## EXECUTIVE SUMMARY ## Source details **Canonical URL:** [EXECUTIVE SUMMARY](https://www.imf.org/-/media/files/publications/cr/2025/english/1fraea2025007.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2025/english/1fraea2025007.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2025/english/1fraea2025007.pdf.json) --- ### Financial sector resilience and prevailing risks - The French financial sector has proven resilient to the stress events of the last five years but faces headwinds from domestic and global uncertainty. - Sovereign debt markets have functioned well; large debt issuance continues to be smoothly absorbed by a deep and well diversified buyer base. - Credit growth has moderated; the housing market is undergoing an orderly adjustment while household and non-financial corporates (NFC) debt remain elevated. - Stability of the French financial sector is an important element of strength amid domestic political uncertainty and rising global geoeconomic risks. ### Banking sector structure, performance, and vulnerabilities - Structure and concentration: - 6 banks account for about 96 percent of total banking assets and 292 percent of GDP. - 4 of the 6 major banks are G‑SIBs. - BNP Paribas and Société Générale have the majority of their credit exposures outside France; Crédit Agricole: 35 percent foreign exposure. - Capital, liquidity, profitability metrics: - Aggregate LCRs (in all currencies) and in euros have remained stable at around 150 percent on average every month since 2020. - LCR: above the Basel requirement; above G‑SIB peer average but below EU peer average. - NSFR: above requirements but below EU and G‑SIB peer averages. - Tier 1 capital: French aggregate below EU average, but above G‑SIBs’ average. - Leverage ratios: below EU peer average and G‑SIB peer average. - ROA: fell to below peers’ average by 2023/2024. - NII: low pass-through for fixed-rate housing loans; aggregate NII declines by 0.3 percent of RWAs in the two adverse scenarios relative to baseline. - LGD for households: aggregate 19 percent; LGD for residential housing loans: 14 percent; about 2/3 of French residential housing loans benefit from a guarantee. - Retail housing loans: - High-quality borrowers, long-term fixed-rate loans, and loan guarantee schemes have resulted in low credit losses and high resilience to interest rate shocks. - Profits on home loans are low and often cross-subsidized. - Sector vulnerabilities: - NFCs vulnerable due to high leverage levels; SME defaults rose as Covid-era measures finished rolling off but remain manageable. ### Solvency stress tests and macroprudential analysis - Coverage and scope: - Solvency stress tests cover the largest 7 SIs (about 96 percent of bank assets) using end‑2024 balance sheets and P&L statements. - Tests are top-down, IMF in-house models, aligned with Euro Area FSAP adverse scenarios (November 2024). - Adverse scenarios and macro shocks: - Geopolitical scenario: 2.4 deviation of real GDP growth from baseline; decline of real GDP of about 5.3 percent after two years; short-term interest rates increase by 1.9 percentage point in the first year; long-term interest rates spike at 5.5 percent in 2024; unemployment to 10 percent after 3 years. - Recession scenario: 2.7 deviation of real GDP growth; decline of real GDP of about 6.3 percent after two years; short-term interest rates decline by 0.6 percentage point in first year; long-term rates spike at 4.8 percent in 2024; unemployment to 13.5 percent after 3 years; term spreads widen by up to 2.1 percent. - Aggregate solvency outcomes: - No banks breach minimum capital requirements (Basel minimum CET1 plus Pillar II) under either adverse scenario. - Four banks, including 3 G‑SIBs among them, dip into additional capital buffers (capital conservation buffer plus G‑SIB and O‑SII buffers). - Aggregate system-wide gap to capital requirements including buffers: - 1.2 percent of risk weighted assets after three years in the geopolitical scenario. - 0.8 percent of risk weighted assets after three years in the recession scenario. - Decline in system-level CET1 ratio: - 490 basis points under the recessionary scenario. - 570 basis points under the geopolitical scenario. - Main drivers of capital depletion (in decreasing order): - Credit risk: 2.0–2.3 percent of RWAs (mainly NFCs). - Market risk: 1.2–1.3 percent of RWAs. - Net fees and commission income (NFCI): 0.7–1.1 percent of RWAs. - G‑SIBs specifics: - G‑SIB capital ratio declines: 520 basis points (geopolitical), 370 basis points (recession). - G‑SIBs have stronger baseline capitalization; credit risk contribution lower (1.6–1.9 percent of RWAs). - Sensitivity and counterfactuals: - Banking system resilient to concentration risks after accounting for credit risk mitigation (CRM). - Sectoral credit risks limited; construction and real estate account for 5 percent and 23 percent of loans to NFCs respectively. - Counterfactual analysis for CCyB calibration: - Precautionary buffers ≈ 6 percent of RWA + current 1 ppt CCyB sufficient to maintain solvency across a wide range of outcomes. - Current CCyB at 1 percent of RWAs can absorb a moderate macroeconomic shock; a 2-percentage point CCyB maps to larger scenario severity (scenario 4: real GDP 3.4 percent below baseline after 3 years). ### Market shocks and market risk design - One-off market shocks applied at start of year 1 and carried over unreversed for 3 years; two sets narrative-aligned to adverse scenarios. - Market shocks include commodity (CM), credit spreads (CR), equities (EQ), interest rates (IR), and FX; include volatility shocks. - Market revaluation approach for FVOCI and FVPL uses bank-specific sensitivities (delta, gamma, vega) from STE. - Conservative floor on market risk losses: max(bank’s own FRTB requirement, 8 percent of bank’s RWAs for market risk). ### Net Interest Income (NII) and interest‑rate pass‑through - NII modelling: semi-structural repricing gaps + econometric pass-through regressions (2000:Q2–2024:Q4). - Regressions R2 examples: Housing Loans 0.718; Loans to NFCs 0.897; Consumer Credit 0.632. - Scenario summary of shocks to interest rates (in percentage points): - Baseline (2025 / 2026 / 2027): Housing Loans -0.4 / -0.1 / 0.0; Loans to NFCs and Consumer Loans -0.6 / 0.0 / 0.1; Other Interest Earning Assets 0.0 / -0.4 / -0.4. - Geopolitical (2025 / 2026 / 2027): Housing Loans 0.3 / 0.0 / -0.8; Loans to NFCs and Consumer Loans 0.3 / 0.0 / -0.8; Other Interest Earning Assets 1.9 / -0.6 / -1.5. - Aggregate NII outcome: decline of 0.3 percent of RWAs in adverse scenarios relative to baseline. ### Liquidity and funding resilience - Structural liquidity: - Aggregate LCR ~150 percent since 2020; lowest observed LCR: 136 percent; median LCRs under stress: 107 percent (all currencies) and 101 percent (euros). - All USD LCRs above 100 percent at end‑2024 though monthly volatility exists. - NSFR: all banks meet minimum but below EU and G‑SIB peers. - Wholesale funding accounts for 50 percent of unweighted available stable funding; retail deposits account for 46.6 percent (weighted). - USD funding: 21 percent of unweighted available stable funding (26 percent for G‑SIBs); ratio weighted/unweighted share of USD funding: 176 percent (all banks), 184 percent (G‑SIBs). - Stress test outcomes: - Under LCR stress with higher run-off rates and HQLA valuation losses: aggregate LCRs (all currencies) = 114 percent; in euros = 112 percent. - 3 banks have LCRs below 100 percent in that stress (2 are G‑SIBs). - Cash flow stress tests: survival horizons exceed one month for all but one bank; one bank experiences a small cash shortfall within 2 weeks in some scenarios. - Reverse cash‑flow stress tests: scenario 2 takes 8 iterations for 4 banks to face funding gaps at a 30-day horizon. - System‑wide liquidity tests (fund sell‑offs): - France‑domiciled funds shock: 1–5 year France sovereign bonds price fall 13 percent (first 2 days) and 19 percent (2 weeks); aggregate LCR declines to 122 percent (2 days) and 120 percent (2 weeks). - Euro Area‑wide shock: AA sovereign bonds 3–5 year: price fall 3 percent (2 days) and 6 percent (2 weeks); aggregate LCR declines to 125 percent (2 days) and 123 percent (2 weeks). - In both exercises, 3 banks experience LCR declines after 2 days. ### Investment funds’ liquidity stress tests and redemption risk - Industry size and sample (end‑Q3/2024 and end‑2024 data): - IFs and MMFs at end‑2024: 13,511 funds with NAV ≈ EUR 2 trillion (60 percent of GDP). - Sample: bond funds 293 (EUR 118 billion), mixed funds 456 (EUR 88 billion), MMFs 62 (EUR 402 billion); total sample NAV EUR 608 billion; NAV of these three types at September 2024: EUR 1,070 billion. - MMF sample coverage 91 percent; bond funds 37 percent; mixed funds 29 percent. - Holdings and interconnections: - IFs held EUR 645 billion in French debt securities and EUR 477 billion in French equities (end‑2024). - Nominal French debt securities outstanding: EUR 5.7 trillion; French equity market capitalization: EUR 2.7 trillion. - A fifth of EUR 1.6 trillion in assets held by IFs invested in shares of other IFs and MMFs; funds-of-funds rose from EUR 320 billion (Q4 2019) to EUR 390 billion (Q3 2024). - Insurance companies: largest investors in IF shares/units; insurance companies hold one-fourth of portfolios in investment funds. - Stress test results: - Under uniform redemption shocks: - 2 percent uniform shock: 19 bond funds and 1 mixed fund (2.5 percent of funds) face liquidity shortfalls; aggregate shortfall EUR 88 million (0.01 percent of NAV). - 5 percent uniform shock: 31 bond funds and 1 mixed fund (3.9 percent) face shortfall EUR 421 million (0.07 percent of NAV). - Scaled to end‑2024 industry: equivalent shortfall at 2 percent = EUR 240 million; at 5 percent = EUR 1.1 billion. - Monte Carlo COVID‑19 distribution: if 5 percent of bond funds face -14.979 percent NAV outflow, one‑third of draws show aggregate liquidity shortfall between 8 and 12 percent of NAV. - MMF tail risk: if largest MMFs face 29 percent redemption shock, aggregate shortfall could reach 33 percent of NAV. - Liquidation strategies and spillovers: - Pro‑rata liquidation (5 percent redemption): EUR 61 billion liquid assets and EUR 8 billion cash disposed. - Waterfall with cash hoarding: funds sell French government securities more aggressively; banks’ securities sales (EUR 11–23 billion) could put significant price pressure. - Conclusions on funds: - Open‑ended bond funds, mixed funds, and MMFs generally have sufficient liquidity to withstand plausible redemption shocks, though tail risks remain if large funds face large redemptions. - Majority of funds have liquidity management tools (LMTs) and operational means to mitigate redemptions. - Recommendations for funds monitoring: - Establish regular data sharing on funds’ liabilities between AMF and relevant regulators. - Study behavior of IFs investing in other IF shares/units to assess amplification risks. - Empirical analysis of past dash‑for‑cash episodes to assess market impact amplification. ### Interconnectedness and contagion analysis - Interbank network and exposures: - Network built from COREP large exposure data (end‑2024) covering 10 French SIs (including 4 G‑SIBs) plus foreign SIs and G‑SIBs. - The 4 French G‑SIBs account for the bulk of bilateral credit exposures internally and with global and Euro Area SIs. - G‑SIBs are net providers of liquidity cross‑border but net borrowers from other French SIs. - Contagion stress test outcomes: - Scenario 1 (resident banks only, 25 connections): no bank fails. - Scenario 2 (adds cross‑border exposures, 71 connections): one smaller French SI fails in first round due to a large funding exposure withdrawal. - Contagion and Vulnerability Indices reported (examples): - Contagion Index Overall: Scenario 1 = 7.687; Scenario 2 = 3.579. - Vulnerability Index French Banks: Scenario 1 = 8.104; Scenario 2 = 6.399. - Vulnerability Index French and Int’l Banks (Scenario 2): 7.208 (credit 6.335; funding 0.873). - NBFI holdings: - Total debt securities, listed shares and fund shares held by banks, insurance companies and investment funds: EUR 4.9 trillion (BdF data). - Insurance companies: hold 43 percent of outstanding securities; within insurance holdings, nearly half are non‑resident securities. - Investment funds: hold about one third of outstanding securities; about half invested in foreign paper. - Domestic banks: account for one fourth of holdings. - Bank contagion model: LGD per claim assumed between 90 and 100 percent for interbank claims; fire sale discount factor δi = 30 percent; selling limit θi = unencumbered marketable securities net of pledges to ECB. ### Non‑financial corporates (NFCs) and households: stress‑test findings - Publicly listed NFCs: - Aggregate debt‑at‑risk (ICR < 1) increases from 6 percent (actual) to 17 percent in baseline, and to about 60 percent of total debt after two years in the adverse scenarios. - Firms with liquidity needs that increase indebtedness: baseline up to 65 percent of outstanding debt; adverse scenarios almost 80 percent. - Large unlisted firms and SMEs (ORBIS sensitivity): - At end‑2022, nearly one fourth of unlisted firms had ICR < 1. - Under stress: large firms share with debt at risk rises from 25 to 47 percent; SMEs: from 21 to 35 percent. - Corporate debt sustainability (debt‑to‑EBITDA): - Close to 30 percent of large firms had debt‑to‑EBITDA > 6 or negative in baseline; under stress increases to 43 percent. - Weighted by debt stocks, debt at risk rises from 44 to 64 percent. - SMEs: baseline distressed share 25 percent; increase under stress by 13 percentage points. - Households: - HFCS‑based household PDs (percent) for France: - 2023: Baseline 1.32 / Geopolitical 1.32 / Recession 1.32 - 2024: 1.32 / 1.32 / 1.32 - 2025: 1.36 / 1.41 / 1.33 - 2026: 1.36 / 1.40 / 1.32 - 2027: 1.36 / 1.38 / 1.30 - Mortgage default rates: highest default in recession scenario 0.7 percent (still below Spain’s current default rate 1.1 percent). - Household LGDs: aggregate 19 percent; residential housing loans 14 percent. - Borrower‑based policy counterfactuals: - Imposing LTV < 90 percent excludes 44 percent of housing loans in HFCS; DSTI < 32 percent excludes 24 percent. - PD reductions: - LTV limit PD decline: 0.07 to 0.12 percentage points. - DSTI limit PD decline: 0.22 to 0.38 percentage points. - CET1 capitalization gains (peak, after 3 years of recession): - LTV limits: 0.7 percent of RWAs. - DSTI limits: 1.6 percent of RWAs. - Conclusion: DSTI limits more effective than LTV limits in containing housing loan default risk in France. ### Sovereign securities market and debt management - Stock and issuance: - Total general government debt outstanding end‑2024: EUR 3.3 trillion; central government marketable debt securities: EUR 2.6 trillion. - Composition of central government marketable debt (percent of EUR 2.6 trillion): OAT 81%, OAT€i 6%, OATi 2%, Green OAT 3%, BTF 8%. - Outstanding nominal French debt securities: EUR 5.7 trillion. - Recent dynamics: - Outstanding stock of government securities increased by 40 percent over past five years and was smoothly absorbed. - Cumulative net debt issuance 2020–24: EUR 825 billion. - Projected cumulative net debt issuance 2025–29: EUR 938 billion. - Eurosystem holdings and roll‑offs: - PSPP net purchases cumulative: EUR 458 billion; PEPP: EUR 292 billion (combined ≈ 30 percent of medium‑ to long‑term outstanding). - Weighted average maturity of Eurosystem holdings: 6.3 years. - Simplifying assumption: about EUR 60 billion in securities could be rolled off annually over next 12 years (20 percent of 2025 gross financing needs EUR 300 billion). - Demand and investor base: - Over half of French government securities held by non‑residents. - Non‑residents hold about 52 percent of OATs (EUR 1,122 billion); 87 percent of BTFs (EUR 174 billion). - Euro Area investors held EUR 0.7 trillion; French insurance companies and pension funds held EUR 360 billion (Q4 2024). - Market liquidity and resilience: - 10‑year OAT‑Bund spreads ~70 bps; spread peaked at 85 bps by early January 2025. - Real yields around 2 percent; term premia risen but below European debt crisis levels. - Auctions: bid‑to‑cover ratio ≈ 3 times in 2024. - Average time to maturity ~9 years (8.5 years including BTFs). - Largest single gross issuance: EUR 42 billion (March 2024); buybacks in 2024: EUR 50 billion maturing 2025–27. - Risks and policy considerations: - Rising public debt and slow fiscal adjustment could raise risk premia and trigger adverse feedback loops affecting NFCs, banks and NBFIs. - Emerging risks: constraints on PDs’ balance‑sheet capacity; repo market scarcity; potential sovereign downgrades raising margins/haircuts. - Policy tools: AFT repo facility, reopening/tap issuances, securities lending by BdF, Eurosystem backstops (OMT, TPI, reinvestments). - Debt management cannot safeguard safe‑asset status without market confidence in a sustainable long‑term fiscal strategy. ### Key FSAP recommendations (summary) - Improve data quality and timeliness on interconnectedness, and on derivative and repo market data, and undertake related risk analysis for banks and markets. (ACPR, BdF, AMF) — ST - Improve liquidity monitoring through integration of liquidity stress in major currencies, and consider higher liquidity buffers to cover wholesale funding outflows within a two‑week horizon. (ACPR, BdF, ECB) — ST - Improve monitoring of investment fund redemption risk through data sharing on fund liability structures. (ACPR, BdF, AMF) — ST *Source: 1fraea2025007 - EXECUTIVE SUMMARY* ### EXECUTIVE SUMMARY __________________________________________________________________________ 8 ### EXECUTIVE SUMMARY ### MACROFINANCIAL BACKGROUND - Sections included: - A. Macro-Financial Landscape and Trends - B. Banking System Structure and Performance - C. Investment Fund Industry Structure and Recent Developments - D. Scope of the Systemic Risk Analysis and Scenarios - Relevant figures and tables cited in this theme: - Figures: 1. Government Debt Markets; 2. Debt Securities Markets; 3. Credit Conditions, House Prices and Private Debt; 4. Non-Financial Corporates; 5. Structure of Banking System; 6. Bank Performance Overview; 7a. Comparative Analysis: Aggregate Buffers; 7b. Comparative Analysis: P&L; 8–14 (investment funds and securities); 15. Scope of the Solvency Stress Tests; 16. Macroeconomic Scenarios for France. - Table: 1. 