## EXECUTIVE SUMMARY (1luxea2024008)

## Source details

**Canonical URL:** [EXECUTIVE SUMMARY (1luxea2024008)](https://www.imf.org/-/media/files/publications/cr/2024/english/1luxea2024008.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2024/english/1luxea2024008.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2024/english/1luxea2024008.pdf.json)

---

### Overview and context
- Luxembourg financial system: highly interconnected, diverse, complex, historically resilient amid heightened economic, financial, and geopolitical uncertainty.
- Macroeconomic signals:
  - Real GDP: October 2023 WEO expected contraction by ½ a percent in 2023 before rebounding to 1½ percent in 2024.
  - National statistical office (December 2023): forecast a recession by 1 percent in 2023 before a recovery by 2 percent in 2024.
  - Unemployment: 5.6 percent in October 2023.
- Financial-sector scale and structure:
  - Financial sector employment: 55,000 people (about 11 percent of total employment).
  - Bank assets: about 12 times GDP.
  - Insurance sector size: about 3 times GDP.
  - Investment fund sector: over 3,300 directly regulated Undertakings for Collective Investment and 122 money market funds registered as of August 2023.
  - Assets (August 2023): non-MMF funds held over 6 trillion euro; MMFs held 0.5 trillion euro.

### Stress-test framework and scenario
- Integrated stress tests across banking, insurance, and investment fund sectors using a common macrofinancial framework.
- Adverse macrofinancial scenario: severe global recessionary scenario coupled with high inflation and a sharp increase in short-term rates in the first year of a three-year horizon (2024–2026). Scenario incorporates domestic and international real estate sector risks.
- Transmission channels modeled: fund redemption shock → bank deposits and securities prices; funding stress and fire sales → solvency impacts.
- Financial Shock Simulator inputs:
  - 27 risk factors; universe includes sovereign and corporate bonds and equities from France, Germany, Japan, Luxembourg, United States, other Euro Area, and the rest of the world.
  - Forward paths simulated 10,000 times for a 1-year horizon.
- Macro scenario severity:
  - Adverse shock envisaged a 5.9 percent cumulative drop in the level of GDP over 2024-25.
  - Adverse scenario corresponds to - 10 percentage points compared to the baseline level of output at the trough, a 2.3 standard deviation shock from the baseline and a 1.4 standard deviation shock from the mean of the historical distribution.
- Coverage:
  - Banks: 39 banks (19 SIs and 20 LSIs).
  - Investment funds: over 1,000 open-ended funds in sample (1,082 reported later).
  - Insurers: 20 insurers (sample covers 80–90 percent of assets).

### Banking sector — solvency stress-test headline findings
- Starting positions and aggregate resilience:
  - CET1 capital ratio: 21½ percent in 2023Q2 (21.7 percent as of December 2023 in stress test tables).
  - Capital adequacy ratio: 23 percent as of June 2023.
  - Tier 1 capital represents 96 percent of total regulatory capital.
  - Average risk weight: 36 percent (35.6 percent reported for the stress test sample).
- Adverse scenario (three-year horizon, 2024–2026):
  - Aggregate CET1 ratio: decline by about 4 percentage points in 2024 to 17.6 percent before gradual recovery.
  - Drivers of capital depletion: credit loss provisions, market losses, and change in RWAs.
  - Expected credit losses (IFRS-9) over three years:
    - Adverse: €3.8 billion, equivalent to 1.9 percent of total banking system RWAs.
    - Baseline: €1 billion, equivalent to 0.5 percent of total banking system RWAs.
  - Market losses on AFS and HFT portfolios: €2.5 billion in first two years, impacting CAR by 1.2 percentage points.
- Capital shortfalls and recapitalization needs:
  - Four banks (together making up less than 10 percent of banking sector assets) would fall below the CET1 hurdle rate of 8 percent; recapitalization needs amount to 0.6 percent of nominal GDP.
  - Leverage ratio outcomes:
    - Tier 1 to total assets declines from 7.7 to 6.8 percent in 2024 and to 7.4 in 2025.
    - Five banks (together making up less than 20 percent of banking sector assets) fall below the 3 percent leverage ratio in 2024; recapitalization needs equivalent to 0.9 percent of nominal GDP.
    - Two banks would be below both CET1 and leverage thresholds.
- Business model and orientation differences:
  - Private banks: largest capital depletion (lower initial profitability, negative repricing gap).
  - Domestically oriented banks: slightly more impacted than internationally oriented banks in year one.
  - Corporate finance banks: lowest average level of capitalization post-shock.
- Key credit-risk PD dynamics:
  - Mortgage retail PD: from 0.9 percent (June 2023) to peak of 3.5 percent under adverse scenario.
  - Corporate loan portfolio PDs: from 1.9 percent (June 2023) to 5 percent in 2025.
- Concentration risk:
  - Simultaneous default of the net largest exposure of each of the thirty-nine banks: CET1 ratio decline of 4.4 percentage points; undercapitalization of five banks; capital shortfall 1.8 percent of nominal GDP.
  - Simultaneous default of five net largest exposures: fourteen banks undercapitalized; capital shortfall 7.8 percent of nominal GDP.
  - Simultaneous default of ten net largest exposures: twenty banks undercapitalized; capital shortfall 14.4 percent of nominal GDP.

### Banking sector — liquidity and funding stress-test findings
- Liquidity buffers and regulatory metrics:
  - Aggregate LCR: 150 percent (aggregate LCR based on stress test sample: 229.1 percent reported for sample; weighted average LCR baseline: Full sample 229 percent; retail/universal sample 180 percent).
  - NSFR: all banks above 100 percent; weighted average NSFR: 187 percent.
- Cash-flow and LCR scenarios:
  - Four liquidity scenarios: baseline, deposit run, combined, weak parent bank.
  - Cash-flow outflow rates (Cumulative over Six Months; selected):
    - Stable retail deposits: Deposit Run 5% ; Combined 5% ; Weak Bank 10%
    - Other retail deposits: Deposit Run 23% ; Combined 23% ; Weak Bank 30%
    - Operational deposits: Deposit Run 26% ; Combined 26% ; Weak Bank 52%
    - Committed credit facilities: Deposit Run 12% ; Combined 12% ; Weak Bank 15%
    - Liquidity facilities: Deposit Run 60% ; Combined 60% ; Weak Bank 75%
  - Haircuts on counterbalancing capacity (selected):
    - Level 1 (CQS 1): Deposit Run 2% ; Combined and Weak Bank 15%
    - Level 2B ABS, covered bonds: Deposit Run 3% ; Combined and Weak Bank 25-30%
    - Non tradable assets eligible for central banks: Deposit Run 25.5% ; Combined and Weak Bank 90%
- LCR stress-test results:
  - Number of banks below LCR threshold (<100%):
    - Baseline: 0
    - Deposit Run: 5
    - Combined: 6
    - Weak Bank: 3
  - Share of total sample assets below LCR threshold:
    - Baseline: 0
    - Deposit Run: 21.7
    - Combined: 23.8
    - Weak Bank: 17.6
- Cash-flow analysis and funding gaps:
  - Number of banks with net funding gap (end-of-period CBC<0):
    - Baseline: 1
    - Deposit Run: 3
    - Combined: 6
    - Weak Bank/Same-name Funds: 5
  - Share of total sample assets with net funding gap:
    - Baseline: 14.2
    - Deposit Run: 23.8
    - Combined: 19.0
  - Time buckets with largest liquidity gaps: 3–6 month and 6–12 month buckets.
- Liquidity–solvency interaction (third-round effects):
  - Fifteen banks experience funding gaps under combined scenarios; nine banks access central bank facilities and pledge HTM securities (assumed penalty 100 bp over main refinancing rate).
    - Additional funding costs: €1.1 billion.
    - Further CET1 ratio decrease for these nine banks: 1.4 percentage points.
    - Caused undercapitalization of one more bank (leverage ratio) in 2024.
  - Six banks not accessing central bank facilities sell HTM and realize market losses:
    - Estimated losses: €88 million.
    - Further CET1 ratio decrease: 0.8 percentage points for five banks referenced.
  - System-wide impact equivalent to decline of 0.5 percent of aggregate CET1 ratio and 1.3 percentage points of CET1 ratio of banks with funding gaps.
- Currency liquidity mismatches:
  - LCRs in USD and GBP below 100 percent for all business models in sample.
  - 11 banks with GBP liabilities fall short of GBP HQLA in the stress test sample.

### Investment funds — liquidity, redemptions, and spillovers
- Sample and coverage:
  - Stress test sample: 1,082 funds with aggregate total net assets of 2,675.5 billion euros (895 UCITS; 187 AIFs).
  - UCITS account for 74 percent of sample total net assets.
- Initial valuation impacts:
  - Aggregate initial valuation losses under adverse scenario: 13.1 percent of aggregate NAV.
  - By strategy (Adverse scenario, initial average NAV fall):
    - Bond funds: -8.0 percent
    - Equity funds: -17.4 percent
    - Fund of funds: -16.0 percent
    - Hedge funds: -12.1 percent
    - Mixed funds: -12.6 percent
    - Real estate funds: -14.4 percent
- Redemption modeling:
  - Flow-performance model calibrated by strategy; where coefficients not significant, historical redemption scenarios used.
  - Redemption Coverage Ratio (RCR) = HQLA / 30-day redemption shock.
  - HQLA distribution: even 10th percentile of overall sample exhibits roughly 40 percent of net assets in HQLA.
- Liquidation strategies examined:
  - Slicing (pro rata), Waterfall cash-first, Waterfall securities-first; main analytical focus on securities-first waterfall to maximize price-impact second-round effects.
- Asset sales and price-impact findings:
  - In adverse scenario, funds may sell up to 150 billion euros in liquid securities to meet redemptions (approx. 50 billion sovereign bonds, 20 billion corporate bonds, 70 billion equities).
  - Price impacts on securities in adverse scenario: bond price impacts range from 0.2 to 1.2 percent for selected Eurozone countries; US/UK impacts more subdued.
  - Aggregate final NAV decline after second-round effects (Adverse): average -13.8 percent.
- Liquidity outcomes and vulnerabilities:
  - Overall: UCITS and majority of AIFs have sufficient liquidity buffers; small second-round effects.
  - Only four AIFs (loan funds) exhibit RCR < 1 in the adverse scenario after second-round effects; these are mixed/loan-fund strategies with longer redemption frequencies and LMTs.
  - Historical redemption-based vulnerabilities:
    - High-yield bond funds: 44 percent of total NAV of high-yield bond funds would have problems meeting large but plausible historical redemptions.
    - Six real estate AIFs have RCR < 1; these have quarterly or longer redemptions and use LMTs (redemption deferral).
- Reverse stress tests:
  - Even with uniform redemption shocks near 40 percent, over 90 percent of funds maintain RCR > 1.
  - Cumulative liquidity shortfall remains minimal up to 40 percent shocks.

### Money Market Funds (MMFs) — solvency (NAV) stress-test findings
- MMF sample (Q1 2023):
  - CNAV: Number of Funds 10; Total Net Assets 72.0 (billions of euros); Average WAM 7.2 (days).
  - LVNAV: Number of Funds 15; Total Net Assets 220.9; Average WAM 14.7 (days).
  - VNAV: Number of Funds 79; Total Net Assets 136.7; Average WAM 1.7 (days).
- Adverse scenario MMF results:
  - Median fall in NAV for LVNAVs: 14 basis points.
  - Four LVNAVs would exceed 20 basis points fall in shadow NAV under the adverse scenario (two USD-based and two GBP-based).
  - Two small CNAV MMFs experienced NAV declines larger than 20 basis points; these two funds total under 2 billion euros in net assets.
- Reverse stress-test findings:
  - Majority of LVNAVs and CNAVs resilient to cumulative shocks up to approximately 200 basis points; majority by NAV can withstand cumulative increases of 300 bps.
  - Almost all LVNAVs cross the 20-bp threshold for combined shocks of 400 basis points.
- Recommendation: consider market guidance to enhance resilience of outlier LVNAVs.

### Insurance sector — solvency and liquidity findings
- Sample and coverage:
  - 22 insurers: 10 life insurers (~83 percent of sector assets) and 12 non-life insurers and reinsurers (81 percent of gross written premiums).
  - Aggregated participant assets: EUR 239bn; EUR 187bn attributable to life insurers.
  - Pre-stress: all 22 participants have solvency ratios above regulatory threshold 100 percent.
- Valuation and solvency impacts (Adverse scenario):
  - Life insurers:
    - Assets decline by 9.1 percent on average; liabilities decline by 8.8 percent.
    - Excess of assets over liabilities (EoAoL) declines by 28 percent (from EUR 3.43bn to EUR 2.46bn).
    - Median life insurer post-stress assets-to-liabilities ratio: 101.6 percent (down 0.3 percentage points).
    - Eligible own funds decline by 22 percent for sample insurers.
    - Median SCR coverage declines from 150 percent pre-stress to 127 percent post-stress; no life insurer falls below 100 percent.
  - Non-life insurers:
    - Assets decline by 2.8 percent on average; liabilities decline by a smaller amount.
    - Eligible own funds decline by 8 percent.
    - Median SCR ratio after stress: 179 percent (down by 17 percentage points).
- Stage 2 (price impact of investment fund redemptions):
  - Assets of life insurers decline by less than 0.5 percent.
  - Median SCR ratio for life insurers further declines by 1 percentage point to 126 percent.
  - Median SCR for non-life: 175 percent (4 percentage points lower than after first-round).
- Insurance liquidity (surrenders/lapses) analysis:
  - Simulated one-month outflow (peak +50 percent of highest historical monthly lapse for guaranteed business): simulated outflow EUR 2.4bn, ~7.6 percent of statutory reserves in guaranteed business (as of June 2023).
  - Total liquidity sources: EUR 32.5bn (Cash EUR 1.2bn; HQLA after haircuts EUR 9.3bn; Reinsurance recoverables EUR 22.0bn).
  - Aggregated RCR: 13.6; least liquid life insurer RCR: 9.6.
  - Derivatives notional: EUR 3.6bn (market value EUR 88m); margin-call risk limited.
- Sensitivity analyses:
  - Return to ultra-low rates: median life insurer SCR deteriorates by 12 percentage points.
  - Equity price shock (-50 percent): reduces median life insurer SCR by 30 percentage points (mainly via future fee income losses).
- Supervisory recommendations for insurance:
  - Improve supervisory reporting data quality, especially S.06.02 (investments) and S.08.01 (derivative holdings).
  - Implement regular top-down stress tests and sensitivity analysis starting with standard formula users and expand to internal model users.

