · 푚푎푥{0, 훴j∈풴 훴푘 휌푖푘 푥푗푖푘 − 훾푖 } ### Model parameters for the Reference Scenario (“Baseline”) - Calibration approach: - Parameters are calibrated in line with Covi, Gorpe, and Kok (2021), except for the funding shortfall rate and the fire sale discount rate. - Specific parameter definitions and assumptions: - Loss given default (휆푖푗푘): calibrated as a ratio of net exposure to gross exposure. Net exposure = exposure value after exemptions and credit risk mitigation instruments; gross exposure = exposure value after exemptions but before application of credit risk mitigation instruments (C.27-C.28). - Funding shortfall rate / short-term funding share (휌푖푘): assumed to be 100 percent. This implies that interbank funding matures by the end of the horizon (30-day scenario). - Liquidity surplus / net liquidity position (휆푖): defined as HQLA assets (C.72.00.a.10), 퐻푄퐿퐴, in excess of net liquidity outflows (C.76.00.a.20). - Pool of assets available for fire sale (휃푖): defined as total unencumbered non-central bank eligible assets (F.32.01). - Fire sale discount rate (훿푖): assumed to be 50 percent. - Bank’s capital surplus / default threshold: defined as excess capital over minimum capital requirements. Bank’s capital is CET1 capital. Minimum capital requirement (hurdle rate) is defined as 4.5 percent of RWAs and the bank-specific P2R set in the SREP process. ### Summary statistics (Annex VIII. Table 3; Billion euros; 2024Q2) - Notes: Sources: ECB and IMF staff calculations. SD denotes the standard deviation, p25, p50, and p75 are the respective percentiles. - Baseline Analysis (Mean, SD, p25, p50, p75): - Gross Exposure: 0.7, 0.9, 0.1, 0.3, 0.9 - Credit Risk Mitigation: 0.1, 0.4, 0.0, 0.0, 0.1 - Net Exposures: 0.5, 0.8, 0.1, 0.2, 0.6 - Loss Given Default: 0.8, 0.3, 0.8, 1.0, 1.0 - CET1: 16.2, 20.6, 3.2, 7.4, 19.0 - CET1 Minimum: 6.1, 8.3, 1.1, 2.6, 7.0 - HQLA: 61.6, 81.0, 10.4, 26.3, 91.8 - Net Liquidity Outflows: 38.9, 56.8, 4.5, 11.6, 56.6 - Unencumbered non-HQLA assets: 215, 346, 39, 66, 204 - Total Assets: 329, 492, 57, 110, 351 - Risk-Weighted Assets: 110, 151, 21, 45, 127 *Source: https://www.imf.org/-/media/files/publications/cr/2025/english/1eurea2025007.pdf* --- _Source: https://www.imf.org/-/media/files/publications/cr/2025/english/1eurea2025007.pdf_