2025 Key FSAP Recommendations. ### SOLVENCY STRESS TESTS OF BANKS - Main components: - A. Stress Testing Approach and Macro-Financial Scenarios - B. Credit Risk Modelling and RWAs - C. Net Interest Income (NII) Modelling - D. Market Risk Modelling - E. Net Fees and Commission Income - F. Other Profit and Loss Items - G. Solvency Stress Test Results - H. Special Topic: Use of Solvency Stress Tests for Calibration of CCYB - Supporting materials and methodological detail: - Boxes: 1. Corporate Credit Risk Model; 2. Modeling LGDs; 3. Structural Model for Household Credit Risk; 4. Provisioning Under IFRS9 Accounting; 5. Interest Rate Pass-Through Econometric Models. - Figures: 17. Cross-Border Exposures: PDs for NFCs; 18. Scenario-Based Solvency Stress Tests; 19. Solvency Stress Tests; 20. Scenario-Based Solvency Stress Tests for G-SIBs; 21. Bank and Sovereign CDS Spreads; 22. Counterfactual Macroeconomic Scenarios; 23. Aggregate Capital Ratios in Each Scenario. - Tables: 2. Market Risk Scenarios; 3. Stylized IRRBB Template; 4. Specific Time Series Regressions for Models of Interest Rates; 5. Scenarios: Shocks to Interest Rates; 6. Large Exposures, as Percent of Tier 1 Capital; 7. Sectoral Credit Risk Sensitivity Analysis: NPL Ratios. ### SPECIAL TOPICS: STRESS TESTS OF NON-FINANCIAL PRIVATE SECTORS - Sections: - A. General Approach - B. Stress Tests of Non-Financial Corporations - C. Solvency Stress Tests of Households’ Balance Sheets - Supporting methodological content: - Boxes: 6. Micro-Simulation Model of Household Default. - Figures: 24. Scenario-Based Stress Tests of Publicly Listed NFCs; 25. Interest Coverage Ratio, Actual and Stressed; 26. Debt Stock Sustainability of Large Firms and SMEs, Actual and Stressed; 27. Comparative Analysis of Households’ Indebtedness; 28. Stress Tests of Households’ Balance Sheets; 29. Counterfactual Solvency Stress Tests. - Tables: 16. NFCI Panel Regressions; 19. Stress Test Matrix (STeM). ### LIQUIDITY STRESS TESTS OF BANKS - Sections: - A. General Approach and Scenarios - B. Structural Liquidity Risks - C. Liquidity Stress Test Results - D. Special Topic: “System-Wide” Liquidity Stress Tests - Supporting content: - Figures: 30. Analysis of the LCR; 31. Analysis of the Net Stable Funding Ratio; 32. Structural Funding Characteristics and Roll-Over Needs; 33. Bank Liquidity Stress Tests; 34. Reverse Cash-Flow Stress Tests. - Tables: 17. Parameters for LCR Scenario Analysis (Based on EBA COREP Templates); 18a–18c. Cash Flow Analysis Scenarios: Scenario 1, Scenario 2, Scenario 3. ### INTERCONNECTEDNESS AND CONTAGION ANALYSIS - Sections: - A. Interconnectedness - B. Bank Contagion Analysis - Supporting materials: - Figures: 35. Network of Bilateral Interbank Exposures; 36. Bank Contagion Model. - Tables: 8. Bank and NBFI Securities Holdings by Type of Issuer; 9. Bank Contagion Stress Test – Contagion and Vulnerability Indices. ### INVESTMENT FUNDS’ LIQUIDITY STRESS TESTS - Sections: - A. Introduction - B. Methodology and Analysis - C. Conclusions and Recommendations - Supporting empirical and scenario analysis: - Figures: 37. Assets Held by Bond Funds and Mixed Funds, end-2024; 38. Performance and Redemptions of Investment Funds and MMFs, 2007-24; 39. Indicator of Liquidity, French 10-year Government Security Price Bid-Ask Spread; 40. Securities Held by Bond, Mixed, and Money Market Funds in the Sample Funds; 41. Interest Rate and Credit Risk Shock Scenarios; 42. Distribution of Fund Liquidity Shortfalls under Scenario where 5 percent of the Funds Face Redemptions Shocks Similar to Levels Experienced during March 2020; 43. Change in Net Asset Value Under Historical Shock Events; 44. Decomposition of the Change in Net Asset Value, by Interest Rate Shock, Credit Shock, Redemption Shock, and Market Impact, by Liquidation Strategy; 45. Cash and Securities Liquidated Under Alternative Liquidation Strategies. - Tables: 10. Distribution of Redemption Flows during March 2020; 11. Asset Classified as Liquid; 12. Historical Extreme Events: Monthly Changes in NAV and its Components (Changes in Flows, Valuation, and Quantity) in Bond Fund, Mixed Fund, Investment Fund and MMF, 2007-2024. ### SPECIAL TOPIC: FINANCIAL STABILITY AND THE FRENCH SOVEREIGN DEBT SECURITIES MARKET - Sections: - A. Introduction - B. Recent Developments - C. Structure of the French Government Securities Market - D. Demand - E. Supply and Distribution - F. Secondary Market Liquidity - G. Sovereign Risk Management - H. Conclusions - Empirical and market structure materials: - Figures: 46. Public Debt, Fiscal Deficit, Primary Balance and Interest Payment; 47. Yield and Spread of French Sovereign Securities; 48. French Government Debt Securities Outstanding at end-2024, by Instrument; 49. Holders of General Government Debt; 50. Holdings of French Government Securities; 51. Domestic Banks Holdings of General Government Debt; 52. Bid-to-Cover Ratio in OAT Auctions; 53. Issue Size of Government Securities; 54. Measures of Liquidity in Government Securities Market; 55. Repo Transactions through LCH Clearnet S.A. (France); 56. Managing Refinancing Risk and Promoting Liquidity; 57. Market Valuation of Government Securities. - Tables: 13. Gross Issuance and Redemption of OATs and Net Purchases of OATs by the Eurosystem, 2015-24; 14. Selected Primary Dealers' Holdings of General Government Debt. ### RISK ASSESSMENT, METHODOLOGICAL TOOLS, AND SUPPLEMENTARY MATERIAL - Risk and policy tools and summaries: - Table: 15. Risk Assessment Matrix. - Boxes and templates across chapters for models and calibration (see Boxes 1–7 and Box 7. Structural Model for Repricing of Net Interest Income). - Figures and tables supporting historical scenarios, stylized templates, and stress-test matrices (including Table 19. Stress Test Matrix (STeM), Tables 18a–18c cash flow scenarios, and Tables 2–7 covering market and interest-rate risk). - Glossary of acronyms and definitions covering terms such as AC; ACPR; AE; AFS; AMF; BdF; CBC; CET1; COREP; CRE; CCyB; DSTI; EA; EAD; EBA; EBIT; ECB; ECL; EMDE; EU; FCI; FINREP; FSAP; FVOCI; FVPL; FX; GDP; G-RAM; G-SIB; GFC; HFCS; ICR; IF; IFRS; IMF; IRB; IRRBB; LCR; LGD; LTV; MFI; MMF; NACE; NBFI; NFC; NII; NPL; NSFR; OECD; O-SII; PD; PiT; P&L; RAM; RoA; RoE; RRE; RWA; SME; SI; SSM; ST; STA; STE; STeM; SSyRB; TTC; TR; UB; WEO. *Source: 1fraea2025007 - EXECUTIVE SUMMARY* ### EXECUTIVE SUMMARY ### EXECUTIVE SUMMARY ### Financial sector resilience and prevailing risks - The French financial sector has proven resilient to the stress events of the last five years but faces headwinds from domestic and global uncertainty. - Sovereign debt markets have functioned well; large debt issuance continues to be smoothly absorbed by a deep and well diversified buyer base. - Banks have pre-funded part of their upcoming roll-over needs. - Credit growth has moderated and the housing market is undergoing an orderly adjustment while household and non-financial corporates (NFC) debt remain elevated. - The stability of the French financial sector is an important element of strength amid domestic political uncertainty and rising global geoeconomic risks. ### Banking sector structure, performance, and vulnerabilities - Large and internationally active banks have high capital and liquidity buffers and have adjusted to the increase in interest rates. - Six major bancassurance conglomerates include four Global Systemically Important Banks (G-SIBs) with important cross-border exposures and market activities. - The banking system features low credit risk from conservative lending practices but limited profitability. - For housing loans: - High-quality borrowers, long-term fixed-rate loans, and loan guarantee schemes have resulted in low credit losses and high resilience to interest rate shocks. - Profits on home loans are low in a very competitive domestic retail market and are often cross-subsidized by other products. - During the rising rate environment net interest margins compressed as fixed-rate housing and, to a lesser extent, corporate loan books saw slow repricing while funding costs on largely floating-rate liabilities rose. - Non-financial corporates are vulnerable due to high leverage levels. - SME defaults have risen as Covid-era measures finished rolling off but remain at manageable levels. ### Solvency stress tests and macroprudential analysis - Solvency stress tests show no banks breach their minimum capital requirements (the Basel minimum CET1 ratio plus Pillar II requirements) under either the geopolitical or the recessionary adverse scenarios. - Four banks, including G-SIBs, dip into their additional capital buffers (the capital conservation buffer plus G-SIB and O-SII buffers). - Aggregate system-wide gap to the capital requirements including buffers is: - 1.2 percent of risk weighted assets after three years in the geopolitical scenario. - 0.8 percent of risk weighted assets after three years in the recession scenario. - Main drivers of additional capital depletion in adverse scenarios: - Credit risk—mainly from vulnerable non-financial corporates. - Market risk—due to market activities of banks, and fees and commission income. - Sensitivity analysis: - Banking system’s exposure to concentration risks is small after accounting for credit risk mitigation measures (CRM). - Sectoral risks are limited. - Sovereign-bank nexus channels of shock transmission appear contained among French banks. - Macroprudential counterfactual analysis: - Solvency stress tests under a range of macroeconomic scenarios indicate the current combination of high precautionary buffers and the CCyB can in aggregate absorb a moderate macroeconomic shock. - Post-shock buffers would likely be large enough to enable banks to continue lending following a range of shocks, although impacts may be heterogeneous across banks. - Counterfactual analysis using the 2021 ECB Household Finance and Consumption Survey shows macroprudential limits on DSTI are more effective than limits on LTV in containing housing loan loss risk in France—the finding aligns with the approach of the French authorities. ### Liquidity and funding resilience - Aggregate LCRs (in all currencies) and in euros have remained stable at around 150 percent on average every month since 2020. - All USD LCRs have been above 100 percent recently, although some monthly LCRs in USD were volatile in the past. - Funding sources are well diversified on average and asset encumbrance is low. - Under scenarios with run-off rates higher than under Basel III and/or with valuation losses on HQLA (which could occur due to a sell-off of fixed income securities by European investment funds), the aggregate LCR in all currencies and in euros remain above the 100 percent requirement. - However, several banks have LCRs that fall below the requirement in the stress scenario. - Cash flow stress tests: - Banks can withstand significant liquidity outflows up to one month under several scenarios. - One bank would experience a small cash shortfall within 2 weeks of the shocks. ### Investment funds and redemption risk - Stress testing of open-ended bond funds, mixed funds, and money market funds suggests they have sufficient liquidity to withstand plausible redemption shocks. - Due to high buffers of liquid assets in these funds, there is only a small additional price impact (beyond the initial interest rate or credit shock) from redemptions leading to sales of less-liquid assets. - A majority of funds have liquidity management tools to mitigate redemptions and established means to prevent disorderly outflows, further buttressing resilience and dampening possible market impact from asset sales. ### Interconnectedness and non-bank financial institutions (NBFIs) - French SIs and NBFIs maintain an extensive network, whereas contagion risks are low. - The four French G-SIBs account for the bulk of bilateral credit exposures among themselves and with other systemic global and euro area banks. - System-wide liquidity and funding risks are low, with only one smaller bank experiencing severe liquidity stress in a contagion risk analysis. - Among NBFIs: - Insurance companies are the largest institutional investors, accounting for close to half of securities holdings. - Insurance companies hold one-fourth of their portfolios in investment funds that are substantially invested in bank debt. ### Non-financial corporates and households - Non-financial corporates are vulnerable to adverse macroeconomic shocks consistent with bank stress tests: - Corporate debt at risk of publicly listed NFCs would increase notably, and to high levels, under the geopolitical and recession adverse scenarios. - Cash shortage of publicly listed NFCs would increase under those scenarios, leading NFCs to increase borrowing. - Sensitivity analysis for non-listed large firms and SMEs confirms the corporate risk assessment findings. - Households: - Households across income levels have accumulated significant debt, but stress tests suggest the risk of default remains limited, including in adverse macroeconomic scenarios. ### Sovereign securities market and debt management - The French sovereign securities market forms the bedrock for financial market stability and continues to function well despite a large increase in issuance since the previous FSAP. - Over the past five years, the outstanding stock of government securities increased by 40 percent but was smoothly absorbed by the market. - The sovereign debt market is deep and well diversified, with over half held by non-resident investors. - Banking sector holdings of government debt are mostly low and near the Eurozone average of 6 percent of assets; banks with larger exposures mainly hold French sovereign bonds and loans to maturity. - Occasional deterioration in market liquidity has not turned into dysfunction during market stress, while repricing of risk has occurred with investor rotation. - Emerging risks and policy tools to monitor/augment: - Secondary market liquidity support by the debt management office (including through its repo facility, reopening and tap issuances of off-the-run securities) may become more critical—but care is needed to avoid distorting private transactions and removing arbitrage opportunities. - Presence of price sensitive investors can support the market when pricing is attractive but could also lead to rapid exits when conditions deteriorate. - Investor diversification and progress on the ongoing pension reform could expand and stabilize the investor base. - Debt management alone cannot safeguard the safe asset status of French government securities without market confidence in a sustainable long-term fiscal strategy. ### Macro-financial background and key metrics - Growth is projected at 0.6 percent in 2025, down from 1.1 percent in 2024, as policy uncertainty amid domestic political fragmentation and rising geoeconomic tensions weighs on confidence and activity. - 10-year OAT-Bund spreads remain around 70 bps (about 25 bps higher than in early June). - The disinflationary process remains on track. - Credit and housing: - Credit growth moderated with household lending stagnating and the credit-to-GDP gap turned negative. - Housing loan issuance declined sharply in 2023-24 and average DSTI increased. - Residential property prices declined by 5 percent y/y and stabilized after a trough in 2024:Q1. - NFC debt is among the highest in Europe, with about one quarter of firms showing inadequate debt service capacity. - Banking concentration and international activity: - The French banking system is large and highly concentrated, with 6 banks accounting for about 96 percent of total banking assets and 292 percent of GDP. - Four of the six major banks are G-SIBs. - BNP Paribas and Société Générale have the majority of their credit exposures outside France; for Crédit Agricole it is 35 percent. - Exposures to sovereign fixed income securities for the 6 largest banks average 5.1 and 5.5 percent of assets respectively, except for La Banque Postale Group. ### Key FSAP recommendations (summary) - Work with relevant European authorities to improve data quality and timeliness on interconnectedness, and on derivative and repo market data, and undertake related risk analysis for banks and markets. (ACPR, BdF, AMF) — ST - Improve liquidity monitoring through integration of liquidity stress in major currencies, and consider higher liquidity buffers to cover wholesale funding outflows within a two-week horizon. (ACPR, BdF, ECB) — ST - Improve monitoring of investment fund redemption risk through data sharing on fund liability structures. (ACPR, BdF, AMF) — ST *Source: 1fraea2025007 - EXECUTIVE SUMMARY* ### 5. Performance of the French banking system has held up well in recent years in a ### 5. Performance of the French banking system has held up well in recent years in a ### Banking system performance: summary findings - Performance has "held up well" in a context of rapidly rising interest rates, but profitability has somewhat declined across banks. - Before the pandemic, all 6 banks had a Return on Assets (ROA) above the EU average, despite lower-than-average Net Interest Margin (NIM). - In 2023, all banks had a ROA somewhat below the EU average and below the average of the other G-SIBs. - French banks’ Tier One ratios are on average above the G-SIB peer average but below the average of EU peers. - Leverage ratios (ratios of capital to total assets) are below the EU peer average and the G-SIB peer average. - LCRs are on average above the average for the G-SIB peers but below the EU peer average; NSFRs are below the EU and the G-SIB peer average. - The below-average ROA reflects a combination of: - a moderate decline of French banks’ profitability, and - an increase of profitability among EU and G-SIB peers. - NIM of French banks has been on the low side compared to peers. - Net fees and commission income has accounted for a share of income above EU and global peers’ average. - Cost efficiency appears to have room for improvement: the cost-to-income ratio is moderately above the average for EU and for G-SIB peers for several banks. - LBP appears as an outlier with a very low cost-to-income ratio, especially after taking over CNP Assurance on its balance sheet. ### Capital, liquidity, profitability metrics (selected details) - Capital requirement note: "The capital requirement does not include the systemic risk buffer, HCSF decision July, 31, 2023." - Comparative metrics (aggregate buffers and P&L) highlighted in figures: - Total Capital Adequacy Ratio and Core Tier 1 Capital Ratio presented for the major banks (visual comparison across 2019 and 2023/2024 in source figures). - Tier 1 Capital Ratio: French aggregate below EU average, but above G-SIBs’ average (Figure 7a). - Leverage Ratio: below peers’ averages (Figure 7a). - Liquidity Coverage Ratio (LCR): above Basel requirement, above G-SIB peers but below EU peers (Figure 7a). - Net Stable Funding Ratio (NSFR): above requirements but below peers (Figure 7a). - Return on Assets (ROA): fell to below peers’ average by 2023/2024 (Figure 7b). - Net Interest Margin (NIM): on the low side since before the pandemic (Figure 7b). - Cost-to-Income Ratio: has declined but remains above some peers’ averages (Figure 7b). - Net Fees and Commissions: now account for a larger share of income than peers’ average (Figure 7b). - Note: Credit Mutuel and Bofa Securities Europe SA are using the latest 2023 data in the comparative charts. ### Investment fund industry: structure and recent developments - The asset management industry comprises investment funds excluding money market funds (IFs), and money market funds (MMFs). - At end-2024: - there were 13,511 IFs and MMFs with a cumulated net asset value (NAV) of almost EUR 2 trillion (60 percent of GDP). - MMFs numbered 120 and manage on average a larger amount of assets compared to IFs. - Market rankings (fund domiciliation, 2023): France ranked 10th largest in the world and 4th in the Euro Area after Luxemburg, Ireland, and Germany. - Consolidation: - In 2023, the top 5 firms accounted for 45 percent of the market share in NAV in France (compare: top 5 in Germany 58 percent, U.K. 41 percent). - Top 10 firms are affiliates of French banks or insurance companies; Amundi (affiliate of Credit Agricole) significantly dominates the industry. - Recent transaction: purchase of AXA IM by BNP Paribas AM in December 2024 noted as the latest consolidation move. - Policy/environmental drivers: - Consolidation is likely to continue in preparation for a potential Savings and Investments (SIU) Union within the EU. - Tax incentives are offered to enhance participation in occupational and private pensions. ### Investment fund holdings, interconnections, and concentrations - Residency and issuer composition (end-2024): - Almost 90 percent of investors in IF shares/units are resident investors (75 percent for MMFs). - About half of IF assets (60 percent for MMFs) are securities issued by French entities. - 30 percent by resident issuers in the EA (20 percent for MMFs). - 20 percent by issuers in the rest of the world (20 percent for MMFs). - IF and MMF balance sheets (end-2024): - IF and MMF asset holdings: IFs collectively held EUR 645 billion in French debt securities and EUR 477 billion in French equities. - Nominal French debt securities outstanding: EUR 5.7 trillion. - Total market capitalization of the French equity market: EUR 2.7 trillion. - Debt securities market composition: 30 percent MFI issuance, almost half by the French sovereign (EUR 2.8 trillion). - Equity market: 90 percent listed shares of non-financial corporations; MFIs account for about 5 percent. - Interfund exposures: - A fifth of the EUR 1.6 trillion in assets