### OFIs, captive financial institutions, and interconnectedness
- OFI sector large and diverse; S127 subgroup (“captive financial institutions and money lenders”) holds over 90 percent of OFI assets.
- S127 entity counts and concentration:
  - BCL-reporting subsample: ~2,700 entities with assets > EUR 500 million representing 85 percent of population by asset size at end-2018.
  - Extrapolated whole population suggests > 50,000 entities in S127.
  - About 80 percent of their assets held by holding companies or entities set up to facilitate intragroup lending.
- Geographic composition (average 2014-19):
  - By affiliated-group domicile: USA 59%; GBR 11%; EA 9%; Other 21%.
  - Balance-sheet counterparty asset side distribution (%): LUX 27, EA 22, GBR 8, USA 7, Other 37.
  - Liability side distribution (%): LUX 35, EA 21, GBR 10, USA 7, Other 27.
- Cross-border linkages and exposures (2022, selected shares):
  - Investment funds: ~80 percent of assets and 90 percent of liabilities with ROW.
  - OFIs: ~60 percent of assets and 45 percent of liabilities cross-border.
  - MFIs: assets related to ROW approach 90 percent; liabilities > 50 percent with ROW.
- Transparency and data:
  - Complexity and limited transparency of linkages hamper full assessment of contagion channels; more granular entity-level data recommended.

### Key policy and supervisory recommendations (selected)
- Banking sector:
  - Improve top-down supervisory stress test models to incorporate liquidity–solvency interactions.
  - Use cash-flow data for all significant currencies and jurisdictions to refine liquidity monitoring and stress testing.
  - Continue monitoring bank liquidity risk from fund/bank connections, including depositary–fund manager group links and exposures to potentially weak parent banks.
  - Consider strengthening treatment of concentration risk in SREP and Pillar II capital reviews.
- Investment funds and OFIs:
  - Carefully analyze interlinkages of investment funds with the OFI sector (including loans) to better understand vulnerabilities and amplification mechanisms.
  - Further monitor liquidity mismatches in newly emerging AIF strategies (semi-liquid structures, LDI-type strategies, “retailization”).
  - Enhance monitoring of AIFs and their interlinkages to OFIs and other funds; tailor methods to assess AIF liquidity risk.
- Money Market Funds:
  - Consider market guidance to enhance resilience of outlier LVNAVs.
  - Engage with market participants employing riskier strategies to encourage alignment with more conservative LVNAV practices.
- Insurance sector:
  - Improve quality of supervisory reporting data, especially on investments and derivatives.
  - Implement regular top-down stress tests and sensitivity analysis for the insurance sector, starting with standard formula users and expanding to internal model users.
- Timing conventions used in recommendations:
  - “C-Continuous”; “I-Immediate” is within one year; “NT-near-term” is 1–3 years; “MT-medium-term” is 3–5 years.

*Source: EXECUTIVE SUMMARY (1luxea2024008) — Luxembourg FSAP executive summary, as contained in the provided content unit.*

### EXECUTIVE SUMMARY ___________________________________________________________________________ 7

### EXECUTIVE SUMMARY

### Major Themes and Sections
- FINANCIAL SYSTEM STRUCTURE AND INTERCONNECTEDNESS
  - A. Investment Funds
  - B. Banking Sector
  - C. Insurance Sector
  - D. Captive Financial Institutions and Money Lenders
  - E. Interconnectedness
- STRESS TEST STRATEGY
  - A. FSAP Stress Testing and Methodology
  - B. Financial Market Shocks (First Round)
  - C. Macrofinancial Scenario—Baseline and Adverse (First Round)
  - D. Market Contagion and Liquidity-Solvency Interactions Effects (Second and Third Rounds)
- BANK SOLVENCY STRESS TESTS
  - A. Types of Stress Tests
  - B. Banking Sector Solvency Stress Test Results
  - C. Results of Sensitivity Analysis
- BANK LIQUIDITY STRESS TESTS
  - A. Methodology
  - B. Results
- LIQUIDITY STRESS TEST FOR INVESTMENT FUNDS
  - A. Objective and Scope
  - B. Methodology
  - C. Results
- MONEY MARKET FUNDS STRESS TESTS
  - A. Objective and Methodology
  - B. Results
- INSURANCE SOLVENCY STRESS TEST
  - A. Scope and Sample of the Solvency Stress Test
  - B. Scenario for the Solvency Stress Test
  - C. Capital Standard and Modeling Assumptions
  - D. Results of the Solvency Stress Test
  - E. Sensitivity Analyses
- INSURANCE LIQUIDITY ANALYSIS
  - A. Approach and Scope for the Liquidity Risk Analysis
  - B. Results of the Liquidity Risk Analysis

### Figures (selected enumeration)
- Figure 1: Financial Sector Size Compared to the U.S. and the Euro Area
- Figure 2: Investment Funds’ Market Share by Type of Structure and Strategy
- Figure 3: Stock of Investment Funds’ Direct CRE Investments by Country
- Figure 4: Investment Funds—Net Redemptions and Change in Assets
- Figure 5: Banks by Business Model
- Figure 6: Banking Sector Balance Sheet
- Figure 7: Insurance Market Structure
- Figure 8: Captive Financial Institutions and Money Lenders
- Figure 9: Interconnectedness
- Figure 10: Stress Test Strategy
- Figure 11: Financial Shock Calibration
- Figure 12: Scenario Severity from a Historic Perspective
- Figure 13: Macroeconomic Baseline and Adverse Scenarios
- Figure 14: Stacking Order for Overall Capital Requirement/Maximum Distributable Amount
- Figure 15: Distribution of Banks’ Exposures by Asset Class (as a Percent of Total Assets)
- Figure 16: PD Projections in the Baseline and Adverse Macroeconomic Scenarios
- Figure 17: Bank Solvency Stress Test Results
- Figure 18: Bank Solvency Stress Test Results: Breakdown Between Domestically-Oriented and Internationally-Oriented Banks
- Figure 19: Bank Liquidity-Solvency Interactions
- Figure 20: Sensitivity Analyses for Sovereign and Credit Concentration Risks
- Figure 21: Bank NSFR Stress Test Results
- Figure 22: Bank LCR Components and Stress Test Results
- Figure 23: Sensitivity of LCRs to Deposit Run-Off Rates
- Figure 24: Cash-Flow Based Analysis Results
- Figure 25: Investment Funds’ HQLA Distribution by Strategy
- Figure 26: Investment Funds—Asset Exposures
- Figure 27: Net Outflows Under Different Scenarios
- Figure 28: Price Impact of Asset Sales for Selected Asset Groups
- Figure 29: Investment Funds’ Reverse Liquidity Stress Test Results
- Figure 30: MMF Stress Test Results—Boxplots with Fall in NAV by Type of MMF
- Figure 31: MMF Reverse Stress Test Results
- Figure 32: Insurance Solvency Stress Test—Valuation Impact
- Figure 33: Insurance Solvency Stress Test—Solvency Impact
- Figure 34: Insurance Sensitivity Analyses
- Figure 35: Life Insurance Lapse Rates
- Figure 36: Life Insurance Redemption Shock

### Tables (selected enumeration)
- Table 1: 2024 FSAP: Key Recommendations
- Table 2: Banks’ Financial Soundness Indicators (Based on the Stress Test Sample)
- Table 3: Stress Test Average Hurdle Ratios vs. Current Ratios
- Table 4: Comparison Between Macro Scenario Projections and Sensitivity Test Assumptions
- Table 5: Bank Solvency Stress Test Results: Breakdown by Business Model
- Table 6: Cash-Flow Based Stress Tests Outflow Rates from Deposits, Liquidity, and Credit Lines
- Table 7: Cash-Flow Based Stress Tests—Haircuts on Counterbalancing Capacity
- Table 8: LCR Stress Test Parameters
- Table 9: Summary of Bank Liquidity Stress Test Results
- Table 10: Investment Fund Stress Test Sample—Statistics by Fund Type
- Table 11: Estimated Initial Change in NAV Following Adverse and GFC Scenarios
- Table 12: Estimated Final Change in NAV Following Adverse Scenario
- Table 13: Results of the Investment Fund Liquidity Stress Test
- Table 14: Money Market Funds—Statistics by Fund Type
- Table 15: Insurance Stress Test Sample
- Table 16: Insurance Stress Test Specification

### Appendices and Technical Materials
- I. Risk Assessment Matrix
- II. Stress Test Matrix
- III. Summary Results of the Stress Tests Across Sectors
- IV. Technical Details on the Stress Test Adverse Scenario Calibration
- V. Methodological Assumptions for Banks’ Balance Sheet and Profit Projections
- VI. Satellite Models for Credit Risk—Technical Details
- VII. Investment Fund Stress Testing Methodology
- VIII. Price Impact of Asset Sales

### Glossary (selected acronyms preserved exactly as in source)
- AIF
- AFS
- AuM
- BCL
- CAA
- CCP
- CDS
- CET1
- CNAV
- COREP
- CRD
- CRR
- CRS
- CSSF
- EBA
- ECB
- EDF
- EL
- ELA
- EM
- EoAoL
- EOF
- EU
- EUR
- FDI
- FE
- FINREP
- FSGM
- FSSA
- FX
- GBP
- HFCS
- HFT
- HQLA
- HY
- IBB
- ICAAP
- IF
- IMF
- IRB
- LGD
- LMT
- LSI
- LVA
- MFI
- NAV
- NFC
- NSFR
- OFI
- OLS
- QRT
- OTC
- O-SII
- PIT
- RAM
- RCR
- RFR
- ROW
- RRE
- RWA
- SA
- SCR
- SI
- STA
- STE
- STeM
- TD
- TN
- TNA
- TTC
- U.K.
- URB
- U.S.
- USD
- UCITS
- VA
- VAR
- VIX

*Source: EXECUTIVE SUMMARY (1luxea2024008) — Luxembourg FSAP executive summary, as contained in the provided content unit.*

### EXECUTIVE SUMMARY

### EXECUTIVE SUMMARY

### Overview and context
- The Luxembourg financial system is described as highly interconnected, diverse and complex, and has displayed a high level of resilience historically amid heightened economic, financial, and geopolitical uncertainty.
- Key macroeconomic signals:
  - Real GDP: According to the October 2023 WEO, output was expected to contract by ½ a percent in 2023 before rebounding to 1½ percent in 2024.
  - National statistical office (December 2023): forecast a recession by 1 percent in 2023 before a recovery by 2 percent in 2024.
  - Unemployment rose to 5.6 percent in October 2023.
- Financial-sector scale and structure:
  - Financial sector employs 55,000 people (about 11 percent of total employment).
  - Bank assets are about 12 times GDP.
  - Insurance sector size is about 3 times GDP.
  - Investment fund sector: over 3,300 directly regulated Undertakings for Collective Investment and 122 money market funds registered as of August 2023.
  - Assets (August 2023): non-MMF funds held over 6 trillion euro; MMFs held 0.5 trillion euro.

### Stress-test framework and scenario
- Integrated stress tests across banking, insurance, and investment fund sectors included:
  - A common macroeconomic and financial scenario: a severe global recessionary scenario coupled with high inflation—a deep recession combined with a sharp increase in short-term rates in the first year of a three-year horizon. The shock incorporates domestic and international real estate sector risks.
  - Transmission channels: fund redemption shock → bank deposits and prices of securities; funding stress and fire sales → solvency impacts.
- External comparisons and precedent scenarios used:
  - Deposit run scenarios akin to those experienced in the US and Switzerland in March 2023 (Credit Suisse and First Republic) and stress linked to some “weak” global parent banks identified in the IMF’s October 2023 GFSR.

### Banking sector: main stress-test findings
- Aggregate resilience:
  - Under the adverse scenario, the system-wide capital ratio would decline significantly but remain “very well-capitalized” due to strong starting capital buffers.
  - Common Equity Tier 1 (CET1) capital ratio was 21½ percent in 2023Q2—6 percentage points above the euro area average.
- Capital shortfalls and recapitalization needs:
  - Four banks (together making up less than 10 percent of banking sector assets) would fall below the CET1 hurdle rate of 8 percent; recapitalization needs amount to 0.6 percent of nominal GDP.
  - Recapitalization needs based on the leverage ratio minimum of 3 percent would reach 0.9 percent of nominal GDP, with five banks (together making up less than 20 percent of banking sector assets) below this threshold.
- Key drivers of capital decline:
  - Credit risk from corporate portfolios.
  - Market risk from bond portfolios (under a “no hedge” assumption).
- Concentration risk:
  - High concentration: several banks would be severely affected if their five largest individual exposures (mostly toward non-parent financial institutions) defaulted simultaneously.