held by IFs are invested in shares of other IFs and MMFs. - Assets managed via funds-of-funds rose from EUR 320 billion in Q4 2019 to EUR 390 billion in Q3 2024. - Of funds-of-funds holdings, almost half (EUR 189 billion) are IF and MMF shares/units issued by domestic MFIs. - Mixed funds and "other" categories have the highest propensity to invest in other funds. - Risk note: fund-of-fund model can diversify but also amplify risk via second-round liquidations; a referenced German study finds exclusive focus on common asset holdings can underestimate vulnerabilities. - Liability and investor concentration: - Insurance companies are the largest investors in IF shares/units, followed by households, and non-MMF investment funds. - For MMFs, insurance companies are also the largest investors, followed by non-MMF IFs and non-financial corporations. - MMF investments are concentrated in bank debt securities and shares in MMFs; net deposits account for 20 percent of MMF assets. - An AMF study found insurers invest mostly in funds managed by asset managers belonging to the same conglomerate; asset managers almost exclusively manage investments from insurers in the same group. - Supervisory data gaps: AMF does not have a systematic database on holders of liabilities at the fund level; reporting of liabilities is not required by law or regulation. Information on net deposits and loans received and their counterparty is unavailable—this information is critical for monitoring redemption risk, stress testing, and interconnectedness analysis. ### Systemic vulnerabilities, scenarios, and stress-testing scope - Key vulnerabilities highlighted: - High and rising public debt. - Growing balance sheet weaknesses in the non-financial sector amid political fragmentation and potential social unrest. - Slow fiscal adjustment could lead to higher risk premia, market repricing, and adverse macro-financial feedback loops affecting NFCs, banks and NBFIs. - Continued weak pace of reforms would weaken potential growth and further deteriorate public and private balance sheets, raising credit risk and corporate default rates. - Global risks: - Rising and more uncertain global risks from geoeconomic fragmentation, commodity and trade shocks, tighter financial conditions, and systemic stability risks. - Leveraged firms and households have been under pressure from higher debt service costs from the tightening period 2022-2023; higher interest rates would further adversely affect debt dynamics, including for the sovereign. - Potential widening of sovereign spreads could raise borrowing rates and heighten credit risk; deteriorating asset quality, lower lending, stress in core financial markets, and contagion from NBFIs are key risks. - FSAP stress-test coverage and scenarios: - The FSAP assessed resilience of interconnected financial sectors and private non-financial sectors’ balance sheets against the 2025 January WEO baseline and two severe but plausible adverse macroeconomic scenarios common to the Euro Area FSAP. - Stress tests consider impacts of non-financial private sectors on bank solvency and interbank contagion risks; liquidity risks assessed for 7 SIs, for investment funds, and for non-financial corporates against several market stress scenarios, including system-wide interactions from investment funds to banks. - Baseline macroeconomic scenario: simulated against the 2025 January update of the WEO (consistent with Euro Area FSAP baseline); this baseline has not been updated since and therefore does not take into account subsequent macro-economic developments. - Geopolitical adverse scenario: tail-risk scenario of deepening geoeconomic fragmentation, global commodity and trade shocks, inflationary trade and tariff shocks ("trade wars"), global loss of confidence, demand shocks, tighter financial conditions, asset price declines, and a "higher for longer" inflation environment with slowing growth and rising short-term interest rates. - Recession scenario with sovereign stress: combines global demand shocks, loss of confidence, tightening financial conditions and domestic fiscal shocks that raise government borrowing costs and term premia, causing a recession and asset market declines while structural shocks to productivity lower potential output growth; shocks to France’s sovereign spreads are considered intermediate among Euro Area countries. *Source: 1fraea2025007 - 5. Performance of the French banking system has held up well in recent years in a, IMF staff (source PDF).* ### 15. In addition, the resilience of banks is assessed against one-off severe market shocks ### 15. In addition, the resilience of banks is assessed against one-off severe market shocks ### Market-shock design and application - Two sets of one-off market shocks are applied at the beginning of the first year of each adverse solvency stress-test scenario and the capital impact is carried over in the two following years of the scenarios; the shocks are not reversed during the 3-year macroeconomic scenarios. - Market shocks are narrative-aligned with each adverse macroeconomic scenario: - Geopolitical-market shocks: increase in interest rates (particularly at the short end of the EUR yield curve), commodity prices and credit spreads, and a drop in equity prices. - Recessionary-market shocks: drop in commodity and equity prices, an increase in the credit spread of mid-risk Euro Area sovereign (which includes France) and of high-risk Euro Area sovereigns, while the short end of the EUR yield curve remains stable. - Both sets of shocks include similar volatility shocks to equities and interest rates. ### Stress-testing approach and macro-financial scenarios - The FSAP stress test is a top-down exercise with projections generated by IMF in-house models developed by the FSAP team; it differs from the constrained bottom-up EBA-SSM-ECB exercise in data granularity and shock calibration. - The stress test considers credit risks, interest rate risks and market risks. - The core bank stress-test results for France utilize adverse scenarios and models from the contemporaneous EA FSAP. - The two adverse scenarios entail very large macro shocks: - Shocks correspond to a 2.4 (respectively 2.7) deviation of real GDP growth from the baseline in the geopolitical scenario (respectively recession scenario). - They result in a decline of real GDP of about 5.3 percent (respectively 6.3 percent) after two years in the geopolitical scenario (respectively recession scenario). - Short-term interest rates: increase by 1.9 percentage point in the first year in the geopolitical scenario; decline by 0.6 percentage point in the recession scenario. - Long-term interest rates spike at 5.5 percent in the geopolitical scenario and at 4.8 percent in the recession scenario in 2024. - Term spreads widen by up to 2.1 percent in the recession scenario (reflecting sovereign stress). - Unemployment rate increases to 10 percent (respectively 13.5 percent) after 3 years in the geopolitical scenario (respectively recession scenario). - Baseline projections are based on October 2024 WEO; adverse scenarios as of November 2024. ### Short-term market stress scenarios and calibration - Two market-distress scenarios were calibrated to capture high-frequency market price and volatility movements using an Expected Shortfall approach; narratives match the geopolitical and recession adverse macro scenarios. - The geopolitical-market scenario: increase in interest rates (particularly short-end EUR), commodity prices and credit spreads, and a sharp contraction in equity prices. - The recessionary-market scenario: drop in commodity and equity prices, larger increase in credit spreads of mid- and high-risk Euro Area sovereigns, short-end EUR yield curve muted. - The market-test scenario is calibrated with specific shocks across commodities (CM), credit spreads (CR), equities (EQ) and interest rates (IR), and includes volatility shocks to equities and interest rates. ### Coverage, data, and scope of solvency stress tests - The solvency stress tests cover the largest 7 SIs with accounts under IFRS9 and included in the EBA stress tests; the 7 SIs account for about 96 percent of bank assets. - The stress tests used balance sheets and P&L statements of these SIs as of the end of 2024. - Data sources: ECB-SSM FINREP and COREP templates, STE files on IRRBB, market sensitivities, and large exposures. - Historical quarterly data back to 2014 were constructed on NPL ratios and recovery rates for loans to NFCs by bank and NACE economic sector for robustness. - Stressed foreign exposures considered (in addition to domestic): Italy, Belgium, Spain, the UK and the US; both private-sector and sovereign exposures included. ### Modeling assumptions and methodology - Balance-sheet approach at consolidated level; based on accounting data (IFRS9) and regulatory capital ratios. IRB and STA portfolios consolidated at bank level. - IFRS9 staging and provisioning: - Stage 1: provisioned on a 12-month horizon. - Stage 2 and Stage 3: provisioned with a life-time horizon. - Transition matrices (TRx-y) estimated where possible; in absence of long historical series, transition flows are estimated based on the beta-linking approach. - Starting-point PiT PDs and TTC PDs are those reported by each SI for IRB portfolios; STA portfolios’ PDs estimated based on recent historical reported transition matrices. - Static balance-sheet assumption: no write-offs and new originations equal maturing loans; nominal balance sheet remains constant. Consequence: defaulted exposures accumulate, producing conservative (upper-bound) estimates of capital ratio declines. - Two hurdle rates for CET1 capital ratio performance assessment: - Bank-specific minimum requirement = Basel III common 4.5 percent minimum plus bank-specific Pillar 2 Requirements. - Second hurdle adds the capital conservation buffer of 2.5 percent, and the systemic buffers (G-SIB buffer, O-SII buffer, and any sectoral buffer). ### Credit-risk modelling for non-financial corporates (NFCs) - Satellite models for non-financial sector exposures use the MCM Corporate Stress Test methodology (Tressel and Ding (2021)). - Model features: - Firm-level dynamic panel regressions of sales growth, return on assets, leverage and interest-coverage ratio on real GDP growth and an FCI from the GFSR. - Accounting identities project earnings, sales, leverage and ICR at firm level over a three-year horizon. - Interest expenses evolve with firms’ historical funding mix and scenario-specific short- and long-term corporate borrowing rate shifts; interest rate shocks applied by debt maturity buckets. - Simulated firm-level indicators are mapped into one-year forward PDs using a Moody’s matrix (benchmarked to historical default rates in the US); aggregated PDs are rescaled to banking-system aggregate PDs on loans to NFCs (based on EBA Risk Parameter Statistics for 2024:Q3). - Model expanded to incorporate 2023 balance-sheet data and roughly 23,000 nonfinancial firms covering 42 countries. - PD outcomes: - PDs increase even in the baseline scenario by up to 170 percent in the second year of the scenario. - In the two adverse scenarios, PDs increase by about 270 percent in the second year, reaching 4.3-4.4 percent. - PDs for NFCs were also constructed for cross-border exposures; PDs reach highest levels for loan exposures to NFCs of Belgium and Italy. ### Loss-given-default (LGD) modelling and key statistics - LGDs modelled using a methodology linking LGDs to PD fluctuations and an initial bank-level LGD by geographical segment (Frye and Jacobs (2012) Vašíček-type approach). - LGD starting point for 2024:Q4 obtained from individual bank supervisory reporting. - Aggregate LGDs for French NFCs from EBA Risk Parameter Statistics as of 2024:Q4 stand at 36.1 percent. ### Additional modelling and scope notes - Market risk analysis was conducted against two short-term market stress scenarios aligned with macro scenarios and calibrated to capture high-frequency market price and volatility movements. - The solvency stress tests considered largest foreign exposures of French SIs in addition to domestic exposures, including private-sector exposures (NFCs, households, financial institutions) and sovereign exposures. - Stress tests follow IFRS9 staging and provisioning conventions, making credit-risk modelling more complex than under the incurred-loss approach. *Source: IMF staff (France FSAP solvency stress tests, end-2024 data and November 2024 adverse scenarios).* ### 26. To achieve consistency with the Euro Area FSAP, the model of households’ PDs for EU ### 26. To achieve consistency with the Euro Area FSAP, the model of households’ PDs for EU exposures is based on the Euro Area FSAP semi-structural approach (Box 3). ### Household PD model and approach - Model basis and data: - Relies on the 2021 ECB Household Finance and Consumption Survey (HFCS). - HFCS includes 83,000 households and 200,000 personal files across 22 countries (20 EA countries, Czech Republic, and Hungary). - Uses a matching procedure to “age forward” the 2021 vintage to 2024 and estimate “durable” consumption. - Approach builds on Valderrama et al (2023). - Modeling scope and outputs: - Forecasts household and personal financial conditions, probability of falling into arrears, and likelihood of credit impairment (arrears>90 days) under FSAP baseline and adverse scenarios. - Battery of logistic regressions at country level identify the financial indicator and the threshold of distress that significantly increase probability of default. - Montecarlo simulations of unemployment shocks at the individual level within the household (controlling for employment status and type of contract), accounting for unemployment benefits. - Maturing loans are replaced by new loans with same DTI at origination and new issuances are repriced at prevailing market rates. - The PDs resulting from the modelling choice result in a relatively flat dynamics of PDs. - PDs reported for France (PDs, in percent): - Baseline scenario / Geopolitical scenario / Recession scenario - 2023: 1.32 / 1.32 / 1.32 - 2024: 1.32 / 1.32 / 1.32 - 2025: 1.36 / 1.41 / 1.33 - 2026: 1.36 / 1.40 / 1.32 - 2027: 1.36 / 1.38 / 1.30 ### Structural model for household credit risk (Box 3) - Four-step credit risk model using HFCS microdata: 1. Forecast household balance sheets, payments, income, and consumption to project ‘vulnerable’ households drawing on scenario macrofinancial projections; perform Montecarlo simulations of unemployment shocks at person level and account for unemployment benefits for unemployed individuals (at around 10 percent of initial income). 2. Estimate link between being financially vulnerable and default risk (PD). Default proxied by arrears over 90 days (stage 3 loans) or less than 90 days (stage 2 loans). Run logistic regressions at individual household level controlling for household income tercile, savings ratio, wealth ratio, age, gender, education, household size, number employed in household, loan to value ratio of main residence, credit constraints, source of income, and family / public financial assistance. 3. Run a horse race across adjusted DSTI thresholds; best performing indicator is a cost-of-living adjusted debt service to income (DSTI) ratio including debt service, essential consumption (food and energy cost) and rents. For most countries, the relative increase in PD is highest when adjusted DSTI exceeds 70 percent of disposable income (“overburdened” household). 4. Project share of banks’ retail loan portfolio with credit default (stage 3) or credit event (stage 2) by forecasting migration of loans held by overburdened households under each scenario. Maturing loans replaced by new loans with DTI at origination; floating rate mortgages reset over life of loan; new issuances repriced at prevailing market rates. - Specific illustrative finding: - When French households are overburdened, the rate of being on arrears (late payments over 90 days) increases from 1.14% to 2.30%; the rate of being on arrears (late payments below 90 days) increases from 3.7% to 6.6%. ### LGD and collateral valuation for households - Household LGDs derived from FSAP methodology combining information on LTVs and collateral values for residential properties reported in supervisory files. - In France: - Aggregate LGDs for exposures to households are relatively low at 19 percent. - LGDs for residential housing loans are 14 percent. - About 2/3 of French residential housing loans benefit from a guarantee; in the event of default, housing loans are taken over by a guarantor (such as Credit Logement) which reimburses the value of the loan to the bank. - Note: losses mainly incur during the transition period between arrears and transfer to guarantor’s balance sheet, and occasionally due to administrative origination mistakes. ### PDs for sovereign, financial and corporate bond exposures - PDs estimated based on scenarios’ projections of sovereign bond yields term spreads and corporate spreads. - Following Euro Area FSAP, a Merton-based transformation converts spread between 10-year sovereign yields and the short-term rate into a PD proxy. - Residual maturities obtained from supervisory files. - Assumed LGD of 45 percent for sovereign and corporate fixed-income exposures (as in many recent FSAPs). - Credit risk estimated on total exposure of each bank (economic approach). ### IFRS9 transition matrices, provisioning and lifetime PDs - Supervisory inputs: - SIs report PiT IFRS9 transition rates and TTC PDs and LGDs for IRB banks. - Adverse scenario uses TTC LGDs and PDs provided by each bank; baseline uses TTC LGDs and shocks in adverse scenario for household exposures according to RRE price projections. - Provisions and expected credit losses: - Under economic approach and IFRS9, provisions computed based on expected lifetime loss of new net flows into Stage 2 and 3 during each period. - Lifetime horizon truncated at 5 years maximum for projecting PDs beyond scenario horizon. - Residual probability of default 푃푃퐿퐿* defined as conditional probability given non-default until previous period; r is a discount factor (such as short-term interest rate). - For each exposure class, transition matrix evolution linked to projected PDs using beta-linking approach with elasticities 훽 of transition rates w.r.t. PDs (e.g., ∆T12 = β12 × ∆PD). - Bank-by-bank starting point annual transition matrices constructed from FINREP supervisory templates based on recent historical transition matrices. - Outcome: projected stocks of exposures in Stages 1–3, required provisions, and dynamics of capital stock. - Provisioning mechanics under IFRS9 (Box 4): - Stock of provisions equals expected credit losses for S1, S2, S3 exposures. - S1 stock = T1→3 × LGD × S1 (annual expected losses). - S2 stock = sum over V = t+1 to t+M of (T2→3_V × LGD_V × S2_{V-1}) / (1+r)^{V−t}, where M is lifetime horizon and r is discount rate. - S3 stock = LGD_t × S3_t (non-recoverable part of defaulted exposures). - Effective maturity assumptions: 7 years for mortgages, 5 years for loans to NFCs, and 3 years for non-mortgage retail loans. ### Credit risk weighting, RWAs and standardized vs IRB treatment - IRB portfolios: - ASRF model for unexpected losses implemented for different exposure types (Basel III). - RWAs subject to PDs and LGDs, provisions, and credit conversion factors for off-balance sheet items. - Regulatory TTC PDs calibrated through scenario as weighted average of PiT PDs for each year of scenarios (weighted by 0.2) and respective TTC PDs of previous year (weighted by 0.8, starting with reported COREP TTC PDs). - Regulatory downturn (DT) LGD = max(reported DT LGD at period 0, estimated PiT LGD). - Standardized (STA) portfolios: - Credit risk charges estimated using density of credit RWA at cut-off date (end of 2024). - Non-performing exposures subtracted from time 0 RWA assuming risk weight of 100 percent. - Ratio of performing RWA to performing exposures calculated; performing exposures multiplied by this ratio and total non-performing exposures added multiplied by non-performing exposure RWA density (assumed 100 percent). - Calculations performed bank-by-bank to reflect individual RWA profiles. ### Net Interest Income (NII) modelling and interest rate pass-through - Methodology: - Semi-structural repricing gaps methodology combining econometric models at portfolio-segment level and maturity “repricing” ladders for each bank. - Identical to Euro Area FSAP methodology. - Econometric models for interest rates on new business (loans and deposits) use Euro Area MFI Interest Rate Statistics to quantify pass-through from market rates to deposit and lending rates. - Use bank IRRBB repricing structure to estimate timing of transmission of market rate shocks to deposit and lending rates by instrument and counterparty. - Combine pass-through and timing at bank-instrument level to simulate interest income and interest expense under baseline, geopolitical, and recession scenarios consistently across countries and bank business models. - Additional assumptions: - For fixed-income securities holdings, market borrowings and other wholesale funding: interest rates move one-to-one with short-term interest rates of each scenario (pass-through of one) and assumed uncorrelated with long-term sovereign bond yields. - NPLs assumed not to pay interest to banks (this lowers NII in adverse scenarios versus baseline). - Data sources for NII modelling: - Most recent bank level NII from FINREP template F 16.01. - Exposures and liabilities by segment and country of counterparty from FINREP templates F 20.04 and F 20.06. - Repricing ladder from STE templates for IRRBB and supervisory IRRBB module template J 05.00, complemented with COREP C 66.01 when needed. - Econometric pass-through models (Box 5): - Regressions run at country level for each portfolio segment at quarterly frequency for 2000:Q2-2024:Q4 (exception: consumer credit model starts 2023:Q2). - Dependent segments: mortgages, non-mortgage household credit, NFC loans on asset side; household sight deposits, household term deposits, NFC sight deposits, NFC term deposits on liability side. - Findings: very low pass-through for sight deposits (especially household sight deposits), significant pass-through for loans to NFC, consumer credit, and term deposits. - Regressions include ∆Euribor, ∆10Y sovereign yield, and real GDP growth (contemporaneous and lagged) as explanatory variables. - Scenario shocks to interest rates (Table 5, in percentage points): - Baseline Scenario (2025 / 2026 / 2027): - Housing Loans: -0.4 / -0.1 / 0.0 - Loans to NFCs and Consumer Loans: -0.6 / 0.0 / 0.1 - Other Interest Earning Assets: 0.0 / -0.4 / -0.4 - Sight Deposits HHs: 0.0 / 0.0 / 0.0 - Sight Deposits NFCs: -0.1 / 0.0 / 0.0 - Term Deposits HHs: -0.7 / -0.1 / 0.1 - Term Deposits NFCs: -0.8 / 0.0 / 0.2 - Other funding: 0.0 / -0.4 / -0.4 - Geopolitical Scenario (2025 / 2026 / 2027): - Housing Loans: 0.3 / 0.0 / -0.8 - Loans to NFCs and Consumer Loans: 0.3 / 0.0 / -0.8 - Other Interest Earning Assets: 1.9 / -0.6 / -1.5 - Sight Deposits HHs: 0.0 / 0.0 / 0.0 - Sight Deposits NFCs: 0.0 / 0.1 / -0.1 - Term Deposits HHs: 0.1 / 0.1 / -0.6 - Term Deposits NFCs: 0.4 / 0.2 / -0.7 - Other funding: 1.9 / -0.6 / -1.5 - Regression sample output (selected coefficients and statistics, Table 4): - Observations: 60 (for each reported regression). - R2 values (by dependent variable): Housing Loans 0.718, Loans to NFCs 0.897, Consumer Credit 0.632, Sight Deposits HHs 0.191, Sight Deposits NFCs 0.559, Term Deposits HHs 0.685, Term Deposits NFCs 0.855. - Example coefficients (∆Euribor (t), ∆Euribor (t-1), ∆10Y Sovereign Yield (t), 10Y Sovereign Yield (t), Real GDP growth (t), Real GDP growth (t-1)): - Housing Loans: ∆Euribor (t) = -0.119, ∆Euribor (t-1) = 0.406***, ∆10Y Sovereign Yield (t) = 0.0802*, 10Y Sovereign Yield (t) = 0.181***, Real GDP growth (t) = -0.592, Real GDP growth (t-1) = 0.300. - Loans to NFCs: ∆Euribor (t) = 0.123, ∆Euribor (t-1) = 0.590***, ∆10Y Sovereign Yield (t) = 0.0676, 10Y Sovereign Yield (t) = -0.0400, Real GDP growth (t) = 2.768***, Real GDP growth (t-1) = 1.602***. - Consumer Credit: ∆Euribor (t) = -0.166, ∆Euribor (t-1) = 0.498***, ∆10Y Sovereign Yield (t) = 0.155*, 10Y Sovereign Yield (t) = 0.177*, Real GDP growth (t) = 3.801***, Real GDP growth (t-1) = 1.054**. - Note on significance: *** p<0.01, ** p<0.05, * p<0.1; robust standard errors. *Source: IMF Staff* ### 38. In a second step, shocks to interest rates summarized in Box 5 are applied to repricing ### 1fraea2025007 - 38. In a second step, shocks to interest rates summarized in Box 5 are applied to repricing ### Interest-rate shocks and Net Interest Income (NII) simulation - Shocks to interest rates summarized in Box 5 are applied to repricing ladders obtained from the IRRBB templates for each bank. - Portfolios are simulated under the assumption of a static balance sheet: portfolio compositions for interest-rate bearing assets and liabilities remain constant over the three years of the scenarios. - Shock application rules: - Shock to interest income for assets (respectively expenses for liabilities) in each time bucket apply to buckets that reprice during that year. - Buckets that have not repriced receive the interest income for assets (respectively bear the interest expense for liabilities) from the time of their origination. - For each year of the scenarios, interest income and expenses are aggregated across repricing buckets. - Aggregate NII outcome: - At the aggregate level, NII declines by “only” 0.3 percent of RWAs in the two adverse scenarios relative to the baseline. ### Accrual versus non-accrual exposures - Under the FSAP stress testing approach, non-accrual exposures (e.g., non-performing or S3 exposures) do not earn interest income. - Net income “before stress” can decline even if other parameters are unchanged due to the accumulation of non-accrual exposures in banks’ balance sheets. ### Market risk modelling and implementation - Market risk assessed using a partial revaluation approach against two market stress scenarios (as in the Euro Area FSAP). - Instruments at fair value (FVOCI and FVPL) revalued using bank-specific sensitivities from the ECB’s Short Term Exercise (STE) conducted in the context of SREP. - Sensitivities include: delta (first-order sensitivity), gamma (curvature), and vega (sensitivity to volatility). - Risk factors covered: commodity risk (CM), credit spread risk (CR), equity risk (EQ), interest rate risk (IR), and FX risk. - FX shock corresponded to the first-year FX depreciation in the adverse macro scenarios. - No market shocks applied in the baseline scenario. - Revaluation formula (notation preserved): - ∆VV = Σ_{j ∈ {CM,CR,EQ,IR,FX}} [ (LeDsm_j^{BB} + LeDsm_j^{TB}) ∙ ∆ε_j + 0.5 ∙ Lgamma_j ∙ (∆ε_j)^2 + Vegam_j ∙ ∆σ_j ] - Where j denotes the risk factors, LeDsm_j^{BB}, LeDsm_j^{LB} are the delta sensitivities for the banking book and the trading book respectively, and ∆ε_j and ∆σ_j are the shocks from the market stress scenarios to risk factor j and its volatility respectively. - Conservative floor on market risk losses: - Floor = maximum between: i. the bank’s own fund requirements based on the Fundamental Review of the Trading Book (as reported in the COREP C 91.00); and ii. 8 percent of the bank’s RWAs for market risk. - Caveat: Banking-book deltas in the STE template may include instruments classified at amortized cost (AC) if the bank regularly calculates a fair value, which can overestimate market losses in the banking book. ### Net Fees and Commission Income (NFCI) modelling - NFCI modelled using bank-level panel regressions on the Euro Area FSAP bank sample, using a panel regression with bank-specific fixed effects. - Dependent variable: ratio of NFCI to assets. Explanatory variables include: - first lag of NFCI ratio, real GDP growth, stock market returns, CPI inflation, residential housing price inflation, first-difference of the 1-month EURIBOR rate, first-difference of the yield of 10-year sovereign bonds, EUR/USD FX depreciation, growth in US stock prices. - Up to one annual lag allowed for all regressors. - Estimation features: - Projected at the highest level of aggregation due to breaks in ECB supervisory component time-series. - Model estimated separately for large banks and for other banks. - LASSO methodology applied to select relevant regressors; almost all regressors identified as relevant and used for simulations. - Results displayed for three specifications: (1) all regressors; (2) LASSO-selected regressors; (3) LASSO-selected regressors with Arellano-Bond dynamic panel estimation. - Coefficient interpretation example: - An increase in stock prices by 1 percent will increase the NFCI ratio by approximately 0.2 basis points for large banks. - In the adverse scenarios, stock prices fall by about 50 percent, so the contribution to the change in the NFCI ratio will be of about -10 basis points. - Key explanatory variables: - Real GDP growth: positive coefficient and significant across specifications. - Change in EURIBOR rate: negative coefficient and significant across specifications. - Stock price growth: positive coefficient significant only for large banks. - Projected NFCI outcomes: - Geopolitical scenario: drop in NFCI ratio of about 0.2 percentage points at the trough. - Context: sample average NFCI ratio is about 0.5-0.6 percent, so a drop of 0.2 percentage points represents about a 33-40 percent contraction in NFCI. - Adverse impact milder in the recession scenario mainly due to lower interest rates. ### Other Profit & Loss items and assumptions - Net trading income (NTI): - Baseline: NTI = 5-year average of NTI to total assets. - Adverse scenarios: 20 percent haircut applied to these revenues. - All other P&L items: - Projected by taking a 5-year average relative to total assets. - Mostly non-interest expenses (wages, operating branch costs). - Kept at the same level in stress scenarios as in the baseline. - Tax treatment: - Historical effective-rate approach: each bank’s average effective tax rate = income-tax expense divided by profit before tax (PBT) over the past five profitable fiscal years. - Five-year average rate applied only to periods with positive PBT. - When a bank records a loss (negative PBT) in a year, no tax expense is recognized for that year. - If a loss year is followed by a profit year, current-year PBT is adjusted to offset accumulated loss carry-forward before calculating tax. - Dividends: - Payout rate applied to total comprehensive income (TCI) = 60 percent of TCI. - If the bank incurs losses, dividends set to “0.” - Dividends further adjusted based on the bank’s capital relative to total regulatory capital requirements. - Specifically, when banks’ capital positions breach minimum regulatory requirements (both Pillar I and II), dividends adjusted to “0”. ### Solvency stress test results (aggregate and drivers) - Aggregate outcomes: - Under adverse conditions, aggregate capital ratio remains above minimum capital requirements in both scenarios. - Baseline: banking system remains well capitalized. - Decline in system-level CET1 ratio: - 490 basis points under the recessionary scenario. - 570 basis points under the geopolitical scenario. - Most quantitatively significant drivers of capital ratio decline in adverse scenarios relative to baseline (in decreasing order): i. credit risk for 2.0-2.3 percent of RWAs (reflecting significant vulnerabilities of NFCs); ii. market risk shocks for 1.2-1.3 percent of RWAs (reflecting significant market activities of French banks); iii. NFCI effect for 0.7-1.1 percent of RWAs (reflecting relatively high share of cycle-sensitive fees and commission). - Other aggregate P&L and RWA effects: - NII: decline of 0.3 percent of RWAs in both adverse scenarios relative to baseline. - Asset-side passthrough specifics: - Small pass-through (and slow repricing) to fixed rate housing loans. - Large pass-through (and fast repricing) of NFC loans and consumer credit. - Liability-side passthrough specifics: - Very small pass-through to retail sight deposits. - Large pass-through (assumed equal to one) to wholesale funding. - Additional observed dynamics (qualitative from figures): - Net Fees and Commission Income decline in both adverse scenarios. - Provisions more than double in the second year of the two adverse scenarios compared to the baseline. - Market losses are more pronounced in the recession scenario than in the geopolitical scenario due to the sovereign stress shock. - RWAs increase in the adverse scenarios as a result of credit risk. ### G-SIBs results and buffer implications - G-SIB capital ratio declines: - 520 basis points in the geopolitical scenario. - 370 basis points in the recession scenario. - Drivers for G-SIBs: - Stronger baseline capitalization than aggregate sample. - Lower credit risk contribution to capital ratio declines (of 1.6-1.9 percent of RWAs). - More differentiated impact on NII across scenarios: - Larger decline relative to baseline in the geopolitical scenario. - Very moderate increase in the recession scenario relative to baseline. - This reflects higher share of wholesale funding with high pass-through in G-SIB funding relative to other banks. - Buffer breaches: - No banks breach their minimum requirements. - Several banks dip into additional buffers (CCoB and buffers for systemic institutions) in adverse scenarios. - Four banks – among which are 3 G-SIBs – accounting for a large share of banking assets breach the buffer requirements. - Aggregate gap to the requirement for those four banks remains small: - 1.2 percent of risk weighted assets on average in the geopolitical scenario. - 0.8 percent of risk weighted assets on average in the recession scenario. - Note clarifying minimum requirement definition: - Minimum requirement defined as the minimum CET1 ratio plus P2R. - Hurdle rate with buffers includes CET1 + P2R + capital conservation buffer + systemic buffers (G-SIBs and O-SIIs). - Sectoral risk buffer on exposures to highly indebted corporates and the CCyB are not included. *Source: IMF staff estimates* ### 57. Sensitivity analysis shows that the banking system is resilient to concentration risks ### 57. Sensitivity analysis shows that the banking system is resilient to concentration risks ### Resilience to concentration and sectoral risks - The banking system is resilient to concentration risks because of risk mitigation techniques; large exposures of banks are multiples of capital but become very small after netting out credit risk mitigation measures (CRM). - Sectoral credit risks appear well contained, with heterogeneity across banks. The construction sector and the real estate sector account respectively for 5 percent and 23 percent of total loans to NFCs (source: FINREP). - Even a doubling of NPLs in the two sectors exposed to real estate fluctuations would result in their NPLs reaching only about 4 percent of CET1. ### Large exposures (Table 6) - Exposure value before application of exemptions and CRM / Exposure value after application of exemptions and CRM: - All Banks - Top 10: 10.7 / 0.6 - Top 5: 7.8 / 0.36 - Top 3: 6.3 / 0.23 - G-SIBs - Top 10: 11.5 / 0.62 - Top 5: 8.1 / 0.37 - Top 3: 6.5 / 0.23 - Source: COREP_LE, C 28.00 - Exposures in the non-trading and trading book ### Sectoral credit risk sensitivity (Table 7) - NPL Ratios (Weighted average / Standard deviation): - Construction: 0.7% / 0.4% (baseline), 0.4% / 0.2% (mid), 1.4% / 0.8% (adverse) - Real estate activities: 1.3% / 0.8%, 0.6% / 0.4%, 2.5% / 1.5% - Total: 2.0% / 1.1%, 1.0% / 0.5%, 3.9% / 2.1% - Source: C 01.00, F 06.01 Breakdown of non-trading loans and advances other than held for trading to non-financial corporations by NACE codes and IMF staff calculations ### Sovereign–bank nexus and market shocks - Direct exposures of French banks to the domestic sovereign and to foreign sovereigns remain small as a share of assets, with the exception of LBP. - Satellite models for NII show no statistically significant pass-through of long-term yields to deposit rates, suggesting deposit funding costs were decoupled from sovereign funding costs in the past. - Prior to June 2024, CDS spreads for French banks were only weakly correlated with sovereign CDS spreads; some co-movement has become apparent since the June 2024 elections. - Market shocks to fixed income securities (sovereigns and corporate fixed income bonds) account for a very large share of solvency losses in market shock scenarios; interpreted as an upper bound of solvency impact from sovereign stress through bond markets at about 1 percent of RWAs for the sample of banks. ### Use of solvency stress tests for CCyB calibration (Special Topic) - Counterfactual macroeconomic scenarios, designed between the baseline and the geopolitical adverse scenario (excluding one-off unreversed market risk shocks in scenario 10), are used to illustrate how solvency stress tests could inform CCyB calibration. - Ten incremental scenarios provide a range of severities for macroeconomic downturns to be considered as target objectives for CCyB setting. - Scenario shock increments (first, second, third year) cited: - Incremental shocks to real GDP growth: -0.37, -0.39, -0.09 - Incremental shocks to unemployment rate (ppt): 0.05, 0.26, 0.37 - Constant balance sheet assumption is kept across scenarios to ensure comparability. ### Comparison with past recessions and mapping CCyB to scenario severity - Scenario 2 is more severe after 3 years than the “moderate” 1975 and 1993 recessions in terms of GDP loss. - Precautionary buffers and current CCyB: - Precautionary buffers of approximately 6 percent of RWA in total, combined with the current 1 ppt CCyB, would be enough to maintain bank solvency across a wide range of outcomes. - In the scenario analysis, aggregate capital reaches the aggregate capital requirement plus the current CCyB at 11.11 percent in scenario 10. - If precautionary buffers were lower, the release of the current CCyB might not be sufficient to ensure banks continue to lend under moderate downturn scenarios. - Mapping CCyB levels to scenario outcomes: - The current CCyB level at 1 percent of RWAs can absorb the solvency impact of a moderate macroeconomic shock. - Aggregate capital ratios decline by 1 percentage point (exhausting the current CCyB) in scenario 1, which has real GDP 0.9 percent below baseline after 3 years. - A 2-percentage point CCyB would correspond to the decline in the aggregate capital ratio in scenario 4, which has real GDP 3.4 percent below baseline after 3 years. - Aggregate capital ratio outcomes (illustrative): - Aggregate capital reaches capital requirements plus buffers (including 1 percent CCyB) after 3 years in scenario 10. ### Stress tests of non-financial private sectors (NFCs and households) - Methodology: - Stress tests of non-financial private sectors use the same baseline and adverse scenarios as bank solvency stress tests; methodologies are sector-specific and rely on microeconomic data. - Resulting PDs for NFCs feed into bank solvency stress tests; household model is separate and used for counterfactual policy analysis. ### Publicly listed NFCs: stress-test results - Under solvency stress test scenarios (end-2023 data): - Aggregate debt-at-risk (ICR < 1) increases in the baseline from 6 percent of total debt to 17 percent. - In two adverse scenarios, debt-at-risk increases to about 60 percent of total debt after two years and declines moderately thereafter. - Liquidity needs: - Baseline: firms accounting for up to 65 percent of total outstanding debt would have liquidity needs and increase indebtedness to meet cash outflows. - Adverse scenarios: this share increases to almost 80 percent. ### Sensitivity analysis for non-listed NFCs (ORBIS samples) - Data considerations: - ORBIS data contain unconsolidated liabilities and are less current (latest vintage end-2022). - A sensitivity analysis (rather than full stress test) was carried out for large firms (3,638 entities) and SMEs (7,330 entities), using stressed ICR outputs from the corporate stress test. - Assumptions and shocks: - Assumed lending rates rise in step with the geopolitical stress test scenario: +2.5 percent. - Resulting median increase in interest payments: 20 percent. - Combined with stressed ICRs, average drop in EBIT: 80 percent. - Outcomes: - At end-2022, close to one fourth of unlisted firms already had ICR < 1 (including negative EBIT cases). - Sensitivity results: - Large firms: share with debt at risk (ICR < 1) almost doubles from 25 to 47 percent. - SMEs: share of vulnerable firms rises from 21 to 35 percent. *Source: IMF staff calculations and France FSAP (sections 57–67).