### Liquidity and funding risks
- Liquidity buffers are sizeable across banks, and the majority would be able to sustain deposit runs comparable to March 2023 episodes, but some institutions require attention.
- In a deposit-run scenario mirroring Credit Suisse and First Republic (March 2023), six banks could have funding gaps.
- Additional stress (from weak global parent banks) would produce funding gaps in a couple more banks.
- Recommendations stress:
  - Use cash-flow data for all significant currencies and jurisdictions to refine liquidity monitoring and stress testing.
  - Continue monitoring bank liquidity risk from fund/bank connections, including depositary-fund manager group links and connections with potentially weak parent banks.
  - Complement market-based indicators with latest EU-level stress test results to monitor subsidiaries of foreign parent banks.

### Investment funds: liquidity, redemptions, and spillovers
- Growth and structure:
  - Since the 2017 FSAP, the directly regulated sector expanded by over 70 percent in total assets.
- Stress-test findings:
  - All UCITS and the majority of AIFs have sufficient liquidity buffers to face rapid redemptions, with small second-round effects.
  - A small number of mixed fund AIFs (loan funds), representing a very small share of total net assets in the stress-test sample, were identified (under an HQLA approach that does not incorporate LMTs and longer redemption frequencies) as not having enough liquid assets to face redemptions.
  - If funds sell securities to stem redemptions, price impacts on securities are small and second-round impacts on the financial sector are manageable.
- Money Market Funds (MMFs):
  - Four Low Volatility Net Asset Value (LVNAV) funds and two small Constant NAV (CNAV) funds experience falls in shadow NAV above 20 basis points under the instantaneous, severe shocks of the adverse scenario—signaling room to further enhance resilience, though not characterized as alarming in this exercise.
- Emerging risks and structural considerations:
  - Increasing share of AIFs, higher interlinkages within the fund sector and with OFIs, rise of less liquid strategies, “retailization,” and strategies akin to Liability-Driven Investments (LDIs) increase heterogeneity and complexity and may generate hard-to-quantify risks.
  - Authorities are encouraged to better identify financial intermediation roles and amplification channels of OFIs and AIFs.

### Insurance sector findings
- Aggregate resilience:
  - Insurance companies were found resilient to increased market volatility and rises in interest rates despite hits to unit-linked business.
  - The sector’s assets decline by 9 percent overall, largely offset by a similarly sized decline in liabilities due to higher interest rates.
- Sources of stress:
  - Guaranteed business: largest capital decline stems from higher sovereign and corporate spreads.
  - Unit-linked business: significant effect from reduction in future fee income as assets under management decline after stress.
- Supervisory recommendations:
  - Continue improving quality of supervisory reporting data, especially on investments and derivative holdings.
  - Implement regular top-down stress tests and sensitivity analysis, starting with standard formula users and later extending, to the extent possible, to internal model users.

### Key supervisory and policy recommendations (selected)
- Systemic risk and bank risk analysis:
  - Improve top-down supervisory stress test models to incorporate liquidity–solvency interactions.
  - Use cash-flow data for all significant currencies and jurisdictions to refine liquidity monitoring and stress testing.
  - Continue monitoring bank liquidity risk from fund/bank connections, including depositary–fund manager group links.
  - Complement market-based indicators with the latest EU-level stress test results for foreign-parent subsidiaries.
- Insurance sector:
  - Further improve quality of supervisory reporting data, especially on investments and derivatives.
  - Implement regular top-down stress tests and sensitivity analysis for the insurance sector, starting with standard formula users and expanding to internal model users.
- Investment funds and OFIs:
  - Carefully analyze interlinkages of investment funds with the OFI sector, including through loans, to better understand vulnerabilities and amplification mechanisms.
  - Further monitor liquidity mismatches in newly emerging investment strategies of large AIF funds, including semi-liquid structures, strategies akin to liability-driven investments, and those promoting “retailization”.
- Timing conventions used in recommendations:
  - “C-Continuous”; “I-Immediate” is within one year; “NT-near-term” is 1–3 years; “MT-medium-term” is 3–5 years.

*Source: EXECUTIVE SUMMARY*

### 8. Luxembourg maintains its status as the top domicile for UCITS, but Alternative

### 8. Luxembourg maintains its status as the top domicile for UCITS, but Alternative

### Fund composition and recent trends
- In 2016, 84 percent of the net assets of Luxembourg's funds were UCITS and 16 percent were AIFs.
- By 2023, UCITS comprise 71 percent and AIFs 29 percent, with AIFs nearly doubling their share since 2016.
- Within AIFs, unregulated structures (indirectly regulated through the investment fund manager) are increasingly preferred, especially for new funds.
- Growth in AIFs is driven by fund strategies investing in less liquid assets such as real estate, private equity, and private debt.
- Direct and indirect exposures of funds domiciled in Luxembourg to crypto assets were roughly €1 bn in 2022.

### UCITS strategy shifts and market share
- The share of assets held by bond UCITS has decreased; this was mostly driven by valuation effects from bond underperformance relative to equity markets rather than by net outflows.
- Equity UCITS grew largely due to expansion of holdings of US equities.
- The largest UCITS categories (bonds, equity and mixed funds) stood at roughly 80 percent of total UCITS’ assets in 2023.
- Note: 2023 numbers refer to March 2023.

### Interconnectedness and systemic risk in the investment fund sector
- Investment funds are more interconnected via other domestic investment funds and OFIs since the last FSAP, partly due to AIF growth (real estate, private equity, debt).
- For a sample of large open-ended funds, mixed funds and funds of funds have the highest holdings of equity of other investment funds.
- Potential systemic risk channel: a surge of net redemptions in some funds could cascade into generalized redemption pressure as funds holding IF equity redeem their shares to meet initial redemptions.
- Liquidity mismatches may be increasing for some funds, as emerging funds of funds invest in funds holding less liquid assets often through OFI intermediaries.
- Some AIFs combine investment in less liquid assets with more frequent redemptions (a “retailization” trend).
- Shares of financial derivatives, interfund loans and loans from OFIs in funds’ liabilities have increased significantly since 2015, but in absolute terms remain small compared to funds’ aggregate balance sheet.

### Money Market Funds and liquidity characteristics
- Money Market Fund assets stand at roughly 5 times domestic GDP.
- Luxembourg is the second largest domicile for MMFs in the euro area after Ireland.
- A considerable share of MMFs is held in Constant NAV (CNAV) and Low Volatility NAV (LVNAV) MMFs.
- LVNAVs must use variable NAV based on mark-to-market/model valuation for subscriptions and redemptions if portfolio market value deviates from book value by more than 20 basis points, which could trigger loss of confidence.
- MMFs have withstood recent interest rate hikes and market turmoil and remain an attractive vehicle for investment in liquid assets.

### Real estate exposures
- Total exposure to the CRE market stood at 330 billion euros in December 2022, roughly 5 percent of total AUM of the entire sector.
- Most CRE exposure is to the European market; Germany accounts for 23 percent of total CRE holdings of investment funds.
- Direct exposure to the domestic real estate market is about 3.3 billion euros.

### Liquidity Management Tools (LMTs) and supervisory work
- Effective usage of LMTs by fund managers can mitigate risks from alternative funds; CSSF work shows swing pricing is effective in containing materialization of risks.
- As the industry becomes more heterogeneous, fund managers need to explore LMT efficacy under different circumstances.
- Monitoring outcomes and developing a framework of best practices is important; future ESMA and IOSCO work in 2024 is envisaged, to which the CSSF is contributing.
- CSSF research examples: papers on LMTs (Carpantier and others (2022)) and on large redemptions during the covid crisis (Carpantier (2021)).

### Developments in risk analysis and data collection
- Since the last FSAP, Luxembourg authorities substantially improved analysis of investment fund sector risks and plan further improvements.
- CSSF and BCL have improved liquidity stress testing frameworks and implemented MMF specific stress testing (Lo and Carpentier (2023) methodology uses a time to liquidation approach).
- CSSF circulars transpose ESMA guidelines on liquidity stress testing and MMF stress tests, and IOSCO recommendations on liquidity risk management for open-ended UCI.
- The data collection framework has been expanded and enhanced.
- Authorities acted swiftly on emerging issues, including risks posed by LDI funds after the UK mini-budget crisis.

### Net redemptions and recent flows
- Net redemptions from investment funds picked up in 2023Q3, cumulatively reaching 2½ percent of net assets since 2022Q2.
- Bond and equity funds faced the highest outflows; real estate saw some positive changes in total net assets.
- Given the size and protracted nature of redemptions to date, they do not pose liquidity risk concerns nor have led to instability, but may leave the sector less robust versus prior positive net flow trends.

### Banking sector: structure and cross-border orientation
- Luxembourg is home to 118 banking institutions.
- Slightly over 70 percent of them have European origin; nine banks are of Luxembourgish origin. Banks from the United States, China, Brazil and other non-European countries are also represented.
- Total assets of the banking system stood at 12 times of GDP in 2023Q1; about 92 percent of them were owned by foreign banking groups and institutions.
- Close to 80 percent of banks offer corporate finance, private banking, custodian banking and activities linked to investment funds.

### Banking balance sheet composition and funding
- The share of loans in total assets has remained fairly stable, fluctuating between 80-83 percent over the last five years.
- Out of the outstanding loan stock of 782 billion euro in 2023Q1, loans to affiliated banks and other deposit-taking corporations represent about 37 percent.
- Loans to other sectors, including nonfinancial corporations, constitute about a third of the total volume.
- In 2023Q1, deposits from other banks, investment funds, non-bank financial institutions, and other financial intermediaries accounted for 78 percent of total bank deposits.
- On average, ¾ of deposits from banks represent deposits from affiliated entities, though this share has declined somewhat recently due to outflows driven mostly by a reduction of deposits by investment funds and banks.

### Banking sector financial soundness indicators
- Capital adequacy ratio of the banking sector was 23 percent as of June 2023.
- Tier 1 capital represents 96 percent of total regulatory capital.
- The average risk weight is 36 percent, with wide dispersion across bank business models: 21 percent for the custodian banking group to close to 62 percent for the corporate finance group.
- Nine banks use internal models (IRB banks), making up 56 percent of total assets of the sample.
- Aggregate Liquidity Coverage Ratio (LCR) is 150 percent.
- Loan-to-deposit ratio is 68 percent overall; corporate finance banks have the highest loan-to-deposit ratio at 95 percent.
- Aggregate nonperforming loan ratio increased from 1.0 percent in 2017 to 1.8 percent in June 2023.
- Stage 2 loans continue to rise following forbearance during the pandemic; forbearance measures have since been withdrawn.
- Profitability has picked up in 2023 after a downward trend between 2016 and 2020; Luxembourg banks' return on assets recently rose (value not specified in supplied content).

*Source: 1luxea2024008*

### 1.1 percent in June 2023, an unprecedented level since the 2007/2008 global financial crisis.

### 1luxea2024008 - 1.1 percent in June 2023, an unprecedented level since the 2007/2008 global financial crisis.

### A. Banking sector: interest rates, margins, and profitability
- Net interest margin widened significantly as banks translated policy rate increases quickly into higher lending rates while funding costs rose more mutedly.
- Only about half of banks’ loan portfolios were immediately repriced because of the share of fixed-rate loans; net interest margin expected to widen further as:
  - current loans mature, and
  - new loans are priced at higher lending rates.
- The negative impact of higher lending rates on borrowers’ creditworthiness has not fully materialized; no sharp rise in the cost of risk observed to date.
- Bank categories and profitability (June 2023):
  - Universal, retail and commercial banking: Return on Assets of 0.8 percent.
  - Private banking: Return on Assets of 0.5 percent.
  - Custodian banks: Return on Assets of 0.4 percent.
- Financial soundness indicators (based on the stress test sample of 39 banks; selected indicators as reported):
  - Regulatory Capital to Risk-weighted Assets: 22.6%
  - Tier 1 Capital to Risk-weighted Assets: 21.7%
  - CET 1 Capital to Risk-weighted Assets: 20.9%
  - Tier 1 capital to Total Assets: 7.7%
  - Average Risk Weight: 35.6%
  - Liquid assets to Total Assets: 39.7%
  - Loans to Deposits: 67.7%
  - Liquidity Coverage Ratio (LCR): 229.1%
  - Nonperforming Loans to Total Gross Loans: 1.8%
  - Return on Equity: 5.3%
  - Return on Assets: 0.8%

### B. Insurance sector: size, structure, assets, and profitability
- Sector size and role:
  - Sector assets equaled €289 bn in June 2023, corresponding to 372 percent of GDP.
  - The insurance sector serves as a European hub, notably in non-life (re)insurance and unit-linked life insurance.
  - Insurance business is conducted mainly cross-border, with more than 90 percent of gross written premiums stemming from cross-border business.
- Liabilities and business mix:
  - Technical provisions are dominated by unit-linked business, which accounts for 63 percent of aggregated insurance liabilities.
  - Technical provisions for traditional life insurance account for 15 percent and non-life insurance 15 percent of aggregated insurance liabilities.
  - Luxembourg is the sixth largest market for unit-linked business in the European Union with around EUR 164bn (absolute numbers).
- Asset allocation:
  - Assets backing unit-linked insurance account for 75 percent of life insurers’ total assets (up from 67 percent in 2017).
  - 59 percent of these assets are investment funds.
  - Non-life insurers’ portfolios are geared toward conservative investments; sovereign and corporate bonds represent 34 percent of total assets in the non-life sector.
  - Unit-linked assets (before look-through) are concentrated in markets with large investment fund sectors (including Luxembourg, France, and Ireland); assets backing traditional business are more diversified.
- Profitability and yields:
  - Profitability is subdued overall.
  - Life insurers: Return on equity for the median life insurer amounted to 5 percent in 2022, slightly below the five-year average.
  - Investment yields for life insurers stood at 1.6 percent on average in 2022.
  - Underwriting remains profitable in the non-life sector despite inflationary impacts in 2022 and some specialized business experiencing rising claims.
  - Competitive pressures in life insurance target high net worth individuals; captive insurers in non-life typically do not aim for high profit margins.