* ### 68. The analysis was extended to assess the impact on the sustainability of firms’ debt ### 1fraea2025007 - 68. The analysis was extended to assess the impact on the sustainability of firms’ debt ### Corporate debt sustainability: methodology - Debt-to-EBITDA ratio computed with and without stress using same parameters as in the ICR calculation: - EBITDA constructed as EBIT with depreciation and amortization (D&A) added back to actual and stressed EBIT (no additional stress on D&A). - ICR-style adjustments: EBIT minus 80 percent, interest payments plus 20 percent. - Debt defined as all liabilities with maturity greater than one year plus short-term loans (excluding trade credit). - Assumed higher interest payments on short-term liabilities repricing within one year are fully financed by additional debt, thereby increasing the debt stock in the debt-to-EBITDA formula. ### Corporate debt sustainability: key findings - Close to 30 percent of large firms had a debt-to-EBITDA ratio greater than the critical value of 6 or negative due to operating losses in the baseline. - Under stress the share of large firms with debt-to-EBITDA > 6 or negative increases to 43 percent. - When weighting by diverging debt stocks (debt at risk), the share of debt at risk rises from 44 to 64 percent — an increase of 20 percentage points. - Interpretation: distressed firms tend to be larger within the group of large unlisted firms. - For SMEs: - Baseline share of distressed SMEs is 25 percent (versus 30 percent for large firms). - Increase in distressed SMEs under stress is 13 percentage points. - Debt at risk for SMEs: more than two thirds already experiencing debt sustainability issues in the baseline, and more than 80 percent under stress (noting that the demise of some larger SMEs biases these numbers upward). - Summary metrics: - Actual unsustainable-debt-fundamentals range: 20 to 30 percent of firms (depending on size and metric). - Under assumed shocks this range shifts to 37 to 47 percent. - Considering debt at risk, the share of distressed firms is higher. - Overall: larger unlisted firms appear more vulnerable than SMEs. ### Households’ solvency stress tests: vulnerability assessment - Data source: 2021 Household Finance and Consumption Survey (HFCS). - Cross-country comparisons: - French household debt-to-income and debt-to-assets ratios similar to Spain and Italy, higher than Germany. - Debt-to-income ratios broadly similar across income groups. - Debt-to-financial-asset ratios significantly higher among lower income households. - Loan characteristic patterns: - Share of loans with LTV>90 percent at origination tends to increase with income. - Share of households with DSTI>35 percent is reasonable but on the high side compared to peers. - Interpretation: pockets of vulnerabilities concentrated among lower income indebted households with limited financial asset buffers and high debt; higher income households have sufficient financial assets to mitigate debt service risks. ### Household micro-simulation model: structure and assumptions (Box 6) - Model based on the 2021 HFCS to estimate probabilities of default (PDs) for households holding mortgage debt. - Core logic: - PD of household i at date t equals probability of unemployment times the probability of default conditional on being unemployed. - Default occurs only if unemployed and gross financial assets are depleted (indicator 1{financial assets < 0} used). - For employed households, financial asset dynamics: - E_{i,t} = (1 + r) × E_{i,t−1} + P_she r_in E_{i,t} + Wmg_e_{i,t} − c_ePS_s e rPPS_{i,t} − S U_eP cP PG_L S_{i,t} (equation form as provided) - For unemployed households, financial asset dynamics include unemployment benefits U U_{i,t} (defined via net replacement rates and wages). - Debt service, consumption, and spending are explicitly modeled; spending truncated at 10th percentile (lower bound) and median (upper bound) of survey distribution to correct errors and allow compression. - LGD at default equals max{0, LTV − 1} where LTV computed from remaining debt and property value (property value evolves with scenario real estate price growth). - Replacement rates and unemployment exit probabilities calibrated from OECD and historical periods: GFC for adverse scenarios, pre-pandemic for baseline. - Practical implication: model assumes full recourse mortgages and households prioritize servicing mortgages; default modeled as outcome of unemployment and asset depletion. ### Household stress-test results: default rates and scenarios - Aggregate default rates on mortgages are low in absolute terms and compared to peer countries. - Example comparative note: the highest default rate reached in the recession scenario of 0.7 percent is still below the current default rate of 1.1 percent on mortgages in Spain. - Time-series of simulated France default rates (mortgages) by scenario and year (as presented): - 2024–2027 series shown with values: 0.2%, 0.3%, 0.3%, 0.4%, 0.4%, 0.5%, 0.5%, 0.6%, 0.6%, 0.7%, 0.7% (presented in the text in sequence). - Conclusion: while default rates increase under adverse macroeconomic scenarios, they remain contained/manageable. ### Counterfactual bank solvency stress tests: borrower-based instruments - Counterfactuals assessed using the household micro-macro model: - (i) 90 percent limit on LTV at origination. - (ii) 32 percent limit on DSTI at origination. - Implementation notes: - For DSTI, current DSTI used as proxy for DSTI at origination (survey lacks DSTI at origination). The current DSTI proxy may correspond to a higher origination DSTI since DSTI tends to decline after origination as nominal income rises. - Imposing LTV<90 percent excludes 44 percent of housing loans from the survey population. - Imposing DSTI<32 percent excludes 24 percent of housing loans from the survey population. - Exposures rescaled to maintain a constant total balance sheet of banks. - Conceptual assumption: characteristics of households below/above the limits are representative of the population that would/could be excluded under activated limits. - Key findings: - Impact on PDs: - Both LTV and DSTI limits reduce default risk on housing loans, but the decline is much larger with the DSTI limit than with the LTV limit. - PD decline with LTV limit: ranges from 0.07 percentage points in the baseline to 0.12 percentage points in the last year of the recession scenario. - PD decline with DSTI limit: ranges from 0.22 percentage points to 0.38 percentage points. - Impact on bank capital ratios (CET1): - Borrower-based limits improve bank capital ratios relative to no limits. - Capitalization gains are small under “normal” conditions: 0.1-0.6 percent of RWAs. - Gains increase under adverse macro conditions and peak after 3 years of the recession scenario: - 0.7 percent of RWAs with LTV limits. - 1.6 percent of RWAs with DSTI limits. - Policy implication: at the extensive margin, limits on DSTIs are more effective than limits on LTVs in containing housing loan default risk in France; supports view that ability to service debt is a crucial metric for borrower creditworthiness in France. - Caveat noted: figures likely overstate true capitalization impact because model assumes the same unemployment risk across all households; in practice households with housing loans have lower unemployment risk due to bank screening at origination. ### Liquidity stress tests of banks: approach overview - Structural liquidity analysis considered Basel III LCR and NSFR: evolution, volatility, structure, and currency composition. - LCR assesses ability to withstand short-term outflows relying on liquid assets. - NSFR gauges longer-term structural refinancing and funding risks given amounts of longer-term, illiquid assets. - FSAP additional analyses: - Near-term refinancing needs, availability of collateral, and funding diversification. - LCRs evaluated under shocks more severe than Basel III parameters to test robustness to higher outflow parameters and HQLA valuation shocks. - Cash flow stress tests conducted under scenarios based on supervisory returns of contractual inflows and outflows across maturity buckets. - Scenarios increase severity of run-off rates up to one year and include very short-term runs for sight and on-demand deposits. *Italic: Source — Orbis, 2021 Household Finance and Consumption Survey, EBA Risk DashBoard 2024:Q4, and IMF staff calculations as presented in the provided content.* ### 76. Several stress scenarios are considered to assess French banks’ resilience to liquidity ### Several stress scenarios are considered to assess French banks’ resilience to liquidity risks ### Stress scenarios considered - LCR risk analysis scenario (Table 10) - Sovereign stress causes valuation shocks to HQLA (Level 1A government bonds and Level 2A corporate bonds). - Contagion to the secured lending market causes higher drawdown of committed facilities by financial institutions. - Cash flow stress tests (detailed parameters in Table 11 Panels A, B and C) - Severe recession scenario (scenario 1) - Adverse macroeconomic confidence shock causes valuation loss on CBC, in particular L2 assets. - More limited access to funding markets and draw-down on credit lines by corporates. - Limited outflows for retail deposits; more severe outflows for non-operational corporate deposits. - Banks protect franchise value and there is no inflow from loans to NFCs, retail, and FIs. - Idiosyncratic bank run scenario (scenario 2) - Adverse idiosyncratic confidence shock (solvency concerns, credit losses, or unprofitable business model). - Credit rating downgrades and reputational risk; severe deposit outflows from retail customers, FIs and corporate non-operational deposits. - Some drawdown on committed credit facilities; bank attempts to protect franchise and roll-over loans to NFCs, retail customers and FIs. - Valuation effects on CBC HQLA would be limited. - Sovereign stress scenario (scenario 3) - System-wide shock with higher haircuts significantly affecting HQLA (sharp increase in CBC haircuts for L1 assets). - Some outflows for retail deposits, FI deposits and corporate non-operational deposits. - Note: scenario 3 excludes potential second-round effects (such as recession and potential concerns for individual banks) which would produce a more severe combined scenario. ### B. Structural liquidity risk findings - LCR coverage - All banks in the FSAP sample meet the LCR 100 percent minimum requirement. - Aggregate LCR for the sample: around 150 percent. - Lowest LCR observed: 136 percent. - French banks’ average LCR: above G-SIB peer average but below European peers average of 190 percent. - Outflow structure vulnerability - Main vulnerability: risks related to unsecured non-retail funding, which account for about ¼ of total unweighted outflows. - Currency dynamics and USD funding - Aggregate LCRs in all currencies and in euros have remained stable at around 150 percent every month since 2020. - All USD LCRs are above 100 percent at the end of 2024, but significant monthly volatility observed; aggregate USD LCRs were below 100 percent some months in 2021-2022. - USD outflows account for a significant share of total unweighted outflows with important point-in-time variation. - French banks could rely on the Fed-ECB swap line in event of USD funding stress, but continued close monitoring of USD funding conditions and buffers in USD is warranted. - Net Stable Funding Ratio (NFSR) - All banks meet the minimum NFSR requirement, but ratios remain below the average of European peers and G-SIB peers. - Wholesale funding accounts for 50 percent of unweighted available stable funding with maturity up to 6 months. - In weighted terms: - Retail deposits account for 46.6 percent of available stable funding. - Wholesale funding accounts for 37,6 percent of available stable funding. - Loans to NFC and retail clients account for 53 percent of weighted required stable funding. - USD funding characteristics: - USD funding accounts for 21 percent of total unweighted available stable funding (26 percent for G-SIBs). - Ratio of weighted to unweighted share of USD funding: 176 percent for all banks, 184 percent for G-SIBs. - Other structural characteristics - Funding sources are well diversified on average and asset encumbrance is low. - G-SIBs have upcoming refinancing needs, including near-term, much of which has been pre-financed. ### C. Liquidity stress test results (cash-flow and LCR scenarios) - LCR risk analysis results - Under an LCR stress scenario with higher outflow run-off rates than Basel III and valuation losses on HQLA: - Aggregate LCRs in all currencies: 114 percent. - Aggregate LCRs in euros: 112 percent. - Median LCRs: 107 percent (all currencies) and 101 percent (euros). - 3 banks have LCRs below the 100 percent threshold, among which 2 are G-SIBs. - Valuation shocks to HQLA account for 2 percentage points of the decline in aggregate LCRs; most decline due to increased outflow rates. - Cash flow stress tests (scenario comparisons) - Bank resilience ranking: sovereign stress scenario (most resilient) > recession scenario > idiosyncratic run (most severe). - Survival horizons exceed one month for all banks under cash flow stress scenarios, except for one bank. - Most severe scenario results: - Net liquidity gap after use of CBC reaches 13 percent of banking assets at a 12-month horizon. - Net liquidity gap does not exceed 4.85 percent of banking assets in the first 60 days of outflows. - USD outflows scenarios: - Up to 2 banks experience net outflows that exceed CBC in the first 30 days. - Net liquidity gap reaches a maximum of 1.30 percent of banking sector assets at a one-year horizon. - Reverse cash flow stress tests (30-day horizon) - Procedure: increase severity in convex steps (capped at 20), with parameter caps at 0 percent inflows and 100 percent outflows; most severe terminal severity is double the starting point severity. - Scenario 3 reverse tests: - Cash flows remain resilient at 30 days for most banks until the final 3 iterations. - Scenario 2 reverse tests: - It takes 8 iterations for 4 banks to face funding gaps at a 30-day horizon. - Aggregate cash flows turn negative after 9 iterations. ### D. System-wide liquidity stress tests (special topic) - Two market-impact exercises focusing on the market liquidity channel of investment fund sell-offs: - Sell-off by France-domiciled investment funds (UCITS) and French MMFs (more severe shocks). - Sell-off by Euro Area–domiciled investment funds (includes cross-border funds). - Valuation shock magnitudes mapped to sovereign bond Level 1 category of HQLA: - France-domiciled funds scenario: - 1–5 year France sovereign bonds: price fall of 13 percent over the first two days and 19 percent over a total 2-week period. - Euro Area–wide scenario: - AA sovereign bonds of residual maturity 3–5 years: valuation shocks of 3 percent over the first two days and 6 percent over the total 2 weeks. - Impact on banks’ liquidity - Despite larger valuation shocks in the France-specific exercise, impacts on banks’ liquidity are comparable across exercises. - Euro Area system-wide shock: - Aggregate LCR declines to 125 percent after 2 days and 123 percent after 2 weeks. - France-specific investment funds shock: - Aggregate LCR declines to 122 percent after 2 days and 120 percent after 2 weeks. - In both stress scenarios, 3 banks experience a decline of their LCR after 2 days. - Funding liquidity channel not assessed due to lack of granular bilateral exposure data of investment funds to French banks. ### Interconnectedness analysis — banks - Network construction and coverage - Network of interbank connections constructed using large exposure data for French SIs compiled by the SSM (data as of end-2024). - Information on exposures of SIs to NBFIs was not readily available in the same dataset and was compiled separately by BdF. - Network constructed from COREP large exposure sheets, including claims on non-resident (foreign) banks and dedicated sheets showing the ten largest funding sources of each bank (each > 1 percent of liabilities). - Total network used: - 10 French SIs (including the 4 G-SIBs) and another credit institution (Crédit Logement). - 6 SIs from other EU countries (of which 3 G-SIBs). - 10 G-SIBs outside the EU. *Source: 1fraea2025007 — IMF staff estimates; COREP, Finrep, Fitch, and 2023–2024 pillar III disclosures.* ### 86. The analysis shows that French SIs maintain an extensive interbank network (Figure ### 1fraea2025007 - 86. The analysis shows that French SIs maintain an extensive interbank network (Figure ### Interbank Network Findings - The 4 French G-SIBs account for the bulk of bilateral credit exposures when considering amounts; this is among themselves but also with global and Euro Area SIs. - The G-SIBs are net providers of liquidity cross-border but net borrowers from the other French SIs, some of which themselves obtain funding from global G-SIBs. - Other French SIs are less connected in terms of credit amounts but maintain a high count of individual claims, including with G-SIBs. - Network construction and visualization reference: Figure 35. Network of Bilateral Interbank Exposures (Source: ECB and IMF staff calculations). ### Non-Bank Financial Institutions: Securities Holdings (as of September 2024) - Total held by banks, insurance companies and investment funds: EUR 4.9 trillion in debt securities, listed shares and investment fund shares or units, and securities issued by other private and public institutions (BdF data). BdF database does not include bank deposits. - Insurance companies: - Hold 43 percent of outstanding securities. - Within insurance holdings, securities issued by non-residents combine to close to half of total holdings. - Insurance companies hold about one fourth of their securities in investment funds that in turn invest in bank securities and deposits (not included here). - Investment funds: - Hold about one third of outstanding securities, with about half invested in foreign paper. - Domestic banks: - Account for one fourth of holdings, mostly in foreign instruments and split almost evenly between private and public securities. - Table 8 (Percent of Total Outstanding Securities, September 2024) — holdings read as left-hand sector having exposure to securities issued by institutions in top line: - Insurance Comp.: 0.4, 10.4, 2.2, 3.8, 6.0, 14.5, 5.7 - Investment Funds: 0.2, 5.2, 3.4, 5.3, 0.8, 16.0, 1.8 - Domestic Banks: 0.0, 0.3, 0.7, 1.4, 2.7, 10.3, 9.0 - Source: BdF. Note: All cells add up to 100 percent. The investment fund industry represented includes bond funds, equity funds, hedge funds, mixed funds, real estate funds and other funds as well as money market funds. ### Bank Contagion Analysis — Methodology - Stress test evaluates capacity to react to credit and funding shocks without violating liquidity and solvency requirements. - Liquidity pass condition: bank can replace withdrawn funding by tapping HQLA buffer after net liquidity outflows, or, having exhausted HQLA, use counterbalancing capacity of unencumbered marketable securities subject to assumed fire-sale haircut. Failure occurs when additional buffer exhausted. - Solvency failure: credit losses from simulated defaults of counterparty banks and fire sale losses exhaust voluntary and additional capital buffers. - Hurdle rate: CET1 capital of 4.5 percent of risk-weighted assets plus the Pillar 2 requirement that varies across banks. - Methodology: IMF contagion stress test methodology by Covi et al. (2021), which expands Espinosa-Vega and Solé (2010) to capture all interbank claims and incorporate credit shocks and funding shocks forcing use of liquid assets or fire sales. - Key model parameters and assumptions: - Loss given default (LGD) is bank-specific and in the exercise ranges between 90 and 100 percent (interbank claims weakly collateralized). - Fire sale discount factor (δi) assumed to be 30 percent. - Upper limit to selling illiquid assets (θi) equals each bank’s marketable unencumbered securities (net of those pledged to the ECB). - Bank becomes illiquid if remaining assets insufficient to match liquidity shortage expressed as funding shortfall ρi xij (with ρi set to 1) from withdrawals less liquidity surplus γi (γi equated to HQLA minus net liquidity outflow given LCR exceeding 100 percent). - Contemporaneous default possible when both solvency and liquidity inequalities are jointly satisfied. - Data sources for exposures and bank-specific items: COREP and FINREP supervisory data at the ECB; large exposure sheets for bilateral credit claims and some bilateral funding exposures; LGD per claim calculated as ratio of net exposure (after deducting mitigants) to gross exposure; supervisory sheets provide total assets, risk-weighted assets, CET1 and Pillar 2, unencumbered marketable securities, and LCR components. ### Bank Contagion Analysis — Scenarios and Simulation Design - Sequential default of debtor banks within bilateral interbank network: after each hypothetical default, capital and liquidity impact on each bank computed; banks failing in a round can trigger subsequent rounds until no further failures. - Two scenarios: 1. Scenario 1: bilateral network of resident banks only (all 10 systemic institutions (SIs) supervised by the SMM, combining for close to 90 percent of system assets). Includes 25 bilateral connections. 2. Scenario 2: adds cross-border exposures to non-resident banks (credit and a few funding exposures), restricted to main international G-SIBs and several European O-SIs particularly connected to French SIs. Includes 71 connections (of which, four funding exposures). ### Bank Contagion Analysis — Results - Overall conclusion: credit and funding risks in the French interbank market are low. - Scenario outcomes: - Scenario 1: no bank fails the contagion stress test. - Scenario 2: one smaller French SI fails in the first round due to a relatively large funding exposure assumed withdrawn. - Contagion and Vulnerability Indices (Table 9: sums of individual CIs and VIs): - Contagion Index Overall: Scenario 1 = 7.687; Scenario 2 = 3.579 - o/w for Credit: Scenario 1 = 7.687; Scenario 2 = 3.579 - o/w for Funding: Scenario 1 = 0; Scenario 2 = 0 - Contagion Index GSIBs: Scenario 1 = 7.610; Scenario 2 = 3.166 - o/w for Credit: Scenario 1 = 7.610; Scenario 2 = 3.166 - o/w for Funding: Scenario 1 = 0; Scenario 2 = 0 - Contagion Index SIs: Scenario 1 = 0.077; Scenario 2 = 0.413 - o/w for Credit: Scenario 1 = 0.077; Scenario 2 = 0.413 - o/w for Funding: Scenario 1 = 0; Scenario 2 = 0 - o/w Contagion Index of Failing SI: Scenario 1 = 0; Scenario 2 = 0.701 (o/w for Credit = 0.701; o/w for Funding = 0) - Vulnerability Index French and Int’l Banks: Scenario 1 = n/a; Scenario 2 = 7.208 - o/w for Credit: Scenario 2 = 6.335 - o/w for Funding: Scenario 2 = 0.873 - Vulnerability Index French Banks: Scenario 1 = 8.104; Scenario 2 = 6.399 - o/w for Credit: Scenario 1 = 8.104; Scenario 2 = 5.526 - o/w for Funding: Scenario 1 = 0; Scenario 2 = 0.873 - Vulnerability Index GSIBs: Scenario 1 = 3.158; Scenario 2 = 2.889 - o/w for Credit: Scenario 1 = 3.158; Scenario 2 = 2.889 - o/w for Funding: Scenario 1 = 0; Scenario 2 = 0 - Vulnerability Index SIs: Scenario 1 = 4.946; Scenario 2 = 3.510 - o/w for Credit: Scenario 1 = 4.946; Scenario 2 = 2.637 - o/w for Funding: Scenario 1 = 0; Scenario 2 = 0.873 - o/w Vulnerability Index of Failing SI: Scenario 1 = n/a; Scenario 2 = 1.716 (o/w for Credit = 0.843; o/w for Funding = 0.873) - Interpretation: - Contagion from French G-SIBs through default in credit exposures dominates in both scenarios (index in Scenario 2 lower due to larger capital base when including non-resident institutions). - VIs show French G-SIBs and other SIs are similarly vulnerable to credit shocks in Scenario 1. - Scenario 2 indicates limited increment in vulnerability of non-resident banks due to their large size relative to claims; funding vulnerability triggers failure of one SI that is also somewhat vulnerable to credit shocks. ### Investment Funds’ Liquidity Stress Tests — Introduction and Risks - Over 80 percent of the NAV of investment funds domiciled in France is managed by open-ended collective investment vehicles offering daily redemptions and susceptible to runs under stressed market conditions. - Open-ended bond funds and mixed funds engage in maturity transformation creating maturity mismatches requiring sound liquidity risk management. - When funds hold sufficient liquid assets, redemptions can be met by sales; large unexpected net redemptions can force fire sales of less-liquid assets, incurring significant losses for remaining shareholders and potentially dislocating funding and market liquidity across banks, firms and governments. - Historical redemption episodes cited: - June 2012 (European debt crisis): bond fund redemptions EUR 9.3 billion, equivalent to 4.4 percent of outstanding shares/units at end-May. - March 2020 (COVID-19 pandemic): net withdrawal EUR 10.3 billion, equivalent to 3.4 percent of IF outstanding shares/units at end-February. - Change in valuation during these events: EUR 0.8 billion (June 2012) and EUR 14.8 billion (March 2020); combined flow and valuation effects led to NAV declines of EUR 10.1 billion (June 2012) and EUR 25.1 billion (March 2020). - Bond market liquidity deteriorated sharply during GFC, European debt crisis and COVID-19 pandemic as indicated by spikes in bond market illiquidity measures (Figure 39: French 10-year government security price bid-ask spread in 100 basis points). ### Investment Funds’ Liquidity Stress Tests — Methodology, Data and Coverage - Scope: bond and mixed funds, as well as MMFs; analysis based on end Q3 2024 data. - Lipper data coverage: - 293 bond funds holding EUR 118 billion. - 456 mixed funds holding EUR 88 billion. - 62 MMFs holding EUR 402 billion. - Total in sample: EUR 608 billion. - NAV of these three fund types as at-September 2024: EUR 1070 billion. - Sample coverage relative to industry: MMFs 91 percent, bond funds 37 percent, mixed funds 29 percent. - Fund asset composition: - Bond funds: predominantly fixed income. - Mixed funds: balanced mix of equities and bonds, large share in other investment funds. - Three-quarters of MMF assets in debt securities (CP, CD, MTN, short-maturity bonds), 40 percent of MMF debt securities issued by French banks. - The three fund types together held EUR 13.2 billion in French sovereign securities; total outstanding sovereign securities EUR 2.8 trillion. - Security-level attributes collected from Bloomberg (ISIN-by-ISIN): asset class, issuer type, credit rating, domicile region, and for bond assets detailed instrument characteristics and end Q3-2024 price and yield; time series for average daily trading volume and price volatility. ### Investment Funds’ Liquidity Stress Tests — Scenarios and Assumptions - Initial market shock based on Euro Area FSAP recession scenario: “synchronized global slowdown amplified by sovereign debt distress in the Euro Area, the widening of credit spreads, term premium decompression, and confidence losses softening aggregate demand” (Figure 41). Scenario encompasses potential downgrade of the French sovereign with knock-on effects on banking and corporate sectors. - Redemption shock design: - Standardized uniform withdrawal assumptions of 2 percent and 5 percent of total NAV (observed during historical extreme events). - Sensitivity analysis draws on distribution of outflows observed during COVID-19 pandemic period (Table 10). - Distribution of redemption flows during March 2020 (Number of firms in sample and percent of end-February 2020 NAV): - Sample counts: Bond = 293; Mix = 411 (or 456 elsewhere for total sample — study notes fund classification corrections); MMF = 62. - Redemption flows by percentiles (Bond / Mix / MMF): - 1%: Number funds in sample — 3, 4, 1; Redemption flows: -48.909, -32.336, -42.146 - 5%: 15, 21, 3; -14.979, -10.068, -29.269 - 10%: 29, 41, 6; -9.525, -4.845, -19.372 - 25%: 73, 103, 16; -3.415, -1.924, -7.682 - 50%: 147, 206, 31; -0.547, -0.251, 0.171 - 75%: 220, 308, 47; 0, 0, 4.117 - 90%: 264, 370, 56; 2.618, 1.366, 10.732 - 95%: 278, 390, 59; 6.541, 4.618, 24.630 - 99%: 290, 407, 61; 55.379, 27.330, 57.120 - Sample totals: 2020 = 883 (bond), 2,024 (mix), 174 (MMF); 2024 = 1503, 4305, 120 (contextual counts from sources). - Interest rate and credit risk shocks illustrated in Figure 41 with spread shocks in bps across regions and EUR swap curve shock in bps by maturity buckets (1 month, EUR-1M, EUR-1Y, EUR-10Y). *Source: IMF staff calculations and BdF, AMF, Lipper, Bloomberg data as presented in the source content.* ### 105. Classification of liquid assets. The adequacy of liquidity coverage is measured by the ### 105. Classification of liquid assets. The adequacy of liquidity coverage is measured by the ### Classification of liquid assets - Liquidity characterized by immediacy of trade, market depth, transaction cost, etc. - Banking sector definition of HQLA (Bouveret, 2017) is applied with marginal modifications to account for illiquid assets such as securities issued by EMDE governments. - Table 11: France: Asset Classified as Liquid (In Billions of EUR) - Total 40.1 131.1 66.1 10.9 137.0 58.7 22.2 22.5 77.7 9.6 30.2 - 1/ Including non-rated securities and securities for which rating information was not available. - Source: Lipper and staff calculations. ### Market impact - Redemptions of investment fund shares can force funds to liquidate assets; simultaneous sales of common assets can materially impact prices. - Modeled price impact function depends on: - average daily volume (higher volume → lower negative price impact), - amount of assets being sold (higher amount → greater negative price impact), - volatility of prices (higher volatility → greater negative price impact). - Reference model note: See Bouchaud, 2010. ### Liquidation strategy - Investment funds (IFs) generally follow vertical slicing (pro-rata) to avoid distorting remaining holdings. - Analysis liquidates only the liquid assets (as defined) to identify the point at which more illiquid assets must be sold. - Two liquidation strategies analyzed: - Pro-rata sale of highly liquid assets (sell assets in proportion). - Waterfall sale: pre-determined ranking of liquid assets, with cash followed by MMF shares/units as most liquid. - Sensitivity: analysis includes a cash hoarding scenario (waterfall without using cash), reflecting behavior seen during COVID-19. ### Output of the analysis — Assessing adequacy of liquidity and quantifying shortfalls - Sample: 811 funds. - Under uniform redemption shocks similar to GFC, European debt crisis, and COVID-19: - 19 bond funds and 1 mixed fund (2.5 percent of total number of funds) face liquidity shortfalls; aggregate shortfall EUR 88 million (0.01 percent of total NAV). - At a uniform 5 percent redemption shock: 31 bond funds and 1 mixed fund (3.9 percent of total number of funds) face a liquidity shortfall of EUR 421 million (0.07 percent of total NAV). - Proportionately scaled to actual total number of funds and NAV as at end-2024: - Equivalent shortfall at 2 percent redemption shock: EUR 240 million. - Equivalent shortfall at 5 percent redemption shock: EUR 1.1 billion. ### Output of the analysis — Sensitivity using COVID-19 experience - Monte Carlo simulation: 1,000 random draws using distribution of outflows experienced during COVID-19; scenarios where 1, 5, 10, 25, and 50 percent of funds face COVID-19 level outflows. - Example: if 5 percent of bond funds face redemption shock of - 14.979 percent of NAV (COVID-19 experience): - One-third of draws show aggregate liquidity shortfall between 8 and 12 percent of NAV. - For MMFs: - 90 percent of funds imply small aggregate liquidity shortfalls of about 3 percent of NAV. - Tail risk: if largest funds face a 29 percent redemption shock (experienced by 5 percent of funds in March 2020), aggregate liquidity shortfall could reach 33 percent of NAV. - Conservatism notes: - Results based on liquid asset definition in Table 11, where single A- rated banks are not treated as liquid assets. - Data issues, including non-availability of ratings, may bias results toward conservatism. ### Valuation impact (decomposition and magnitudes) - Total change in NAV decomposed as: - ∆Valuation_ti = ∆ due to initial impact_ti + ∆ due to market impact_ti - ∆NAV_ti = ∆ due to valuation_ti + ∆ due to liquidation (net redemption flows)_ti - Initial market price impact: - Function of modified duration and cash flow weights (longer cash flows → higher sensitivity). - Based on sample holdings: - Bond funds: decline in asset prices of about 2.5 percent under the interest rate shock, and 3.2 percent under the credit shock. - Mixed funds: decline by 0.8 - 1 percent under the two shock scenarios. - MMFs: face 0.2-0.3 percent in price declines. - Price impact from asset sales: - Function of quantity sold, average daily traded volume, and volatility. - Short-maturity securities treated as liquid despite limited trading. - Empirical evidence (Figure 38 and Table 12): valuation effects for MMFs have been minimal even during extreme stress episodes. - Market impact in analysis likely underestimates potential price impact during stress because market liquidity may deteriorate rapidly. ### Historical extreme events (Table 12 highlights) - Table 12 reports monthly changes in NAV and components (Flows, Valuation change, Quantity change) for Bond Fund, Mixed Fund, Investment Fund, and MMF covering 2007-2024, rescaled to end-2024 NAV. - Selected entries (NAV and NAV change examples, In Billions of EUR): - Bond Fund: 2020-03 NAV 273.6; NAV change -25.1; Flow -10.3; Valuation -14.8. - Investment Fund: 2020-03 NAV 1,166.9; NAV change -113.1; Flow -10.3; Valuation -102.8. - Money Market Fund: 2020-03 NAV 301.8; NAV change -52.7; Flow -52.4; Valuation -0.3. - Source: BdF and IMF staff estimates. ### Change in NAV under stress and liquidation strategies - Adverse stress test resembles March 2020 COVID-19 changes in NAV. - Under pro-rata liquidation: - Change in NAV under adverse redemption scenario explained approximately equally by valuation change (interest rate or credit shock) and flows (liquidation); market impact smaller. - For MMFs: valuation impact minimal; most change explained by asset liquidation and some cash liquidation. - Under waterfall liquidation: - MMFs use greater amount of cash to meet redemptions compared with pro-rata. ### Assessing securities liquidated and possible spillovers - Pro-rata strategy, adverse redemption shock (5% of assets sold): - EUR 61 billion in liquid assets and EUR 8 billion in cash disposed to pay investors. - MMFs could sell French bank and corporate debt securities as well as non-French debt securities. - Bond funds would mainly liquidate non-French debt securities. - Mixed funds could sell French IF and MMF shares/units, ETF and equities, as well as non-French securities. - Waterfall strategy, adverse scenario: - Funds would use EUR 40 billion in cash and sell EUR 29 billion in liquid securities; French and non-French government securities liquidated after cash. - Waterfall strategy with cash hoarding (waterfall strategy does not utilize cash): - All funds would sell French government securities more aggressively; MMFs would also sell more French bank, corporate debt, and non-French debt securities; bond funds predominantly sell non-French securities; mixed funds would additionally liquidate MMF and IF shares/units and equity. - Market impact and market capacity: - French government securities average daily trading volume: EUR 15 billion. - Sale of EUR 1-5 billion (redemption shocks of 2-5 percent for bond and mixed funds, and 15 percent of MMF) unlikely to dislocate markets. - Sale of EUR 11-23 billion in French bank securities can put significant price pressure and impact banks’ funding market. - If IFs and MMFs hoard cash by selling assets and not using cash to meet redemptions: - French government securities market could absorb sales albeit with greater price impact. - Banking sector securities are more vulnerable to market dislocation. ### Conclusions and Recommendations - Key conclusions: - French IFs and MMFs have important liquidity buffers to weather plausible redemption shocks. - Sample covered open-ended bond funds, mixed funds, and money market funds offering daily redemptions. - IFs have sufficient liquidity to meet plausible redemption shocks, supported by regulatory framework and a deep liquid securities market. - Tail risk remains large if a large fund faces large redemptions. - Majority of investment funds have liquidity management tools (LMTs) in prospectus and can operationalize them to: - ensure fair treatment between redeeming and remaining investors, or - prevent disorderly redemptions that could affect investors’ best interests. - Recommendations: - Establish more regular and periodic data sharing arrangement on funds’ liabilities between AMF and other relevant regulators to better understand and monitor redemption risks. - Conduct a study on behavior of IFs investing in other IF shares/units to assess whether redemptions of such shares/units have amplification effects. - Perform empirical analysis of past stress episodes to examine whether dash-for-cash episodes lead to greater market impact than episodes without dash-for-cash, to explain divergent flow-performance relationships in crisis episodes. - Operational evidence: - Activation of liquidity management tools by funds in 2022 helped mitigate some redemptions (AMF Annual Report, 2024). *Source: IMF staff and sources as cited in the original content.