### C. Captive Financial Institutions and Other Financial Intermediaries (OFIs)
- Sector characteristics:
  - The OFI sector is large and diverse; much of its size reflects multinational corporations’ use of holding companies and intragroup financing structures.
  - Over 90 percent of OFI assets are held by the largest subgroup termed “captive financial institutions and money lenders” (sector S127 in the European System of Accounts 2010).
- Entity counts and asset concentration:
  - Based on BCL-reporting subsample, about 2,700 captive financial institutions and money lenders had assets of more than 500 million euros representing 85 percent of the population (by asset size) at end-2018.
  - Extrapolated to the whole population, this suggests more than 50,000 entities in the S127 sector.
  - About 80 percent of their assets were held by holding companies or institutions set up to facilitate intragroup lending.
- Economic activity and geographic composition:
  - Seven industries account for roughly 80 percent of assets held by sector entities, with the largest shares in:
    - Finance and insurance: 24%
    - Chemicals and non-metallic products: 24%
    - Electrical, medical, and optical equipment: 16%
    - IT-related services: 13%
    - Mining, drilling, and quarrying: 9%
    - Food, beverages, and tobacco: 8%
    - Wholesale and retail trade: 6%
  - By affiliated-group domicile (average for 2014-19):
    - USA: 59%
    - GBR: 11%
    - EA: 9%
    - Other: 21%
  - When considering the balance sheet counterparty of S127 firms (average for 2014-19):
    - Asset side geographic distribution: LUX 27, EA 22, GBR 8, USA 7, Other 37 (percent).
    - Liability side geographic distribution: LUX 35, EA 21, GBR 10, USA 7, Other 27 (percent).
  - Close to 50 percent of parent companies have 2 or more entities registered in Luxembourg, indicating significant intra-group exposures within the S127 sector.

### D. Interconnectedness of sectors and cross-border linkages
- General observations:
  - The financial system is complex, highly interlinked domestically, and highly connected to the rest of the world (ROW).
  - Investment funds (IFs) have the highest exposure to ROW in absolute balance sheet positions.
- IF and OFI external exposures (2022, selected shares):
  - IFs: close to 80 percent of assets and 90 percent of liabilities are with ROW.
  - OFIs: about 60 percent of assets and 45 percent of liabilities represent cross-border positions.
  - MFIs: share of assets related to ROW approaches 90 percent; over 50 percent for liabilities.
- Cross-sector linkages:
  - Nonfinancial corporations have significant balance sheet positions with MFIs, OFIs, and IFs.
  - Funds’ holdings of MMF shares and bank deposits account for about one-fourth of MFIs’ financial liabilities.
  - Households’ assets are predominantly held in MFIs and IFs; household liabilities are almost exclusively directed at the MFI sector (mortgage and consumer loans).
  - OFIs issue sizable amounts of debt securities mostly held by other OFIs and the ROW; OFIs also grant loans to ROW and other domestic sectors.
- Data and transparency:
  - Given the complexity and limited transparency of linkages, aggregated data make it difficult to fully assess channels of contagion; more granular, entity-level data would be beneficial.
  - Di Filippo (2024) provides an example analysis of FDI links between IFs and affiliated OFIs at the entity level for funds with affiliated entities connected to private equity and real estate.

### E. FSAP stress testing strategy and methodology
- Objective and scenarios:
  - The FSAP stress testing exercise assesses the capacity of the financial system to withstand baseline and adverse macroeconomic shocks.
  - The baseline relates to the October 2023 IMF World Economic Outlook (WEO) macroeconomic projections for 2024-2026.
  - The adverse scenario comprises a set of severe but plausible macroeconomic shocks to explore potential weaknesses and transmission channels.
- Purpose and outputs:
  - Tests aim to identify priorities for policy actions, such as reducing specific exposures or building capital and liquidity buffers.
  - The FSAP process helps identify informational and methodological gaps and assesses preparedness for financial distress.
- Methodological notes:
  - FSAP stress tests differ from other tests (CSSF, BCL, EBA, ECB) in credit risk models, bank samples, and assumptions relating to macro hedges and other parameters.
  - The FSAP team conducted tests in close cooperation with the CSSF, the BCL and the ECB, with access to supervisory data in a secure data room at the ECB’s premises, on either aggregate or individual basis.

*Sources: ECB; and IMF staff calculations.*

### 34. Although stress tests are useful to explore vulnerabilities in a financial system, results

### 34. Although stress tests are useful to explore vulnerabilities in a financial system, results must be interpreted with caution

### Caveats, Purpose, and Scope
- FSAP stress tests have a macrofinancial perspective aimed at assessing the resilience of the banking system as a whole to a set of macroeconomic and financial shocks; they differ from microprudential stress tests and are not intended to lead to supervisory actions at the level of individual banks.
- FSAP credit loss estimates and solvency projections in the adverse scenario are subject to data and methodological limitations, including:
  - no reflection of possible economic hedges,
  - use of benchmarks taken from other countries due to a lack of relevant historical data (see paragraph 46 for further details).
- Choices must be made regarding the severity of shocks; adverse scenarios typically combine external and domestic shocks that (ex-ante) have a very low probability of occurrence and, by construction, should not be interpreted as macroeconomic "forecasts."
- Possible mitigating actions, including those under the recovery and resolution legal framework, are not considered.

### Consistency and Comparability across FSAPs
- Consistency is sought across FSAPs in terms of stress test methodology; however:
  - country-specific environments and idiosyncrasies preclude complete comparability between stress test results across FSAPs,
  - differences in methodologies and hurdle rates may translate into different results for similar risk profiles.

### Top-Down Stress Tests: Coverage, Horizon, and Approaches
- The top-down (TD) stress tests examined resilience to solvency, liquidity, and contagion risks.
- Banking sector stress tests:
  - included a top-down exercise based on macroeconomic scenarios and sensitivity analyses conducted by the FSAP team,
  - assessed impact of severe but plausible external and domestic shocks over a three-year horizon (2024–2026), based on data available through June 2023,
  - used satellite models based on international banking crisis experience and IMF methodologies to assess effects on individual banks' profitability and capitalization,
  - included sensitivity stress tests for individual shocks.
- Liquidity tests:
  - assessed banks' capacity to withstand large withdrawals of funding using a maturity ladder analysis and supervisory information,
  - performed both on an aggregate basis and by currencies.
- Investment fund stress tests:
  - designed to evaluate liquidity risk against severe yet plausible redemption shocks,
  - aim to identify fund categories more vulnerable to liquidity risks and potential propagation of shocks through the financial system,
  - incorporate second-round effects from asset sales (macroprudential perspective) in addition to microprudential focus on individual funds,
  - include “solvency” stress tests on deviation of constant NAV and shadow NAV of Money Market Funds (MMFs) to evaluate resilience to interest rate hikes and widening credit spreads, and on impact on value of assets and liabilities for all funds in the sample,
  - do not take into account use of liquidity management tools (LMTs) and the often lower redemption frequency of AIFs (typically monthly or quarterly for several ones), which could mitigate the impact of shocks to some extent (as was seen during the pandemic in 2020).

### Financial Shock Calibration (First Round)
- Calibration uses the Financial Shock Simulator (Gross, Henry, and Rancoita 2022), which:
  - employs estimation of an empirical copula via a Monte Carlo simulation,
  - assumes a number of risk factors, each following a stochastic process with independent increments (Markov property),
  - uses a Monte Carlo simulation/bootstrapping routine to estimate a joint probability density function and calculate conditional expected shortfall at the tail of the conditional distribution.
- In this application:
  - 27 risk factors are included in the simulation.
  - Sovereign and corporate bonds and equities from France, Germany, Japan, Luxembourg, United States, other Euro Area, and the rest of the world at daily frequency constituted the universe of risk factors.
  - U.S., Germany, and U.K. corporate bonds and equities were chosen to be conditioning variables.
  - Forward paths are simulated 10,000 times for the horizon of 1-year and conditional expected shortfalls are calculated for the tail of the estimated empirical distribution.
- Note: The Financial Shock Simulator was calibrated separately from the main adverse scenario, which was based on Bayesian VAR.

### Macrofinancial Scenarios: Baseline and Adverse (First Round)
- Integration:
  - The three segments—investment funds, insurance, and banking sectors—were subjected to stress tests incorporating information from the same macro scenarios.
  - Solvency stress tests assess whether institutions have enough capital to withstand economic and financial shocks.
  - The three stress tests are integrated through a common macro scenario and methodological choices to model interactions; sensitivity tests assessed the impact of individual shocks.
- Baseline:
  - Based on the October 2023 World Economic Outlook projections.
- Adverse scenario design and drivers:
  - A severe recessionary scenario coupled with inflation affects solvency of banks, funds, and insurers.
  - Driven by external geopolitical and financial shocks, increase in investors’ risk aversion and confidence losses.
  - Based on the Risk Assessment Matrix; assumes three external shocks transmitted via international macroeconomic spillovers and pure financial channels.
- External risks most likely to materialize:
  - Escalation of war in Ukraine leading to complete disruption of gas supply in context of already low gas reserves, causing severe recession in the euro area and amplifying inflationary tensions, plus retrenchment in trade and financial flows.
  - Rising geopolitical concerns and inflationary pressures increasing short-term rates, initially benefiting banks’ interest margins but producing marked-to-market losses on bond portfolios and higher private sector loan losses over the next three years; funds and insurers facing marked-to-market losses leading to redemptions and lapses by policy holders.
  - Abrupt tightening of financial conditions translating into sharp rise in risk aversion and risk premia, flight to quality and dollar assets, and sharp redemption flows from international investors in investment funds.
- Scenario composition:
  - Three layers: (i) global layer (geopolitical risks in systemic advanced economies), (ii) euro area layer (severe recession with high inflation and escalating risk aversion), (iii) Luxembourg-specific layer (shock to Luxembourg’s financial sector feeding through domestic real estate market and domestic economic confidence).
- Key transmission channels in Luxembourg:
  - External services demand, financial flows into investment funds, and the real estate market.
  - Recession in the euro area leads to a drop in Luxembourg exports.
  - Decompression of credit spreads from rising interest rates aggravates market volatility and liquidity.
  - Higher financial volatility stresses domestic funding markets, contraction in banking intragroup exposures between foreign parents and local subsidiaries, intragroup deposit outflows, and reduction in large intragroup assets of Luxembourgish subsidiaries.
  - Domestically oriented banks experience higher funding costs due to rises in deposit rates and switch from sight to term deposits by non-financial customers.
  - Rise in market rates passes imperfectly to lending rates; variable rate loans still account for a large share of total loans; effects on demand for credit and corporate creditworthiness; impact on mortgage demand and borrower creditworthiness.
  - Decline in residential and commercial real estate prices causes adverse wealth effects and negative feedback to domestic demand and a deflationary process.
- Fiscal effects:
  - Shocks affecting the financial sector adversely affect the fiscal balance, increase credit risk of the sovereign debt portfolio, widen credit spread against Germany, limit fiscal space for countercyclical fiscal policy, and amplify domestic consumption shock and house price correction.
- Stress test coverage and horizon:
  - Tests covered 39 banks (19 SIs and 20 LSIs), over 1000 open-ended funds subject to the full reporting scope, and 20 insurers—with 80–90 percent of asset coverage in each sector.
  - Bank sample included: 8 universal, retail and commercial banks; 14 private wealth management banks; 9 corporate finance banks; 5 custodian banks; 2 covered bond banks; and one payment services bank.
  - Stress test examined a baseline and an adverse macroeconomic scenario over a three-year horizon. The first year of the shock is 2024 and the scenario runs until 2026.
  - UCITS covered in the sample fulfill at least one of the following two criteria: net assets above EUR 500 million; or use of VaR for the calculation of global exposure covered by Article 42(3) of the 2010 Law with an average gross leverage during the semester greater than 250 percent.
  - A three-year projection was chosen because forecast errors appeared too large over periods longer than three years.
- Adverse scenario macro projections and severity:
  - Models used: IMF Global Macro-Financial Model (GFM) for external variables and a Bayesian VAR developed for domestic variables for this FSAP (Appendix III).
  - Variables included: real GDP growth, HICP and CPI inflation rate, unemployment rate, the three-month interbank rate, the nominal government bond rate, lending rates, the euro-dollar exchange rate, and real estate price growth.
  - Adverse shock envisaged a 5.9 percent cumulative drop in the level of GDP over 2024-25, with spikes in inflation requiring sharp increases in global short-term rates, increasing short-term interest rate in Luxembourg and decreasing prices of residential and commercial real estate.
  - The GDP shock corresponds to - 10 percentage points compared to the baseline level of output at the trough, representing a 2.3 standard deviation shock from the baseline and a 1.4 standard deviation shock from the mean of the historical distribution.
- Impact on funds and insurers:
  - Valuation losses on assets due to financial shocks, with related redemptions and lapses.