* ### 118. The French sovereign debt securities market forms the bedrock for financial market ### 118. The French sovereign debt securities market forms the bedrock for financial market stability ### Role and functions of the sovereign securities market - Sovereign debt securities: - serve as a benchmark for pricing financial assets, - are eligible as collateral for borrowing and lending activities, - act as safe-haven assets in times of crisis. - These functions underpin: - zero risk weighting in capital adequacy calculations, - classification as high-quality liquid assets (HQLA) for liquidity coverage ratios in the banking sector. - The market facilitates price discovery and the repricing of risk for all financial assets—both domestically and cross-border. - French sovereign securities provide a safe-haven asset role for French institutions and for the Euro area as a whole, supporting broader Euro Area financial stability. - For these functions to hold, the market must be liquid and well-functioning. - Potential emerging concerns: rising sovereign debt vulnerabilities in France, domestic political discourse, and global geopolitical fragmentation. ### How market dysfunction and stress manifest; market participants and roles - Manifestations of dysfunction/stress: - significant deterioration in bond market liquidity, - market failure to adequately price risks. - Key drivers of stress and dysfunction: - supply-demand imbalances, - shifts in investor sentiment, - deleveraging, - reduced intermediary balance sheet capacity, - funding market stress. - Principal market participants and roles: - Government / AFT (issuer): central role in preserving market functioning and preventing dislocation while managing sovereign risk. - Investors (buyers): heterogeneous objectives and price/time preferences. - Intermediaries / dealers: provide liquidity; supported by derivatives and repo markets which underpin funding and position-taking. - Central bank (BdF) and the Eurosystem: contribute to maintaining market functioning and safeguarding financial stability. - Market infrastructure, regulation, derivatives and repo markets facilitate risk mitigation, arbitrage, and efficient allocation of resources. ### Recent developments and fiscal context - General government debt trajectory and fiscal balances: - General government debt rose from 97.6 percent of GDP in 2019 to 114.6 percent of GDP in 2020. - Debt is projected to continue rising to a high of 128 percent of GDP in 2030 (IMF 2025 Article IV). - France has not run a primary surplus since 2007. - Fiscal deficits and interest dynamics: - Primary deficit increase during COVID-19 contributed to sharp debt increase. - Supportive unconventional monetary policy historically kept interest costs low; reversal of that environment has contributed to rising interest cost while growth remains subdued, worsening “r-g” dynamics. - Yields and spreads: - Yields on French medium- to long-term government securities (OATs) remained high and rising despite monetary easing in the Euro Area in 2024. - The spread between the 10-year OAT and Bund widened to historic highs, reaching 85 basis points by early January 2025. - OAT spread vis-à-vis periphery sovereigns tightened at times, with Spain’s yield briefly falling slightly below that of France. - Current French real yields are around 2 percent; during the GFC they reached 4.5 percent. - Term premia have risen but remain far below levels seen during the European debt crisis. - Sovereign credit ratings: - France’s sovereign credit rating was downgraded to AA-/Aa3. - Fitch downgraded France to AA- in April 2023; S&P followed in May 2024; Moody’s downgraded in December 2024. - Fitch noted France’s debt to GDP ratio stands more than twice the AA median of 50.1 percent at end-2023. - Despite the sovereign downgrade, banks were not downgraded by Fitch in April 2023 and by S&P in June 2024; banks were downgraded by Moody’s in December 2024. - Market resilience: - Despite weakening sovereign finances, the OAT market remains resilient and continues to maintain its safe-haven status. - Contributing factors: diversified investor base; stable and predictable primary issuance practices supported by the primary dealers’ (PDs) system; a liquid money market and secondary market; secure infrastructure; AFT policies balancing costs and risks of the sovereign debt portfolio. ### Structure of supply (end-2024 stock and issuance patterns) - Total general government debt outstanding at end-2024: EUR 3.3 trillion. - Central government marketable debt securities: EUR 2.6 trillion. - Remainder: debt of entities such as CADES and other government bodies. - Composition of sovereign debt securities (share of central government marketable debt; In percent, total outstanding EUR 2.6 trillion): - OAT: 81% - OAT€i: 6% - OATi: 2% - Green OAT: 3% - BTF: 8% - Net issuance and fiscal deficits: - Fiscal deficit increased from EUR 58 billion (2.4 percent of GDP) in 2019 to EUR 207 billion (8.9 percent of GDP) in 2020. - In 2024, fiscal deficit was EUR 173 billion (5.9 percent of GDP). - Cumulative net debt issuance totaled EUR 825 billion over 2020-24. - Projected cumulative net debt issuance over 2025-29: EUR 938 billion (source: 2024 Article IV). - Eurosystem holdings and rolling-off impact: - As at December 2024: - Cumulative net purchases under PSPP: EUR 458 billion. - Cumulative net purchases under PEPP: EUR 292 billion. - Combined amount is about 30 percent of French medium- to long-term government securities outstanding. - Weighted average maturity of Eurosystem holdings: 6.3 years. - Simplifying assumption: about EUR 60 billion in securities could be rolled off annually over the next 12 years, representing 20 percent of the 2025 government gross financing needs totaling EUR 300 billion (to be absorbed by private sector or non-Eurosystem official sector). - In 2020, the Eurosystem’s net purchase of French securities amounted to 64 percent of gross securities issuances. - Since 2023, the Eurosystem’s net purchase has turned negative; starting in August 2024, entered a phase of passive redemptions. - In 2023 and 2024, Eurosystem redemption of French government securities accounted for 19 percent and 31 percent of total redemptions, respectively, absorbed by private sector or foreign official sector. - Table 13 (selected figures; In EUR millions, Net purchase as share of Gross issuance/Redemption): - 2020: Gross OAT issuance 289,509; OAT Redemption 171,578; Net purchase under PSPP 63,375; Net purchase as share of Gross issuance 64.3 percent. - 2023: Gross OAT issuance 303,094; OAT Redemption 165,599; Net purchase under PSPP (29,797); Net purchase as share of Gross issuance -10.3 percent; Net purchase as share of Redemption 18.9 percent. - 2024: Gross OAT issuance 339,804; OAT Redemption 168,612; Net purchase under PSPP (46,225); Net purchase as share of Gross issuance -15.7 percent; Net purchase as share of Redemption 31.5 percent. ### Demand: investor base, holdings, and recent flows - Investor base and evolution: - Split between holdings of general government debt between non-residents and residents remains approximately evenly divided since the last FSAP. - Change in investor landscape dominated by Eurosystem purchases; resident private investors remained relatively stable. - Non-resident official, banks, and non-bank investors have increased holdings of French government debt. - Holdings of central government securities (shares and EUR amounts, Banque de France): - Non-residents hold: - about 52 percent of OATs (EUR 1,122 billion), - 32 percent of Euro inflation-linked OATs (EUR 54 billion), - 17 percent of French inflation-linked OATs (EUR 10 billion), - 87 percent of BTFs (EUR 174 billion). - Historically non-resident share reached as high as 60 percent (post-European debt crisis). - Non-resident composition and behavior: - Arslanalp and Tsuda database: official sector investors (foreign central banks and sovereign wealth funds) accounted for 37 percent of non-resident holdings. - Official sector investors likely motivated by liquidity and are relatively stable holders. - Non-resident private sector investors are the most price sensitive and have stepped in when prices became attractive. - ECB analysis: Eurosystem holdings increasingly being replaced by hedge funds that contribute significantly to absorbing net supply across the Euro Area; BdF analysis suggests hedge funds prefer ultra-long maturities. - Recent transaction patterns (net flows since 2021): - Q4s tend to see net outflows, but there have been sustained net inflows since 2021. - Euro Area and other non-resident investors ex-Japan and the US have sustained inflows. - Euro Area investors accumulated EUR 183 billion in French government securities (source: ECB) during the period referenced. - Japanese private sector investors: net sellers; sold EUR 46 billion cumulatively between 2021-24 (source: BoJ), with selling accelerating in the second half of 2024. - US investors: net sellers; cumulative liquidation EUR 28 billion (source: US Treasury). - Net sales by Japanese and US investors have been more than offset by net purchases by Euro Area and other non-resident investors. ### Secondary market liquidity, interactions, and mitigating factors - Liquidity and pricing risks arise from interactions of supply, demand, and intermediaries, with potential vulnerabilities in secondary market liquidity. - Mitigating factors supporting market functioning and containing risks: - Diversified investor base with different time preferences and price elasticity ensures demand across maturities and price points. - Stable and predictable primary issuance practices supported by the primary dealers’ system. - Liquid money market and secondary market; secure market infrastructure. - AFT policies aimed at ensuring smooth market functioning while balancing costs and risks of the sovereign debt portfolio. - Role of BdF and the Eurosystem in maintaining market functioning and financial stability. - Derivatives and repo markets that support funding and position-taking, enabling dealers to provide liquidity. *Source: IMF staff report text (France FSAP chapter excerpt).* ### 134. Excluding central bank and general government holdings, French and Euro Area ### Excluding central bank and general government holdings, French and Euro Area ### Holdings and investor concentration - French and Euro Area private sector investors held EUR 1.3 trillion in French government securities (Figure 50). - As at Q4 2024, French insurance companies and pension funds together held EUR 360 billion. - Their holdings declined from EUR 400 billion in Q1 2021, but this trend modestly reversed starting Q1 2024. - French investment funds including MMFs held EUR 129 billion in government securities, just 6 percent of total NAV. - Euro Area investors’ holdings of French government securities totalled EUR 0.7 Trillion (Holdings by Euro Area Investors, Total Holdings at end-2024: EUR 0.7 Trillion). - ECB study note: insurance companies, investment funds, and pension funds tend to increase purchases when yield rises; absorption capacity tends to decrease in times of elevated financial market uncertainty. - Large holdings by other Euro Area investors indicate a safe-asset role for French government securities beyond domestic investors. ### Domestic banks and sovereign exposures - French commercial banks held EUR 463 billion in general government debt as at Q2 2024: - EUR 251 billion in securities (mostly issued by the central government). - EUR 212 billion in loans (to local governments). - These holdings accounted for less than 5 percent of total banking sector assets (IMF). - ECB suggests Euro Area bank holdings of sovereign bonds relative to total Tier 1 capital is at a 10-year low. - Banks predominantly hold government securities in held-to-maturity portfolios (EBA). ### Supply and auction mechanics - The annual financing plan is published with the budget toward the end of the previous calendar year and presents total gross amount planned with breakdown of OATs and BTFs and net cash balance. - Auction calendar and dates are fixed and predictable: - OATs auctions on the first and third Thursdays every month. - BTFs auctions every week. - Non-competitive bids are accepted one day after the BTF and OAT auctions. - Pre-auction announcements and significant post-auction transparency are practiced. - Some off-the-run securities are offered on tap to minimize “special” status in repo markets. - The increasing size of securities offered in each auction has not been accompanied by an increase in issuance frequency. ### Primary market demand and syndication - Auctions faced excess demand; the auction bid-to-cover ratio centered around three times in 2024 (Figure 52). - The 2024 bid-to-cover ratio is above the 2019-2024 average and European average that have fluctuated between 2-2.5 (source: AFME). - Excess demand reduces the risk that Primary Dealers cannot offload securities purchased at auction and encourages aggressive bidding. - Syndications complement auctions for new products, new benchmarks, green bonds, and ultra long bonds. - Proportion of issuance through syndication has been increasing, but France's use of syndication remains in the lower range among Euro Area sovereign issuers (AFME). - Syndications have generally faced excess demand and been successful. ### Primary Dealers (PDs), capacity, and incentives - The PD system is governed by a three-year agreement defining privileges and obligations (direct access to AFT, access to syndication, tap issue, prestige; obligations to participate competitively in auctions, provide firm quotes in secondary market, and market government securities). - Number of PDs in France: 15 (constant over the past 5 years), despite significant increase in government securities. - Regulatory changes (supplementary leverage ratio) have reduced PDs’ capacity to absorb risk; the effect was muted when Eurosystem and foreign official sector absorbed issuance. - Concerns that, with increasing free float and competing issuers, PD balance sheet constraints could become more acute. - PDs’ businesses have become more order driven rather than position taking, evidenced by decline in positions in trading books (Table 14). - ECB survey (early 2024): dealer intermediation capacity in European government bond and repo markets remained strong; most dealers reported increased capacity; leverage ratio remained main constraint for some; profitability drives balance sheet allocation. - Incentives offered to PDs include syndication, non-competitive bids, and ability to tap securities on demand; prestige alone may not be sufficient. ### Secondary market liquidity: size, measures, and mixed signals - Individual French bonds outstanding are among the largest in Europe; particularly around the 5-year segment, OATs are the largest individual securities outstanding with 16 bonds exceeding EUR 50 billion in size. - Weighted average amount outstanding for French OATs: EUR 204 billion, compared with EUR 169 billion (Spain), EUR 148 billion (Germany), and EUR 129 billion (UK). - France introduces 4 or 5 new conventional bonds as new benchmark bonds every year (2 or 3 medium- and 2 long-term bonds). - Liquidity measures show mixed signals: - French 5-year government bond bid-ask spread has been one of the tightest in Europe after peaks in March 2020 and January 2023, and increases in mid-June and Decembre 2024 during the political crisis. - Dispersion of off-the-run securities (deviations from fitted curve) suggests deterioration in off-the-run liquidity. - Average daily trading volume of French government bonds appears comparable to Germany and Spain, and half of the UK’s (data not strictly comparable). - ESMA notes market liquidity tends to be concentrated on bonds with original maturity less than 12 years for Euro Area sovereigns. ### Drivers of liquidity stress and contagion risks - Off-the-run securities liquidity suffers when coupon rates differ significantly from current coupon rates; rising market interest rates widen differences and make off-the-runs more illiquid. - Widening French spreads against the German Bund has increased price dispersion between off-the-run and benchmark securities, whereas Spanish and Italian spreads tightened, converging on-run and off-run yields. - Eurosystem holdings of OATs and reduced free float created collateral scarcity, contributing to liquidity squeezes. - Larger holdings by buy-and-hold investors (insurance companies, sovereign wealth funds) may explain lower turnover ratio relative to other Euro Area sovereigns (AFME). - Market liquidity and funding liquidity can reinforce each other, potentially creating liquidity spirals (ECB). - The Euro Area and French repo market remains largely overnight with limited term activity (ESMA), heightening sensitivity to calendar effects and regulatory reporting cycles. - Hedge funds (many offshore) have become more active in French government bond repo and futures markets, relying on leveraged strategies that depend on stable repo funding and low collateral volatility; these could become sources of instability if volatility rises or funding tightens. - Contagion from other markets (US Treasuries, JGBs) could be a vulnerability. ### Risks from sovereign rating changes and clearing - A downgrade of French government bonds could raise financial stability concerns due to increased volatility and higher margin requirements. - Potential changes in collateral treatment by central counterparty clearing houses (where 90 percent of repo transactions take place, ESMA) can cause market dislocation. - Repo transactions through LCH Clearnet have increased multiple fold in 2021 and continue to grow (Figure 55). - In case of sovereign downgrades, clearing houses may raise margin demands, reassess eligibility of French bonds, or increase haircuts, straining market liquidity and affecting primary market funding. ### Market and official backstops - AFT has a standing repo facility with PDs to provide security in short supply or enable short-term positions to manage inventory; AFT has offered securities on tap to minimize specialness. - BdF’s securities lending facilities avail securities under the PPP and PEPP as a lender of last resort. - Studies show these facilities have helped alleviate scarcity in the repo market and enhanced cash market liquidity. - Eurosystem tools to safeguard market functioning include Outright Monetary Transactions (OMT) and the Transmission Protection Instrument (TPI). - ECB retains flexibility to restart reinvestments under the PSPP or PEPP if needed. *Source: IMF staff compilation from Banque de France, ECB, EBA, AFT, AFME, ESMA, and IMF data as presented in the supplied chapter content.* ### 149. Notwithstanding the rising deficit and debt, and large benchmark bond size, gross ### 149. Notwithstanding the rising deficit and debt, and large benchmark bond size, gross ### Debt management, issuance, and buybacks - France’s regular issuance across the curve, especially in medium- and long-term segments, helps to maintain a long average maturity, currently over 9 years (8.5 years including BTFs). - Largest single gross issuance to date: EUR 42 billion (March 2024). - Larger benchmarks (after re-openings) of EUR 60 billion can present significant single-day refinancing risk. - Buybacks are allowed for securities up to two years ahead. - In 2024, EUR 50 billion in securities maturing between 2025-27 were bought back. - Effect on original maturities: - Original maturities: EUR 168 billion (2024) and EUR 198 billion (2025). - After buybacks: EUR 155 billion (2024) and EUR 135 billion (2025). - For 2025, EUR 48 billion will need to be bought back to bring 2026 maturities to 2025 levels. - A further similar amount (EUR 48 billion implied) is needed to bring 2027 maturities to 2025 level. - Buyback objectives: - Smooth near-term (single-day and annual) redemptions. - Improve liquidity in the secondary market by taking out illiquid off-the-run short-residual maturity securities. ### Maturity profile, refinancing risk, and interest-rate risk - Average time to maturity at 9 years contains refinancing risk for the government. - Effective interest rate of the government debt portfolio at end-2024: 1.5 percent. - Marginal funding cost in 2024: 3 percent. - Gross financing needs are over 10 percent of GDP. - Rising term premia and increasing global bond supply may require adjustment of the issuance strategy toward shorter- and medium-maturities and away from the very long end. - Note: average maturity of social security debt is significantly shorter, with higher refinancing risk; AFT extended financing through an emergency liquidity facility to address this. ### Market valuation impact on investors - Since the last FSAP, the yield curve has shifted up 200-300 basis points, implying significant valuation loss on a 9-year average maturity portfolio. - From end-December 2024 price level: - A 1 percent increase in interest rate will incur another valuation loss of EUR 12.5 billion. - A 2 percent increase will imply EUR 23.4 billion in valuation loss. - Most investors (including banks and insurers) typically hold government bonds to maturity or on an amortized cost basis, which limits transmission of valuation losses to broader financial stability concerns. ### Conclusions and market resilience - Sovereign securities are a bedrock for financial market stability: benchmarks for pricing, eligible collateral, and safe-haven assets. - Robustness supported by: - A diversified investor base. - A well-oiled network of intermediaries and distribution mechanism. - Prudent sovereign risk management, including long average maturities and active buyback operations. - Auctions and syndications remain competitive with persistently strong bid-to-cover ratios despite repricing of risk and rating downgrades. - Some measures of secondary market liquidity suggest deterioration, but benchmark bonds remain among the most liquid in the Euro Area. - Eurosystem backstop as lender of last resort and readiness to intervene in severe dislocation reinforce investor confidence. ### Emerging risks and policy recommendations - Monitor potential constraints on primary dealers’ (PDs’) balance sheet capacity to absorb growing auctions; if binding: - Recalibrate mix between auctions and syndications or expand the pool of PDs. - Short-term funding market freeze-ups for PDs could exacerbate secondary-market liquidity drying up. - Policy tools to support secondary market liquidity may become more critical: - Repo facility at the AFT. - Reopening and tap issuances of off-the-run securities. - Care needed to avoid distorting private transactions and removing arbitrage opportunities. - Price-sensitive investors can both support markets and be first to withdraw on negative shocks. - Investor diversification and progress on ongoing pension reform could expand and stabilize investor base. - Debt management alone cannot safeguard safe-asset status without market confidence in a sustainable long-term fiscal strategy. - While ECB liquidity backstop may be called to stabilize markets, market discipline should be allowed to operate to reprice risk and preserve market integrity. *Source: AFT and staff calculations.