### Market Contagion and Liquidity-Solvency Interaction Effects (Second and Third Rounds)
- Second-round effects:
  - Investment funds selling cross-border liquid assets (assuming they do not sell Luxembourg securities) could impact prices and trigger second-round impacts on NAVs of funds and valuation of securities in banks and funds.
- Third-round effects:
  - Selling liquid assets to stem bank deposit outflows could entail capital costs for banks if standing facilities are not available at the central bank.
  - Almost half of the banks with 20 percent of assets in the stress test sample (all foreign subsidiaries with diverse business models) do not have standing facilities set up with BCL.
  - If banks sell held-to-maturity (HTM) assets—almost 80 percent of large banks’ securities portfolios with unrealized losses—there could be liquidity-solvency interactions from realizing losses on capital; stress tests explored this.
  - Large-scale securities sales, especially if multiple sectors sell to stem redemptions and lapses, could produce fire-sale implications on securities prices with spillovers across the financial sector.

### Bank Solvency Stress Tests: Types, Frameworks, and Hurdle Rates
- Types:
  - Scenario-based solvency stress tests assess banks' robustness to shocks to capital related to credit and market risks.
  - Banking sector stress tests included a Top-down (TD) exercise based on macroeconomic scenarios and sensitivity analyses.
- Regulatory frameworks:
  - Stress tests were based on applicable international and national regulatory frameworks, conducted mainly with reference to the Basel III framework, and also the European Union and national frameworks as defined by CRD V, Regulation on Prudential Requirements, national law and CSSF regulation.
- Main hurdle rates:
  - Hurdle rates related to the Common Equity Tier1 (CET1) capital ratio and the leverage ratio, set according to Basel III fully-loaded definitions of capital requirements.
  - Stress tests based on minimum capital ratios under Pillar I and bank-specific Pillar II requirements.
  - Hurdle rates used:
    - For CET1: the Overall Capital Requirement Ratio (OCR, including the minimum requirement of 4.5 percent, the capital conservation buffer of 2.5 percent, average Pillar 2 requirement of 0.5 percent as well as an average Countercyclical buffer of 0.5 percent, and triggering restrictions for dividend distributions.), averaging 8 percent in the sample.
    - For the leverage ratio: the minimum requirement of 3 percent.
- Assessment methods:
  - Effects of shocks on individual bank profitability and capitalization were assessed using satellite models and methodologies developed by IMF staff; satellite models link credit risk parameters (PDs) with various macrofinancial variables (see Appendix IV).

*Source: IMF staff, Luxembourg FSAP chapter text excerpt.*

### Appendix V).

### Appendix V)

### Sensitivity analysis: scope and assumptions
- Sensitivity stress tests assessed vulnerabilities to individual shocks: a decline in prices of domestic sovereign securities; an increase or decline in interest rates affecting banks' net interest income; a depreciation or appreciation of the euro nominal effective exchange rate that triggers direct gains or losses in banks with net open FX positions; and a decline in stock prices.
- Indirect effects of a nominal depreciation of the euro on credit quality were not assessed because the share of foreign currency lending in total loans is very low in Luxembourg.
- Sensitivity tests also assessed sensitivity to default of the 2–10 largest exposures (concentration risk).
- Sensitivity tests are static: they assessed the instantaneous impact of shocks on banks' balance sheet positions as of June 2023, with risk-weighted assets assumed constant after shocks.
- Assumptions for sensitivity tests are usually harsher than macro scenario projections (Table 4):
  - Lending rate shock: Adverse scenario 324 bps; Sensitivity test +/- 500 bps
  - Domestic sovereign interest rate shock: Adverse scenario 510 bps; Sensitivity test +/- 500 bps
  - Foreign exchange rate shock: Adverse scenario -7 percent; Sensitivity test +/- 30 percent
  - Equity price shock (to Luxembourg stock price index): Adverse scenario -48 percent; Sensitivity test -50 percent

### Banking sector solvency stress test: headline results
- Baseline scenario:
  - Aggregate CET1 ratio would remain stable slightly below 22 percent.
  - Profit creation would be offset by strong asset growth and dividend distribution.
  - Two banks—three banks if hedging is not considered—would struggle to meet the 3 percent leverage ratio hurdle rate due to faltering profitability and a low initial leverage buffer.
- Adverse scenario:
  - Aggregate CET1 ratio would drop by about 4 percentage points in 2024 to 17.6 percent before gradually recovering from 2025.
  - Credit loss provisions, market losses and the change in risk-weighted assets are the main drivers of capital depletion.
  - Four banks making up less than 10 percent of total banking sector assets would see CET1 ratio fall below the 8 percent hurdle rate, translating into a capital shortfall equivalent to 0.6 percent of nominal GDP.
  - Regarding the leverage ratio hurdle rate, five banks would become undercapitalized in 2024 after the first-round macrofinancial shock; a sixth bank would do so after additional funding costs from liquidity stress and pledging Held-to-Maturity securities with the BCL.
  - Two banks would have capital ratios below both CET1 and leverage thresholds.

### Credit risks in the loan book
- Total loans represent 51 percent of the assets of the 39 banks in the stress test sample.
- Loans to households: 12 percent of total assets; 19.5 percent of assets of the five domestically oriented banks.
- Debt securities: 12.5 percent of total assets, most held-to-maturity.
- Loan book exposures extend to euro area corporates outside Luxembourg (Belgium, France, Germany, the Netherlands).
- Largest exposures of the thirty-nine banks are to non-financial corporations, followed by credit institutions (including parent banks).
- Point-in-Time (PiT) PDs under the adverse scenario:
  - Mortgage retail segment PD increases from 0.9 percent as of June 2023 to a peak of 3.5 percent under the adverse scenario (model uses international benchmark from the 2008/2013 Irish banking crisis).
  - Floor imposed on PDs for financial institutions so they could not go below the level experienced during the 2008/2009 global financial crisis.
- PDs in the corporate loan portfolio climb from 1.9 percent in June 2023 to 5 percent in 2025 in the adverse scenario.
- Expected credit losses under IFRS-9 over the three-year horizon:
  - Adverse scenario: €3.8 billion, equivalent to 1.9 percent of total banking system RWAs.
  - Baseline scenario: €1 billion, equivalent to 0.5 percent of total banking system RWAs.
  - In the baseline, new provisions are more than offset by net income before losses (6.4 percent of total RWAs).

### Market risks and scenario analysis
- Market risks included: interest, exchange rate and equity risks. No allowance for macro hedges in the main exercises.
- Losses or gains arise from open positions (currency, maturity, time-to-repricing mismatches) and valuation changes in securities (Available For Sale and Held For Trading).
- Valuation effects on debt security holdings assessed using a modified duration approach and average maturity proxies from BCL data.
- In the adverse scenario, losses due to decline in price of sovereign and corporate securities in AFS and HFT portfolios amount to €2.5 billion in the first two years, impacting the Capital Adequacy Ratio by 1.2 percentage points.
  - Explanatory factors: average AFS and HFT exposure of 2.8 percent of total assets and 22.6 percent of the total securities portfolio; moderate average maturity of bond portfolio averaging 4 years.
- Sensitivity analysis: for three weaker banks, assumption of 50 percent of marked-to-market bond portfolio being hedged was tested based on confidential data shown to the mission team.
- FX positions: banks benefit from modest net FX gains in the adverse scenario; euro depreciation against the U.S. dollar results in a small gain of €45 million.

### Interest rate risk: time-to-repricing and valuation channels
- Net interest income assessed using time-to-repricing buckets; expected losses/gains computed as product of repricing gap and interest rate changes.
- The 10-year German government bond rate is projected to increase by 3.2 percentage points in 2024 and then decline by 1 and 2 percentage points in 2025 and 2026, respectively, in the adverse scenario.
- Thirteen banks display a negative time-to-repricing gap for maturities shorter than 1 year (liabilities repriced faster than assets), leading to losses when interest rates rise.
- Aggregate repricing gap for the thirty-nine banks: €118 billion as of June 2023 for maturities below one year.
  - In the adverse scenario, this gap and the change in interest rates translate into a positive effect on the CAR of €5.4 billion, equivalent to 2.7 percent of Risk-weighted Assets, over the stress horizon.
  - For the thirteen banks with negative gap (all private or custodian banks), the gap amounted to -€ 7.5 billion in June 2023; it would translate into a decline in CET1 capital by €1.2 billion, equivalent to -4.4 percent of their RWAs, after the first year of the scenario.
  - Negative interest income in the first year for these banks reflects the negative repricing gap, negative interest income at one bank at the starting date, and a surge in funding costs; these banks would start benefiting from positive net interest margin from the second year.

### First-round effects and drivers of capital change
- Under the adverse scenario, expected credit losses are the main channel through which risks materialize.
- Aggregate CET1 ratio change and drivers (key findings):
  - Aggregate CET1 ratio would drop by 4.1 percentage points, from 21.7 percent as of December 2023 (after integration of expected net result based on data as of June 2023) to 17.6 percent in 2024, before gradual recovery from 2025.
  - Drivers of CET1 decline: credit losses (-1.9 percentage points of RWAs), decline in other comprehensive income (-0.6 percentage points of RWAs), and change in risk-weighted assets (effect of -0.6 percentage points).
  - Four banks (less than 10 percent of banking sector assets, including 1 SI and 3 LSIs) would fall below the CET1 hurdle rate of 8 percent, entailing recapitalization needs amounting to 0.6 percent of nominal GDP.
  - If capital conservation buffer and countercyclical capital buffer were available, two banks would still be below the CET1 hurdle rate.
  - If the Basel III minimum CET1 ratio of 4.5 percent were used as threshold, one small bank would not pass it.
- Business model differences:
  - Domestically oriented banks would be slightly more impacted than internationally oriented banks in the first year.
  - Private banks would experience the largest capital depletion, reflecting lower initial profitability and negative repricing gap.
  - Corporate finance banks would end up with the lowest average level of capitalization.
- Leverage ratio:
  - Tier 1 to total assets (leverage ratio) would decline from 7.7 to 6.8 percent in 2024 and to 7.4 in 2025.
  - Five banks would fall below the 3 percent hurdle rate in 2024, by an amount equivalent to 0.9 percent of nominal GDP, with two banks below both CET1 and leverage hurdles.

### Baseline scenario: vulnerabilities identified
- In the baseline scenario, the aggregate banking system would remain well capitalized overall.
- However, a few banks were identified as weak:
  - Three banks (8½ percent of total assets of the sample, with 1 SI and 2 LSIs) would struggle to meet the leverage ratio of 3 percent in the baseline in 2024.
  - One bank would see its CET1 ratio fall below the regulatory CET1 hurdle rate in 2025.

*Source: IMF staff calculations (Appendix V).*

### 2026. This is due to its low initial level of capitalization and low profitability.

### 1luxea2024008 - 2026. This is due to its low initial level of capitalization and low profitability.

### Caveats, data, and methodological limitations
- Top-down stress test did not take into account loan write-offs and cures.
- Intragroup risk mitigants such as economic hedges and financial guarantees were not considered because of the assumption on their enforceable nature during a crisis.
- With a hedging assumption of 50 percent of the AFS and trading books for three weaker banks, the average CET1 ratio of these three banks whose main hedging counterparties are not intra-group entities would be improved by 0.3 ppts but they would remain under the hurdle rate of 8 percent.
- Some assumptions were made for Loss Given Default and other supervisory parameters due to lack of historic depth and relevant crisis experience.
- Some items on banks' balance sheets, such as the derivatives, were not subject to stress.

### Bank solvency stress test: key results and dynamics
- Common Equity Tier 1 Capital Ratio (Fully-Loaded, in percent): reported across Dec -23, Dec -24, Dec -25, Dec -26 with Baseline, Adverse, Hurdle rate series depicted (hurdle rate referenced as 8 percent).
- Leverage Ratio (Fully-Loaded Tier 1/Total Assets, in percent): reported across Dec -23 to Dec -26 with Baseline, Adverse, Hurdle rate series.
- Net profit components in the adverse stress scenario (EUR billions): categories shown include Net Interest income, Loss provisions (Credit Risk), Net non-interest income, Other comprehensive income (AFS portfolio), Others, Taxes, Net effect.
- Contribution to Change of Capitalization Ratio in the Adverse Stress Scenario (points of RWA): components include Net Profit before stress, Loss provisions (Credit Risk), Trading and AFS book repricing risk, Change in Risk weighted assets, Net effect (values shown across 2024–2026 timeline).
- Capital Shortfalls (based on CET1 and leverage ratios, as a percent of nominal GDP): breakdowns across 2024–2026 with categories CET1>=10%, CET1<10%, CET1<8%, CET1<5.5%, CET1<4.5%, CET1<2%, CET1<0%.
- Banks' CET1 capital distribution in the Severe Stress Scenario (percent of banking system's assets) presented by deciles of capitalization.

### Breakdown by business model and orientation
- Results shown separately for Domestically-oriented Banks and Internationally-oriented Banks.
- CET1 ratios by business model (selected figures provided):
  - CET 1 ratio before stress: 20.9%, 25.2%, 42.2%, 14.6%, 42.4% (across business model categories).
  - CET 1 ratio - baseline (end of 1st year): 22.0%, 26.7%, 43.5%, 15.2%, 47.6%.
  - CET 1 ratio - adverse (end of 1st year): 17.0%, 16.8%, 42.6%, 13.8%, 46.5%.
  - Capital depletion in the adverse scenario: -3.8%, -8.5%, 0.4%, -0.9%, 4.1% (with -/+ implying declines/increases).

### Household stress and micro-data complementary test
- Household stress test used to benchmark banks' credit loss projections in the mortgage segment.
- Luxembourgish banks' internal models estimate low probabilities of default and loss given default in the housing loan segment, attributed to lending practices, households' payment culture, and benign recent macroeconomic conditions.
- A stress test using micro data and household characteristics was performed to analyze non-linear effects and to provide a complementary assessment of solvency risk (see Technical Note on Macroprudential Policies referenced in source).

### Second and third rounds of shocks: investment fund sales and liquidity-solvency feedback
- Investment fund fire sales price impact assumed equivalent to a 1 percent drop in the market value of German and French bonds.
  - Aggregate market loss: €156 million for the 39 banks in sample.
  - Impact on aggregate CET1 ratio: -0.1 percentage point.
  - No additional undercapitalization in 2024 from this channel.
- Liquidity-driven actions in the third round:
  - Sample of 39 banks: fifteen would experience a funding gap in the combined liquidity stress test scenario.
  - Nine banks would access central bank facilities and pledge HTM securities to BCL; assumed penalty of 100 bp over the main refinancing rate of the ECB.
    - Additional funding costs: €1.1 billion.
    - Further decrease in their CET1 ratio: 1.4 ppts for these nine banks.
    - Caused undercapitalization of one more bank with regard to the leverage ratio hurdle rate in 2024.
  - Six banks not accessing central bank facilities would sell part of HTM securities and realize market losses.
    - Estimated losses: €88 million (noting source formatting shows a gap before 88 million).
    - Further decrease in their CET1 ratio by 0.8 ppts for these five banks mentioned in context.
  - System-wide impact equivalent to a decline of 0.5 percent of the aggregate CET1 ratio and of 1.3 percentage points of the CET1 ratio of the banks with funding gaps.
- Charted liquidity-solvency interaction snapshots:
  - Dec-2024 (39 banks) CET1 ratio: 17.6%, 17.5%, 16.7%, 15.4% shown across scenarios and impacts.

### Sensitivity analysis: market risks and other shocks
- Sovereign interest rate sensitivity (domestic sovereign AFS portfolio):
  - Shock: 500 basis points (bp) increase in interest rates.
  - Proxy approach: average maturity of AFS portfolio used due to absence of duration data; Finrep exposures to Luxembourgish general government used as proxy.
  - Result: Luxembourgish banks very little exposed to domestic sovereign bond risks; CET1 ratio in the system would remain flat in isolation.
  - Only one bank has a moderate marked-to-market exposure; no bank would become undercapitalized with regard to the hurdle rate of 8 percent.
  - If entire Held-to-Maturity securities were marked to market under the adverse scenario:
    - Additional market losses: €13.1 billion in 2024.
    - Aggregate CET1 ratio drop: 6.5 percentage points.
    - Recapitalization needs equivalent to 8 percent of GDP.
- Interest rate risk: net interest income effects (maturity ladder gap analysis):
  - Shock: increase in interest rates by 500 bp.
    - Increase in total banking system's net interest income: €5.9 billion.
    - Positive impact on aggregate CET1 capital ratio: 3 percentage points.
    - No undercapitalization in any of the thirty-nine banks.
  - Symmetric shock: decline in interest rates by 500 bps.
    - Aggregate decline in net interest income: €5.9 billion.
    - Decline in aggregate CET1 ratio: 3 percentage points.
    - No undercapitalization of any bank with regard to CET1 hurdle rate of 8 percent.
- Foreign exchange rate risk (net open FX positions as of June 30, 2023):
  - Net long open position in foreign currency for the thirty-nine banks: €4.7 billion, making up 11.5 percent of their CET1 capital.
  - A 30 percent depreciation of the euro nominal effective exchange rate:
    - Gain: €255 million.
    - Moderate impact on aggregate CET1 ratio: 0.1 percentage points.
    - No undercapitalization.
  - A 30 percent appreciation of the euro nominal effective exchange rate:
    - Loss: €255 million.
    - No undercapitalization.
- Decline in stock prices:
  - Marked-to-market equity investments average 0.1 percent of banks' CET1 capital.
  - A 50 percent decline in stock prices:
    - Loss: €430,000.
    - No material impact on aggregate CET1 ratio.
    - No undercapitalization in any of the thirty-nine banks.
- Concentration risk: simultaneous default of large exposures
  - Single largest exposure net of credit mitigation on average reaches 21 percent of CET1 capital.
  - Simultaneous default of the net largest exposure of each of the thirty-nine banks:
    - Aggregate CET1 ratio decline: 4.4 percentage points.
    - Undercapitalization of five banks.
    - Capital shortfall: 1.8 percent of nominal GDP.
  - Simultaneous default of the five net largest exposures:
    - Fourteen banks undercapitalized (CET1 hurdle rate of 8 percent).
    - Capital shortfall: 7.8 percent of nominal GDP.
  - Simultaneous default of the ten net largest exposures:
    - Twenty banks undercapitalized in terms of CET1 capital.
    - Capital shortfall: 14.4 percent of nominal GDP.

### Policy recommendations
- CSSF and BCL should collaborate to improve top-down supervisory stress test models to incorporate liquidity-solvency interactions.
- Supervisory reviews, including for SREP and Pillar II capital requirements, should consider strengthening the banks’ concentration risk treatment in their large exposures to non-intragroup clients.

### Bank liquidity stress tests: methodology and sample
- Three types of liquidity stress tests conducted: LCR, NSFR, and cash-flow based tests.
  - LCR-based stress test: measures capacity to meet short-term liquidity needs in a 30-day horizon against initial level of high-quality liquid assets (HQLA).
  - NSFR: analyzes whether available stable funding covers required stable funding over one year.
  - Cash-flow based stress test: uses maturity profile across time buckets from overnight to more than one year to assess maturity mismatches and counterbalancing capacity.
- Tests used June 2023 data for the sample of 39 banks (same sample as bank solvency stress tests).
  - Sample composition: eight universal, retail and commercial banks; 14 private wealth management banks; nine corporate finance banks; five custodian banks; two covered bond banks; and one clearing, treasury and payment service bank.
  - Liquidity stress tests also consider a sub-sample of 16 banks including 15 banks with parent banks or subsidiaries in Europe identified as weak in the October 2023 Global Financial Stability Report.

*Source: IMF staff calculations and IMF text provided in the source content.*

### Chapter 2 and two banks with deposits of funds from the same group (one of the two overlaps with

### Chapter 2 and two banks with deposits of funds from the same group (one of the two overlaps with the 15 banks with weak parents)

### Scenarios and methodology
- Four scenarios used for both LCR and cash-flow based stress tests: baseline, deposit run, “combined,” and weak parent bank.
- Baseline:
  - Regulatory parameters as applied by EBA for NSFR and LCR calculations.
  - Cash-flow parameters aligned with the ECB 2019 Sensitivity Analysis of Liquidity Risk baseline.
- Deposit run:
  - Higher deposit outflow rates based on empirical data from past deposit runs (October 2023 Global Financial Stability Report Box 2.1 provides more details on size of deposit outflow rates in past bank runs).
- Combined:
  - Combination of high deposit run-off rates and market stress.
  - Applies larger haircuts on asset values (HQLA for LCR, counterbalancing capacity for cash-flow analysis) and lower inflow coefficients from wholesale and retail segments, in addition to outflow parameters from the deposit run scenario.
- Weak parent bank:
  - Applied only to the weak parent bank sub-sample.
  - In addition to combined scenario shocks, uses even higher deposit outflow rates to capture contagion effects and deposit runs triggered by parent companies or other group entities.
- Historical calibration:
  - Deposit run-off rates and paths reflect bank liquidity stress periods in history, notably late 2022–early 2023 (Credit Suisse 2022Q4 and First Republic 2023Q1).
  - Historical events adjusted to match 30-day horizon for LCR analysis (Table 8).
  - Cash-flow analysis assumes outflows over six months with 40 percent of total outflows occurring in the first five days; runoff rates calibrated using similar daily rates as the ECB (2019).

### Cash-flow based stress test parameters (selected)
- Table 6: Outflow rates (Cumulative over Six Months, Percent)
  - Stable retail deposits: Deposit Run 5% ; Combined 5% ; Weak Bank 10%
  - Other retail deposits: Deposit Run 23% ; Combined 23% ; Weak Bank 30%
  - Operational deposits: Deposit Run 26% ; Combined 26% ; Weak Bank 52%
  - Non-operational deposits from non-financial corporates: Deposit Run 40% ; Combined 40% ; Weak Bank 60%
  - Non-operational deposits from other counterparties: Deposit Run 40% ; Combined 40% ; Weak Bank 60%
  - Committed credit facilities: Deposit Run 12% ; Combined 12% ; Weak Bank 15%
  - Liquidity facilities: Deposit Run 60% ; Combined 60% ; Weak Bank 75%
- Haircuts on counterbalancing capacity (Table 7)
  - Level 1 (CQS 1): Deposit Run 2% ; Combined and Weak Bank 15%
  - Level 1 (CQS2, CQS3, CQS4+): Deposit Run 3.5-10% ; Combined and Weak Bank 17-23%
  - Level 2A tradable assets: Deposit Run 3.5-4.5% ; Combined and Weak Bank 17-18%
  - Level 2B ABS, covered bonds: Deposit Run 3% ; Combined and Weak Bank 25-30%
  - Level 2B (CQS 3-5): Deposit Run 13% ; Combined and Weak Bank 50%
  - Central government (CQS1): Deposit Run 14% ; Combined and Weak Bank 29%
  - Central government (CQS 2 & 3): Deposit Run 25.5% ; Combined and Weak Bank 39%
  - Other tradable assets: Deposit Run 25.5% ; Combined and Weak Bank 39%
  - Non tradable assets eligible for central banks: Deposit Run 25.5% ; Combined and Weak Bank 90%
  - Undrawn committed facilities received: Deposit Run 30% ; Combined and Weak Bank 75%
- LCR stress test parameters (Table 8) — selected run-off rates (Percent)
  - Stable retail deposits: Baseline 5% ; Deposit Run 10% ; Combined 10% ; Weak Bank 10%
  - Other retail deposits: Baseline 10% ; Deposit Run 23% ; Combined 23% ; Weak Bank 30%
  - Retail deposits subject to higher outflows: Baseline 15-20% ; Deposit Run 25-40% ; Combined 25-40% ; Weak Bank 25-40%
  - Operational deposits: Baseline 5-25% ; Deposit Run 15-35% ; Combined 15-35% ; Weak Bank 15-45%
  - Excess operational deposits: Baseline 20-40% ; Deposit Run 30-60% ; Combined 30-60% ; Weak Bank 30-80%
  - Non-operational deposits: Baseline 20-40% ; Deposit Run 30-60% ; Combined 30-60% ; Weak Bank 30-80%
  - Committed facilities to retail customers: Baseline 5% ; Deposit Run 10% ; Combined 10% ; Weak Bank 10%
  - Committed facilities to corporate customers: Baseline 10-30% ; Deposit Run 20-40% ; Combined 20-40% ; Weak Bank 20-50%
  - Committed facilities to credit institutions: Baseline 5-40% ; Deposit Run 10-50% ; Combined 10-50% ; Weak Bank 10-60%
  - Change in liquidity assets weights: Level 1 assets Baseline // -5/0% ; Deposit Run -5/0% ; Combined -5/0% ; Weak Bank -5/0%
    - Level 2A assets Baseline // -30/-10% ; Deposit Run -30/-10% ; Combined -30/-10% ; Weak Bank -30/-10%
    - Level 2B assets Baseline // -15% ; Deposit Run -15% ; Combined -15% ; Weak Bank -15%

### Cash-flow approach to liquidity-solvency interaction
- Sequence for counterbalancing negative funding gaps:
  1. Existing cash and central bank reserves
  2. Undrawn committed facilities (excluding facilities from intragroup and other counterparties)
  3. Available-for-sale securities
  4. Held-to-maturity (HtM) securities
- HtM use:
  - Amounts of HtM to sell or pledge computed as a function of approximated shares of HtM in banks’ counterbalancing capacities and additional liquidity needed when counterbalancing capacities are exhausted.
  - Amounts used equal maximum cumulative use of HtM over a one-year period in the cash-flow analysis.

### Results — NSFR and LCR outcomes
- NSFR:
  - All banks have regulatory NSFRs above 100 percent.
  - Weighted average NSFR as of mid-2023: 187 percent.
  - No single bank below the 100 percent threshold.
- LCR:
  - HQLA concentrated in Level 1 assets; Level 1 comprised more than 90 percent of total HQLA for all business types.
  - Central bank reserves account for around 70 percent of total HQLA.
  - Major reliance on wholesale deposits; retail deposits concentrated in traditional banks and private banking.
  - Weighted average LCR in baseline scenario:
    - Full sample: 229 percent
    - Sample of retail and universal banks: 180 percent
  - Number of banks below LCR threshold (<100%):
    - Baseline: 0
    - Deposit Run: 5
    - Combined: 6
    - Weak Bank: 3
  - Share of total sample assets below LCR threshold:
    - Baseline: 0
    - Deposit Run: 21.7
    - Combined: 23.8
    - Weak Bank: 17.6
  - Among six banks identified in the combined scenario, four are SIs (subsidiaries) and many are subsidiaries of global banks that upstream excess liquidity to the group.
- Currency mismatches:
  - LCRs in USD and GBP below 100 percent for all business models.
  - Net outflows in CAD, CHF, DKK, GBP, PLN, SEK, USD, and YEN; 11 banks with GBP liabilities fall short of GBP HQLA in the stress test sample.
  - In case of large outflows in non-EUR currencies, banks would need to monetize EUR HQLA and/or use FX swaps.