* ### Box 7. Euro Area: Structural Model for Repricing of Net Interest Income (concluded) ### Box 7. Euro Area: Structural Model for Repricing of Net Interest Income (concluded) ### Structural model definitions and recursion - Exposure-weighted average for bucket [k−1,k]: - Equation (3): 푃푃푡[푘−1,푘] = 휌휌푡 푟푟푡−1[푘,푘+1] + (1−휌휌푡) 푟푟푡 푛푛푛푛 - Where 휌휌푡 = 퐸퐸푡−1[푘,푘+1] / 퐸퐸푡[푘−1,푘] - Initial condition requirement: - Initial interest rate in all buckets 푟푟0[푘−1,푘] is set equal to the average interest income rate of the portfolio at T0, denoted 퐼퐼퐼퐼푇0. ### Step 3 — Interest income calculation - Case without NPEs: - Equation (4): 퐼퐼퐼퐼푡 = ∑_{푘=0}^{3} 푟푟푡−1[푘,푘+1] 퐸퐸0[푘,푘+1] + (1−휔휔) (푟푟푠푠 푃푃푃푃 − 푟푟푠푠−1[0,1]) 퐸퐸0[0,1] - 휔휔 = 푉푉푅푅 푎푎 퐸퐸푉푉푥푥푠푠 푡푡푅푅 / 365 - Interpretation: - First term: base rate determined in year (푠−1) and unaffected by year 푠 interest shock. - Second term: effect of year 푠 interest rate shock on interest-sensitive assets (those repricing during year-푠). These assets earn old rate 푟푟푡−1[0,1] for fraction 휔휔 of the year and change by (푟푟푡 푛푛푛푛 − 푟푟푡−1[0,1]) for fraction (1−휔휔). - Exposures indexed with T0 due to static balance sheet assumption. - Incorporating NPLs / NPEs: - Simplifying assumption: NPE ratio is the same across buckets. - Equation (5): 퐼퐼퐼퐼̂푡 = (1 − 푁푁푃푃퐸퐸푟푟̄푡) ∙ 퐼퐼퐼퐼푡 - Interest income from equation (4) is multiplied by the exposure-weighted average share of performing exposures. ### Rewriting in terms of interest rate “deltas” - Define interest rate delta 푐푐푟푟 relative to 퐼퐼퐼퐼푇0 (e.g., 푐푐푟푟푡 푛푛푛푛 = 푟푟푡 푛푛푛푛 − 퐼퐼퐼퐼푇0). - Constant-rate hypothetical (no interest shocks): 퐼퐼퐼퐼̂푡 = 퐼퐼퐼퐼푇0 ∙ 퐸퐸0. - Difference form (4PP): - 퐼퐼퐼퐼푡 − 퐼퐼퐼퐼̂푡 = ∑_{푘=0}^{3} 푐푐푟푟푡−1[푘,푘+1] 퐸퐸0[푘,푘+1] + (1−휔휔) (푐푐푟푟푠 푃푃푃푃 − 푐푐푟푟푠−1[0,1]) 퐸퐸0[0,1] ### Empirical NFCI panel regressions (Table 16) — selected coefficients and significance - Dependent variable: NFCIR - Key coefficients (All regressors / Lasso-selected regressors / Arellano-Bond estimator) for Large banks and Small/Medium-sized banks: - NFCIR (t-1): 0.707*** / 0.707*** / 0.623*** (Large banks); 0.717*** / 0.714*** / 0.463*** (Small/Medium-sized banks) - RGDP growth: 0.00891*** / 0.00862*** / 0.00880*** (Large); 0.00700*** / 0.00612** / 0.00584*** (Small/Medium) - RGDP growth (t-1): 0.00186 / 0.00174 / 0.00277* (Large); 0.000274 / 0.000286 / 0.00188 (Small/Medium) - Stocks growth: 0.00189** / 0.00190** / 0.00222*** (Large); 0.000349 / 0.000204 / 0.000523 (Small/Medium) - D.EURIBOR: -0.0270*** / -0.0270*** / -0.0295*** (Large); -0.0162** / -0.0177*** / -0.0184** (Small/Medium) - CPI inflation (t-1): 0.00739* / 0.00764* / 0.00501 (Large); 0.00626* / 0.00652** / 0.00577** (Small/Medium) - HPI inflation (t-1): 0.00237 / 0.00184* / 0.00225** (Large); 0.00287*** / 0.00170 / 0.00217 (Small/Medium) - Sample sizes and statistics: - r2: 0.609 / 0.609 / 0.608 (Large); 0.605 (Small/Medium) - N: 595 / 595 / 568 (Large); 942 / 942 / 867 (Small/Medium) - Note: "* p<0.10 ** p<0.05 *** p<0.01". All variables are expressed in percentages. ### LCR parameters and scenario calibration (Table 17 — HQLA and Panel A highlights) - Panel A. HQLA (COREP 72.00) — selected asset haircuts / recognition: - Coins and banknotes: 100.00% / 100.00% - Withdrawable central bank reserves: 100.00% / 100.00% - Central government assets: 100.00% / 95.00% - Recognisable domestic and foreign currency central government and central bank assets: 100.00% / 95.00% - High quality covered bonds (CQS2): 85.00% / 85.00% - Corporate debt securities (CQS1): 85.00% / 75.00% - Qualifying CIU shares/units: underlying Level 2A: 80.00% / 75.00% - Extremely high quality covered bonds: 93.00% / 93.00% - Qualifying CIU shares/units: underlying extremely high quality covered bonds: 88.00% / 88.00% - Asset-backed securities (residential, CQS1): 75.00% / 75.00% ### Outflow and inflow parameters — Scenario frameworks (Tables 17 Panel B; Tables 18a–18c Scenarios 1–3) - Panel B. Outflow parameters (COREP 73.00) — selected entries (Basel III / Stress scenario): - ID 0040: 0.00% / 0.00% - ID 0060 (item 7): 10% / 10.00% - ID 0070 (item 8): 15% / 15.00% - ID 0080 (item 9): 5.00% / 5.00% - ID 0090 (item 10): 3.00% / 5.00% - ID 0110 (item 12): 10.00% / 10.00% - ID 0150 (item 16): 25.00% / 25.00% - ID 0180 (item 19): 100.00% / 100.00% - Several items set at 100.00% across both columns (multiple IDs). - ID 0450 (item 48): 50.00% / 50.00% - ID 0480 (item 51): 5.00% / 5.00% - ID 0490 (item 52): 10.00% / 10.00% - ID 0510 (item 54): 5.00% / 10.00% (for funding promotional loans of retail) - ID 0520 (item 55): 10.00% / 15.00% (for funding promotional loans of non-other) - Several memorandum and contingency items with 0.00% or 5.00% entries across IDs 0940–1100 and beyond. - Scenario 1 (Table 18a) — selected outflow/inflow entries (COREP 66.01): - Liabilities resulting from securities issued (unsecured bonds / regulated covered bonds / securitisations / other): 50% / 50% ... 50% ... 50% - Level 1 central bank assets and level 1 categories: 100% / 100% ... 100% ... 100% - Liabilities from deposits received — selected stable shares: - Stable retail deposits: 19% / 0.16149% / 19% - Other retail deposits: 44% / 0.44087% / 44% - Operational deposits: 15% / 0.51446% / 15% - Non-operational deposits from credit institutions: 50% / 0.524% / 50% - Non-operational deposits from central banks: 25% / 0.218% / 25% - Inflows (Panel B highlights): - Monies due from loans and advances to retail customers: 2% / 0.0154% / 2% - Non-financial corporates: 2% / 0.0154% / 2% - Credit institutions: 100% / 100% / 100% - Other counterparties: 3% / 0.0154% / 3% - Haircuts to counterbalancing capacity (Panel C highlights): - Level 1 central bank: 100% / 100% / 100% - Level 1 (CQS 1): 98% / 98% / 98% - Level 1 (CQS2, CQS3): 95% / 95% / 95% - Level 2A tradable assets (corporate bonds, covered bonds, public sector): 90% / 90% / 90% - Non tradable assets eligible for central banks: 62% / 62% / 62% - Undrawn committed facilities Level 1 facilities: 25% / 25% / 25% - Committed credit facilities considered Level 2B by the receiver: 40% / 40% ... 40% - Outflows from uncommitted funding facilities: 0% / 0% ... 0% - Outflows due to downgrade triggers: 25% / 25% ... 25% - Scenario 2 (Table 18b) — selected differences vs Scenario 1: - Liabilities resulting from securities issued: 100% / 100% ... 100% ... 100% - Liabilities from deposits received — selected stable shares: - Stable retail deposits: 5% / 0.03885% / 5% - Other retail deposits: 10% / 0.07979% / 10% - Operational deposits: 49% / 0.51475% / 49% - Non-operational deposits from credit institutions: 100% / 100.00000% / 100% - Non-operational deposits from non-financial corporates: 76% / 1.06602% / 76% - Inflows: same treatment for monies due from loans and advances as Scenario 1 (e.g., Retail 2% / 0.0154% / 2%). - Haircuts to counterbalancing capacity: same patterns as Scenario 1 (e.g., Level 1 (CQS 1) 98%). - Scenario 3 (Table 18c) — selected differences vs Scenarios 1–2: - Liabilities resulting from securities issued: 50% / 50% ... 50% ... 50% (similar to Scenario 1) - Liabilities from deposits received — selected stable shares: - Stable retail deposits: 5% / 0.03885% / 5% - Other retail deposits: 10% / 0.07979% / 10% - Operational deposits: 15% / 0.12304% / 15% - Non-operational deposits from credit institutions: 50% / 0.52374% / 50% - Haircuts to counterbalancing capacity (Panel C differences): - Level 1 (CQS 1): 90% / 90% / 90% (Scenario 3 lower than 98% in other scenarios) - Level 1 (CQS2, CQS3): 85% / 85% / 85% - Level 1 (CQS4+): 60% / 60% / 60% - Level 2A tradable assets (corporate bonds / covered bonds / public sector): 80% / 80% / 80% - Level 2B ABS / covered bonds / corporate bonds: 50% / 50% / 50% - Non tradable assets eligible for central banks: 50% / 50% / 50% - Committed credit facilities considered Level 2B by the receiver: 20% / 20% ... 20% - Outflows due to downgrade triggers: 75% / 75% ... 75% ### Stress test matrix reference - Table 19 indicates subsequent content: "A. Banking Sector: Solvency Stress Test — Top-down by IMF" (table header present; details not included in the supplied excerpt). *Source: Box 7. Euro Area: Structural Model for Repricing of Net Interest Income (concluded) — extracted tables and equations as presented.* ### 1. Institutional ### 1. Institutional ### Perimeter and institutions included - Institutions included: 7 SI banks, 4 of which are G-SIBs. - Market share: Around 96 percent of the banking sector assets. - Scope of consolidation: banking activities of the consolidated banking group for banks having their headquarters in France; consolidated group basis (CRD V). Insurance activities are excluded; banking associates are included. ### Data, baseline date, and data sources - Data vintage / baseline date: 2024 Q4 (starting point for PL, balance sheet and capital). - Supervisory data: Bank balance sheet and supervisory statistics (including FINREP and COREP), information on interest rate risk in the banking book (IRRBB), short-term exercise (STE), provided by the ECB. - PDs for non-financial corporates: estimated based on the Corporate Stress Test and complemented for some foreign exposures with Expected Default Frequency sourced from Moody’s. - Household analysis: relies on the 2021 Household Finance and Consumption Survey. - Market and public data: ECB statistical data warehouse (funding and lending rates for new business (front-book) by type of asset and funding portfolios), Capital IQ and Orbis for corporate sector analysis. - Coverage of sovereign and non-sovereign securities exposures: debt securities measured through fair value (FVPL and FVOCI) and amortized cost (AC) account. ### Solvency stress-test methodology and channels of risk propagation - Overall approach: Top-down by IMF using FSAP team satellite models and methodologies; balance-sheet regulatory approach. - IRB exposures: projection of PiT and TTC PDs, PiT and DT LGDs, EAD, and RWA. - SA exposures: projection of new flows of defaulted exposures and RWA based on risk weights for performing and non-performing loans separately. - Provisioning: IFRS9 transition matrix approach for IRB and SA. - Market risk: revaluation of trading assets (FVPL) and FVOCI securities assessed using a modified duration approach or sensitivities to market risk factors (Greeks) with hedging strategy considered; equity and derivative exposures assessed using sensitivities (Greeks). - Interest income and expense: time-to-repricing approach. - Satellite models for macro-financial linkages: models for credit losses, funding costs, lending rates; PD analysis using micro-data at individual household (HFCS) and non-financial corporate (commercial corporate database) within EA; outside EA, EDF or Corporate Stress Test model used as proxies for PDs. - LGD shocks for collateralized exposures linked to real estate price paths using a smoothing factor to account for TTC regulatory approach. - Interest income projection: structural approach applying interest rate shocks on new originations and loans’ repricing ladder to outstanding volumes. - Funding costs projection: portfolio-level using funding structure by product (retail and wholesale deposits, secured and unsecured debt securities, repo, etc.) and maturity bucket (overnight vs. term). ### Tail shocks, scenarios, and horizon - Stress test horizon: 2024 Q2– 2027 Q2 (three years). - Scenarios: - Baseline: drawn from the October 2024 WEO macroeconomic projections. - Adverse scenario 1: A geopolitical scenario (or higher-for-longer) featuring an escalation of geopolitical conflicts. - Adverse scenario 2: A recessionary scenario showing a synchronized global slowdown amplified by sovereign debt distress in EA. - Macroeconomic modeling: the two adverse scenarios rely on GFM, a structural macro-econometric model disaggregated into forty national economies. - Real GDP shock magnitudes: geopolitical scenario entails a shock over two years of 2.4 times the standard deviation of 2-year real GDP growth over 1970-2024; recession scenario entails a shock over two years of 2.7 times the standard deviation of 2-year real GDP growth over 1970-2024. - Market risk shocks: modeled as an add-on materializing at the beginning of the first year of each adverse scenario. - Second-round effects and sensitivity analysis: counterfactual policy analysis of household borrower-based instruments and mortgage default; sovereign spreads shocks incorporated in market risk scenarios; exposures to large counterparties documented; variations to inform calibration of the positive neutral CCyB. ### Risk coverage, behavioral adjustments, and calibration - Risks covered: credit (on loans and debt securities), market (valuation impact of debt instruments through repricing and credit spread risk and P&L impact of net open positions such as foreign exchange risks), and interest rate risk on the banking book (IRRBB). - Behavioral adjustment for balance-sheet growth: quasi-static approach — asset allocation and composition of funding remain the same while balance sheet grows in line with nominal GDP paths of major geographical exposures; floor of zero percent on rate of change of balance sheets to prevent deleveraging (constraint binding in the adverse scenario). - FX shock: revaluation effects on foreign currency loans specified in the stress test scenario. - RWA projection: standardized and IRB portfolios differentiated. Standardized portfolios: RWAs change due to balance sheet growth, new inflows of non-performing loans, exchange rate movements, and conversion of a portion of off-balance sheet items to on-balance sheet items. IRB portfolios: through-the-cycle PDs, downturn LGDs and EAD for each asset class/industry used to project risk weights. - Calibration specifics: - Interest income from nonperforming loan is not accrued. - Dividends: paid out by banks that remain profitable and adequately capitalized throughout the stress; dividend rate equals the average ratio between observed dividends and profits after tax over the last five years. - Tax rate: set at 30 percent in line with 2023 EBA methodology. ### Regulatory and market-based standards and parameters - National regulatory framework: Basel III regulatory minima on CET1 (4.5 percent) and include any requirements due to systemic buffers (SyRB, G-SII buffer, O-SII buffer), with and without capital conservation buffer (CCoB), and Pillar II requirement. - Leverage ratio: assessed during the stress test horizon against the 3 percent Basel III minimum requirement. - Liquidity regulation (LCR-based tests): hurdle rate set at 100 percent at the aggregate currency level (per Basel III and domestic regulation). ### Reporting and outputs for solvency tests - Output presentation includes: - Capital path under various scenarios by groups of banks, categorized by business model. - System-wide capital shortfall. - Number of banks and percentage of banking assets in the system that fall below regulatory minima or breach capital buffers. - Information on impact of different result drivers, including profit components. ### Liquidity test (Banking Sector) - Institutions included: 7 SI banks, of which four are G-SIBs. - Market share: Around 96 percent of the banking sector assets. - Data and horizon: Data vintage: 2024 Q4; supervisory data from ITS files (FINREP, COREP). - Methodologies: - LCR-based tests, using regulatory parameters and more severe scenarios; breakdown by significant currency where available. - Cashflow-based liquidity stress test, breakdown by significant currency where available. - Share of large depositors assessed to describe concentration risks. - Stress-test horizons: 30 days for LCR-based tests, and up to 1 year for cashflow analysis. - Type of analyses: scenario analysis (various stress scenarios), main risks analyzed include market upheaval and tightening of market liquidity conditions (linked to solvency adverse scenario where possible), deposit run-offs, outflows from top funding sources; reverse stress tests. - Behavioral adjustments and buffers: liquidity from the central bank is not considered; capacity of banks to generate liquidity from inflows and from assets under stress (counter-balancing capacity) assessed. - Reporting format for liquidity results: outputs include Average LCR, Net Liquidity Position and survival period, and number of institutions with LCR below regulatory limits. ### Mutual funds sector — liquidity risk (Top-down stress test assumptions) - Institutional perimeter: all open-end debt-oriented schemes. - Supervisory and commercial data: fund-level characteristics and AUM, cash flow data, fund investment portfolio, bond market trading data; other commercial data sources: Bloomberg. Sample period: From December 2017 to September 2024. - Liquidity resilience metric: Redemption Coverage Ratio (RCR) based on value of high-quality liquid assets and calibration of redemption shock. - Redemption shock calibration: both historical simulation and flow-performance approaches. - Historical simulation: instantaneous shocks simulated based on historical net flows under fund homogeneity, fund heterogeneity and fund family assumptions. - Flow-performance approach: exogenous market shocks trigger change of NAV which lead to additional redemption outflows; redemption shock triggered from NAV change via interest rates and credit spreads. - Market impact estimation: based on assumptions on different fund liquidation strategies and segmental-market characteristics; second-round redemption shock triggered if market sale causes significant price impacts leading to asset devaluation of funds. - Scenarios: - Baseline: historical simulation calibrating redemption shock based on time series cash flow data under four years horizon. - Adverse: exogenous market shock triggers asset depreciation through interest rate risk and creates additional redemption shocks. - Fund liquidation responses: prorate approach and waterfall approach. - Market-impact estimation under three market scenarios: peak, normal and low trading activity. - Sensitivity analysis: reverse stress test showing total number of funds failure with levels of redemption shock applied to funds homogeneously. - Risks assessed: interest rate risk, market risk, liquidity risk. - Mutual funds reporting outputs: - Redemption Coverage Ratio and liquidity shortfalls at fund level. - Number of funds that cannot survive the shocks (with the RCR ratio below one and liquidity shortfall larger than zero). - Total value of assets sold under different scenarios. - The percent of price decline under different market conditions and the mitigation effect of central bank lending facilities. *Source: 1fraea2025007 - 1. Institutional* --- _Source: https://www.imf.org/-/media/files/publications/cr/2025/english/1fraea2025007.pdf_