### Results — sensitivity and reverse stress testing
- Retail deposit run sensitivity:
  - Banks resilient up to 20 percent retail deposit outflows: all banks maintain LCR above 100 percent at runoff rates below 20 percent.
  - At 21 percent retail deposit run-off, one bank’s LCR falls below 100 percent.
  - At 40 percent run-off, three banks (accounting for 14 percent of total assets in the sample) have LCRs below 100 percent.
- Reverse stress testing:
  - Increments of 5 percent from 0 percent to 40 percent retail deposit outflow rates were tested; LCR computed for each bank at each increment.

### Results — cash-flow analysis and net funding gaps
- Counterbalancing capacity quality:
  - High quality, mainly central bank reserves across business models, followed by Level 1 liquid assets.
- Use of counterbalancing capacity under stress:
  - Many banks (15 out of 39 in the “Combined” scenario) would use up cash and central bank reserves and available-for-sale securities, necessitating sale, repo, or pledge of HtM assets.
- Number of banks with net funding gap (eop CBC<0):
  - Baseline: 1
  - Deposit Run: 3
  - Combined: 6
  - Weak Bank/Same-name Funds: 5
- Share of total sample assets with net funding gap:
  - Baseline: 14.2
  - Deposit Run: 23.8
  - Combined: 19.0
- Time buckets:
  - Liquidity gap widest for 3–6 month and 6–12 month buckets, which are not directly monitored by LCR (30-day) or NSFR (one-year).
  - Five out of six banks identified with potential net funding gaps are subsidiaries of significant banking groups from an SSM perspective.

### Findings on bank business models and vulnerabilities
- Business model funding composition:
  - High reliance on deposits and other liabilities from financial institutions across all business models.
  - For universal, retail and commercial, and private banking, corporate deposits and retail deposits are also important.
- Subsidiaries of global banks:
  - Some subsidiaries operate with relatively lower liquidity ratios due to centralized liquidity management and upstreaming of excess liquidity to the group.
- Currency liquidity shortfalls:
  - Limited or no liquidity buffers in many non-EUR currencies, notably USD and GBP.

### Investment funds liquidity stress test — objective, scope, and approach
- Objective and scope:
  - Evaluates liquidity risk within Luxembourg investment funds to determine resilience to severe but plausible redemption shocks.
  - Time horizon: one month.
  - Identifies fund categories more susceptible to liquidity risk and gauges potential for system-wide shock propagation.
  - Does not account for liquidity management tools (LMTs) nor for redemption frequency of AIFs.
  - Test is primarily microprudential but incorporates macroprudential second-round effects from asset sales.
- Five key components of the test:
  1. Estimating initial impact on balance sheet through Net Asset Value (NAV).
  2. Estimating resulting redemption shock.
  3. Determining composition of asset sales to meet redemptions.
  4. Estimating price impact of those sales.
  5. Estimating second-round impact on NAVs due to price impact of securities sales.
- Interconnectedness and spillovers:
  - Direct spillovers assessed via fund holdings in other domestic sectors, mostly bank deposits—fund redemptions lead funds to withdraw deposits from Luxembourg banks, creating liquidity pressure on banks.
  - Indirect spillovers assessed via asset sales’ impact on market prices; given fund sector size, indirect spillovers primarily run from funds to other sectors.
  - Second-round impacts imposed on banks’ and insurers’ securities holdings are estimated to be negligible.

*Source: IMF staff calculations and assessment as presented in the Chapter.*

### 105. The stress test includes nearly 70 percent of the universe of UCITS falling withing the

### 105. The stress test includes nearly 70 percent of the universe of UCITS falling withing the

### Sample coverage and composition
- The stress test sample comprises 1,082 funds with aggregate total net assets of 2,675.5 billion euros.
- The sample includes all open-ended investment funds with NAV greater than 1 billion euros:
  - 895 UCITS
  - 187 AIFs
- UCITS account for 74 percent of the total net assets in the sample.
- MMFs are included in the initial stage to assess impact on NAV and consequent need to sell assets, although they are not featured in the liquidity stress results.
- Seven funds were dropped from the analysis due to data consistency issues.

### Methodology: redemption shocks and scenarios
- Scenarios used:
  - Adverse macroeconomic scenario (main integrated stress test).
  - Scenario mirroring the Global Financial Crisis (GFC), using September–October 2008 asset price changes.
  - Historical redemption scenarios based on past fund flows.
- Redemption shock calibration:
  - Adverse scenario redemption shocks constructed using historical flow-performance relationships at the fund strategy level to translate NAV shocks into outflows.
  - Historical redemption shocks use two approaches:
    - Homogeneity approach: shocks determined from net flow distributions of all funds in a category, applied uniformly within the category (varies by broad investment strategy).
    - Heterogeneity approach: fund-specific shocks from each fund's historical net flow distribution.
  - For both historical approaches, shocks are derived from the first and fifth percentiles of the historical net outflow distribution, expressed as a percentage of previous total net assets using a VaR assumption.
- If historical flow-performance relationship for a fund type is not statistically different from zero, it is assumed zero outflows following the adverse and GFC scenarios; resilience assessed via historical redemption scenarios.

### Asset grouping, price shocks, and balance sheet data
- Balance sheet assets (data from March 2023) are organized into 42 groups, experiencing asset-class-specific price shocks.
- Asset classes include Equities, Sovereign Bonds, Corporate Bonds, Deposits, MMF investments, and others, and are categorized by key regions (Luxembourg to Rest of the World).
- Initial price shocks applied to tradable assets via a financial simulator model; liquid short-term deposits assumed to retain value.
- Market-sensitive assets aligned with market price indices and adjusted for asset price scenario consistency with historical movements.
- Real estate price shocks derived from a House Price at risk model for CRE to forecast commercial real estate price changes in key economies.

### Liquidity buffers (HQLA) and liquidation strategies
- Liquidity buffers measured via a liquidity buckets approach using Basel III HQLA criteria to assign liquidity weights to portfolio assets.
- HQLA distribution (10th, 50th, 90th percentiles) shows:
  - Even the 10th percentile of the overall sample exhibits HQLA of roughly 40 percent of net assets.
- Strategies with considerably lower HQLA buffers include:
  - High-yield bond funds
  - Real estate funds (which are AIFs)
  - Other funds (with a relatively high share of AIFs)
- Assumed liquidation strategies when meeting redemptions:
  - (i) Slicing (pro rata) — selling HQLA assets in line with portfolio HQLA weights.
  - (ii) Waterfall: using highly liquid securities before cash.
  - (iii) Waterfall: using cash first, before moving to highly liquid assets.
- Main analytical focus: waterfall selling securities first (maximizes price impact for second-round effects). Cash-first waterfall would have greater implications for banks’ liquidity risk.

### Asset exposures and market impact assumptions
- Investment funds’ holdings relative to outstanding securities:
  - Own 2.5 percent of Luxembourg's sovereign debt.
  - Own nearly 15 percent of Luxembourg equities.
  - Hold 1.5–2.5 percent of sovereign debt and 3–5 percent of equities in France, Germany, and the United Kingdom.
  - In emerging markets, funds in Luxembourg hold more than 2 percent of total sovereign debt for a number of countries.
- Price impact estimation:
  - Volume of sales measured against market depth with estimates from previous FSAPs, academic literature, and specific methodologies by security type.
  - Estimates consider volatility and asset sale period; primary figures based on market depth during stress, assuming a 7-day sale period.
  - For funds, asset price impact triggers NAV recalculation, leading to potential second-wave redemptions via the flow-performance relationship, and further NAV adjustments for asset liquidation costs.

### Resilience metrics
- Redemption Coverage Ratio (RCR):
  - RCR = HQLA / 30-day redemption shock.
  - RCR > 1: sufficient liquidity to manage redemptions.
  - RCR < 1: need to sell less liquid assets, possibly at significantly reduced prices.
- Liquidity shortfall:
  - Measures difference between redemption shock and fund’s level of HQLA for funds with RCR < 1.

### Results — initial impact on balance sheets and NAV
- Under the adverse scenario financial shocks:
  - Aggregate valuation losses of 13.1 percent of the aggregate NAV.
  - Aggregate decline in NAV from the stress test (13.1 percent) broadly aligns with the Bayesian VAR model decline of 14.1 percent used in the Macro scenario.
- Heterogeneity by strategy:
  - Equity funds: average NAV decline of over 17 percent.
  - Funds in the 95th percentile of the distribution: average decrease in NAV of 21.8 percent.
  - Aggregate NAV of funds beyond the 95th percentile equals 142.8 billion euros, roughly 5 percent of the total NAV of the sample.
- Under the GFC scenario:
  - Average NAV falls by 19.8 percent.
  - Equity funds are the most adversely impacted in the GFC scenario.

*Source: BCL, CSSF and IMF staff calculations.*

### 27.9 percent, while bond funds fall by 10.6 percent. Funds in the 95

### 1luxea2024008 - 27.9 percent, while bond funds fall by 10.6 percent. Funds in the 95

### Estimated Initial and Final Valuation Changes in NAV (Adverse and GFC scenarios)
- Aggregate initial average fall in NAV (Adverse scenario): 13.1 percent; median -14.1; StdDev 5.7; P95 -21.8; P99 -27.5; Total NAV of funds above percentile... 142.8 (Billions of euros); 35.9.
- By strategy (Initial, Adverse scenario, average / median / StdDev / P95 / P99 / Total NAV):
  - Bond funds: -8.0 / -7.4 / 2.6 / -14.1 / -18.1 / 26.7 / 6.0
  - Equity funds: -17.4 / -17.1 / 4.2 / -23.2 / -28.0 / 64.6 / 11.8
  - Fund of funds: -16.0 / -16.5 / 3.8 / -22.0 / -22.9 / 11.5 / 1.0
  - Hedge funds: -12.1 / -10.4 / 8.1 / -29.4 / -34.7 / 3.7 / 2.5
  - Mixed funds: -12.6 / -12.8 / 4.0 / -17.5 / -22.0 / 21.5 / 3.1
  - Other funds: -14.4 / -15.7 / 5.7 / -22.4 / -26.5 / 3.1 / 1.7
  - Real estate funds: -14.4 / -13.3 / 4.9 / -22.2 / -24.1 / 9.2 / 8.1
- Aggregate initial fall in NAV (GFC scenario): average -19.8 percent; median -19.6; StdDev 10.4; P95 -39.5; P99 -41.3; Total NAV of funds above percentile... 133.9 (Billions of euros); 28.4.
- By strategy (Initial, GFC scenario, average / median / StdDev / P95 / P99 / Total NAV):
  - Bond funds: -10.5 / -10.6 / 4.8 / -18.8 / -23.2 / 27.2 / 6.1
  - Equity funds: -27.9 / -26.0 / 6.8 / -40.4 / -40.8 / 50.9 / 8.8
  - Fund of funds: -22.9 / -20.8 / 8.4 / -39.0 / -53.1 / 13.2 / 1.8
  - Hedge funds: -15.9 / -14.8 / 9.8 / -36.3 / -39.3 / 3.7 / 2.5
  - Mixed funds: -18.0 / -17.8 / 7.2 / -30.3 / -39.7 / 18.5 / 3.1
  - MMF: -6.8 / -7.0 / 3.5 / -12.0 / -13.2 / 14.4 / 6.1
  - Other funds: -23.6 / -24.0 / 12.4 / -41.8 / -52.5 / 2.5 / 1.3
  - Real estate funds: -23.1 / -19.1 / 15.6 / -54.5 / -59.3 / 9.2 / 8.1
- Final valuation change in NAV after second-round price effects (Adverse scenario, aggregate): average -13.8 percent; median -14.4; StdDev 5.5; P95 -22.5; P99 -27.9; Total NAV ... 140.7 / 35.9.
- By strategy (Final, Adverse scenario):
  - Bond funds: -9.1 / -8.5 / 2.7 / -14.4 / -19.1 / 35.1 / 6.0
  - Equity funds: -17.8 / -17.4 / 4.2 / -24.1 / -28.3 / 64.6 / 11.8
  - Fund of funds: -16.4 / -16.7 / 3.9 / -22.9 / -23.2 / 9.6 / 1.0
  - Hedge funds: -12.8 / -11.6 / 8.4 / -31.3 / -35.7 / 3.7 / 2.5
  - Mixed funds: -13.3 / -13.3 / 3.8 / -17.9 / -22.8 / 21.5 / 3.1
  - Other funds: -14.7 / -15.7 / 5.6 / -22.4 / -26.8 / 3.1 / 1.7
  - Real estate funds: -14.5 / -13.4 / 4.9 / -22.3 / -24.1 / 9.2 / 8.1

### Net Outflows, Redemption Shocks, and Liquidity Management
- Redemption shocks estimated from scenarios and historical redemption data provide alternative stress-test starting points; histograms of net outflows are shown by investment strategy.
- Net outflows estimated via flow-performance relationship are, on average, more severe for equity funds than for bond and mixed funds.
- Adverse and GFC scenario outflows are considerably milder than historical redemption data, reflecting a relatively low elasticity of NAV to net outflows.
- The relatively low elasticity contributes to sector stability in managing liquidity risk.
- Note: Distribution charts report flows as a percentage of NAV of the funds in the sample.

### Impact of Asset Sales on Market Prices and Geographic/Asset Composition
- In the adverse scenario, funds may sell up to 150 billion euros in liquid securities to meet redemptions, assuming liquidation of liquid securities first.
  - Composition: nearly 50 billion euros from sovereign bonds, 20 billion from corporate bonds, and 70 billion from equities.
  - Geographic breakdown: US securities 44 billion euros; France 16 billion euros; Germany 13 billion euros.
- In the GFC scenario, total liquidated increases to 200 billion euros; composition and country pattern largely consistent but more extreme price swings.
- Asset sales from banking and insurance sectors considered too small relative to investment funds and were excluded from price-impact estimation.
- Price impacts (Adverse scenario) for bond prices (corporate and sovereign) in Germany, France, and other Eurozone countries range from 0.2 to 1.2 percent (approximate yield increase of 5 to 30 basis points).
- US and UK markets show more subdued impacts (except US corporate bonds).
- Under the GFC scenario, price declines can reach almost 4.5 percentage points for some securities.

### Second-round Effects on NAV and Net Outflows
- Incorporating second-round price effects increases average fall in NAV in the Adverse scenario from 13.1 to 13.8 percent.
- For bond funds, NAV fall increases from 8.0 percent (first round) to 9.1 percent (second round).
- The increase in NAV losses from second-round effects is not large enough to materially change estimated net outflows versus initial impacts.
- For the GFC scenario, average NAV fall increases from 19.8 to 20.8 percent with second-round effects; still insufficient to meaningfully alter net outflows relative to the initial impact.

### Final Stress Test Results and Vulnerabilities
- Overall conclusion: Investment funds are resilient to redemptions following the main stress scenarios.
- Only four AIFs exhibit a Redemption Coverage Ratio (RCR) below one in the adverse scenario after the final redemption shock including second-round effects; these four are loan-fund strategy AIFs with longer redemption frequencies (typically monthly) and liquidity management tools (LMTs) like gating and redemption deferral.
- Final stress-test results are identical to results without second-round effects, indicating second-round channel unlikely to pose relevant additional risks to the sector; results for the GFC scenario are also identical despite higher outflows and second-round effects.
- Historical redemption-based vulnerabilities:
  - A considerable share of high-yield bond funds—making up 44 percent of total NAV of high-yield bond funds—would have problems meeting large but plausible historical redemptions.
  - Six real estate AIFs have RCR below one; these real estate AIFs have redemption frequencies quarterly or longer and possess liquidity management tools (mainly redemption deferral).
- Table 13 results summary (Adverse and GFC; Historical homogeneity first and fifth percentiles): selected entries:
  - High-yield: Historical (Homogeneity, First Percentile) shows Funds with RCR<1 = 17; % NAV with RCR<1 = 58.6; % Funds with RCR<1 = 44.1 (others zero in adverse/GFC columns).
  - Mixed Funds: Adverse and GFC: Funds with RCR<1 = 4; % Funds with RCR<1 = 2.5; % NAV with RCR<1 = 0.3.

### Sensitivity and Reverse Stress Testing for Investment Funds
- Reverse stress tests apply uniform exogenous redemption shocks (percentage of NAV) across funds and compare net outflows to HQLA holdings.
- Even with redemption shocks nearing 40 percent, over 90 percent of funds maintain RCR above one.
- Cumulative liquidity shortfall as a fraction of total net assets remains minimal up to 40 percent shock.
- Beyond 40 percent shocks, marked increase in number of funds with RCR<1; redemptions exceeding 40 percent are uncommon, making such widespread outflows unlikely.

### Policy Recommendations for Investment Funds
- Continue monitoring liquidity mismatches in the AIF space, especially in newly emerging investment strategies.
- UCITS generally prepared for liquidity crises; minority of UCITS (mainly high-yield) might struggle with very high redemptions but represent a small share of the sector.
- Enhance monitoring of AIFs and their interlinkages to the OFI sector and other investment funds; develop methods better tailored to assess AIF liquidity risk.

### Money Market Funds (MMFs) — Objective, Methodology, and Results
- MMFs undergo separate solvency analysis due to CNAV and LVNAV regulatory features and amortized cost accounting.
- MMF stress test assesses likelihood of CNAV and LVNAV MMFs breaking the 20 bp threshold using instant interest rate and spread shocks consistent with the adverse macrofinancial scenario.
- MMF sample statistics (data as of 2023 Q1):
  - CNAV: Number of Funds 10; Total Net Assets 72.0 (billions of euros); Average WAM 7.2 (days); Median 1.4; Maximum 51.0; 25.8.
  - LVNAV: Number of Funds 15; Total Net Assets 220.9; Average WAM 14.7; Median 6.1; Maximum 101.1; 23.2.
  - VNAV: Number of Funds 79; Total Net Assets 136.7; Average WAM 1.7; Median 0.8; Maximum 18.5; 57.9.
- Adverse scenario MMF results:
  - Median fall in NAV for LVNAVs: 14 basis points.
  - Four LVNAVs would ‘break the buck’ (exceed 20-bp threshold) under the adverse scenario; two USD-based and two GBP-based.
  - Two small CNAV MMFs experienced NAV declines larger than 20 basis points (regulatory CNAV threshold is 50 basis points); these two funds total under 2 billion euros in net assets.
- Stress test shocks: interest rate shock in range 130 to 150 basis points for main currencies; spread shock homogeneous 200 basis points to corporate spreads; no spread applied to public sector MM instruments.
- Reverse stress-test findings for MMFs:
  - Majority of LVNAVs and CNAVs resilient to cumulative shocks up to approximately 200 basis points.
  - Majority (by percentage of total NAV) can withstand cumulative increases of 300 bps.
  - Almost all LVNAVs cross the 20-bp threshold for combined shocks of 400 basis points.

### MMF Recommendations
- Consider market guidance to enhance resilience of outlier LVNAVs to mitigate tail-risk spillovers.
- Engage with market participants employing riskier strategies to encourage alignment with more conservative LVNAV practices.

### Insurance Solvency Stress Test — Scope, Scenario, and Key Results
- Sample: 22 large insurers (10 life insurers covering ~83 percent of assets in sector; 12 non-life insurers and reinsurers with market coverage 81 percent of gross written premiums).
- Aggregated balance sheet assets of participants: EUR 239bn; EUR 187bn (78 percent) attributable to life insurers.
- Of total technical provisions in life sector, 83 percent stem from unit-linked business.
- Pre-stress: all 22 participants have solvency ratios above regulatory threshold 100 percent.
- Capital quality: 88 percent of eligible own funds are unrestricted Tier 1; 1 percent Tier 3.
- Long-Term Guarantee (LTG) measures: 11 insurers use the Volatility Adjustment (VA); for the median life insurer using VA, SCR ratio at end-2022 would have been 2 percentage points lower without VA.
- Adverse scenario tailored for insurers includes shocks to RFR, equity and property prices, credit spreads by credit quality and geography, and a one-category downgrade for one third of corporate bonds in SCR recalculation.
- Scenario shock specifics (selected):
  - Equity shocks: Luxembourg -22.3%; Euro Area -23.8%; United States -13.1%; Other advanced economies -19.5%; China -16.8%; Emerging economies -16.8%; Unlisted -12.0%; Infrastructure -4.0%.
  - Property: Real estate-13.5%; RRE domestic -15.0%; CRE domestic -18.0%; RRE other countries -10.0%; CRE other countries -13.0%.
  - Corporate bonds (financials) by Credit Quality Step: 0 +50 bps; 1 +70; 2 +90; 3 +120; 4 +180; 5 +300; 6 +420; Unrated +120.
  - Sovereign bonds (in bps): Luxembourg +60; Euro Area +75; United States +60; Other advanced economies +70; China +100; Emerging economies +110; Supranational +/-0.
  - Interest rates: EUR 1Y +50; EUR 10Y +5; USD 1Y +20; USD 10Y +36.
  - Loans and mortgages: Mortgages +45; Other collateralized loans +60; Uncollateralized loans +100.
- Capital and modeling assumptions:
  - Solvency II market-consistent valuation used; LTG measures (VA) and transitional measures incorporated where applicable.
  - Stress applied to assets and liabilities via haircuts and stressed term structures; unit-linked liabilities decline mirrored market value loss of underlying assets.
  - SCR partially recalculated after stress for market risk and life underwriting risk; other SCR components assumed unchanged except approximations for internal model users.
  - Loss-absorbing capacity of deferred taxes recalculated based on modeled valuation losses.
- Results:
  - Life insurers: assets decline by 9.1 percent on average; liabilities decline by 8.8 percent on average; excess of assets over liabilities (EoAoL) declines by 28 percent (from EUR 3.4bn to EUR 2.5bn). Median life insurer assets-to-liabilities ratio after stress: 101.6 percent (down 0.3 percentage points).
  - For guaranteed life business, largest decline in capital stems from higher corporate and sovereign spreads; for unit-linked business, significant effect is reduction in future fee income as assets under management decline.
  - Non-life insurers and reinsurers: assets decline by only 2.8 percent and liabilities by a smaller amount (sector exhibits lower sensitivity and high resilience in adverse scenario).

### Insurance Policy and Interpretation Notes
- Stress tests are macrofinancial diagnostics, not pass-fail exercises nor implying additional regulatory capital requirements for individual insurers.
- TD stress test limitations: static balance sheet assumption excludes reactive management actions (asset reallocation, underwriting changes, reinsurance adjustments, group capital support) and some data granularity limitations (derivatives, hedges, product features).
- Sensitivity tests included single-factor shocks: interest rates ±200 bps; stock prices -50 percent (unlisted -30 percent); Luxembourg sovereign spread +500 bps; currency ±20 percent; counterparty defaults with specified haircuts and LGDs.

*Source: BCL, CSSF and IMF Staff calculations (excerpts from IMF FSAP Luxembourg chapter).*

### 1.3 percent. The ratio of assets to liabilities declines from 124.2 to 122.6 percent, but both before

### 1.3 percent. The ratio of assets to liabilities declines from 124.2 to 122.6 percent, but both before

### Valuation Impact
- The ratio of assets to liabilities declines from 124.2 to 122.6 percent; dispersion across the sample is very large.
- The excess of assets over liabilities declines by 9 percent, driven mainly by higher corporate spreads.
- Life sector valuation changes:
  - Assets in the life sector decline by 9.1 percent.
  - Liabilities in the life sector decline by 8.8 percent (through higher interest rates).
  - The ratio of assets to liabilities in the life sector shrinks by 0.3 percentage points to 101.6 percent for the median life firm.
  - The excess of assets over liabilities in the life sector declines from EUR 3.43bn to EUR 2.46bn (-28 percent), mainly driven by higher sovereign and corporate spreads and the impact of lower assets under management on future fee income in unit-linked business.
- Non-life sector valuation changes:
  - Non-life insurers see relatively small valuation changes in their balance sheet.
  - The ratio of assets to liabilities for the median non-life insurer declines by 1.6 percentage points to 122.6 percent.
  - The excess of assets over liabilities in the non-life sector declines by 9 percent (from EUR 9.95bn to EUR 9.05bn), with the largest impact coming from higher corporate bond spreads.
- Notes: “Other effects” comprise mainly the valuation impact on reinsurance recoverables for guaranteed life insurance.
- Source: IMF staff calculations based on CAA data.

### Solvency Impact
- Life sector solvency effects:
  - Eligible own funds (EOF) decline by 22 percent for insurers in the sample.
  - The median life insurer records a coverage of its post-stress SCR of 127 percent, down from 150 percent prior to the stress.
  - No life insurer falls below the regulatory threshold of 100 percent, although a few approach it.
- Non-life sector solvency effects:
  - Eligible own funds decline by 8 percent.
  - The post-stress SCR remains almost unchanged.
  - The median SCR ratio after stress amounts to 179 percent, down by 17 percentage points.
- Stage 2 (price impact of investment fund redemptions):
  - Assets of life insurers decline by less than 0.5 percent.
  - The median SCR ratio for life insurers declines by another 1 percentage point to 126 percent.
  - In the non-life sector, the median SCR coverage ends up at 175 percent, 4 percentage points lower than after the first-round impact of the adverse scenario.
- Visual summary points (as presented):
  - The median life insurer’s SCR ratio declines from 150 to 127 percent after stress.
  - Eligible own funds (EOF) would decline by 21 percent for the median life insurer and by 7 percent in the non-life sector, while the SCR would only decline marginally (and remains almost unchanged for non-life insurers).
- Source: IMF staff calculations based on CAA data.

### Sensitivity Analyses
- General finding: Broad resilience of the insurance sector, with large effects from changes in interest rates and stock prices on life insurers.
- Life sector sensitivities:
  - A return to (ultra-)low rates would deteriorate solvency positions for the median life insurer by 12 percentage points.
  - A 50 percent decline in stock prices would reduce the SCR ratio by 30 percentage points, mainly through the impact on future fee income in unit-linked business.
  - A default of the largest, and even of the three largest banking counterparties would have a limited first-round impact on solvency.
  - A steep increase in the credit spread for Luxembourgish sovereign bonds would leave solvency rates almost unchanged given very low exposures.
- Non-life sector sensitivities:
  - Impact of single-factor shocks is lower than in the life sector.
  - Only the 200 basis points decline in the EUR risk-free rate would move the median SCR ratio by more than 10 percentage points.
  - Lower stock prices and banking counterparty defaults have a very minor effect.
  - Exchange rate appreciation/depreciation effects are mixed due to some insurers preparing accounts in U.S. dollars.
- Specific scenario labels preserved from figures:
  - pre-stress; RFR parallel +200bps; RFR parallel -200bps; Equity -50%; EUR +20%; EUR -20%; LUX sov. spread +500bps; Default largest bank; Default 3 largest banks.
- Source: IMF staff calculations based on CAA data.

### Policy Recommendations
- Improve supervisory reporting data quality, especially on investments (S.06.02) and derivative holdings (S.08.01).
  - Conduct horizontal reviews and automatically cross-check reported data against other data sources (e.g., the database on Legal Entity Identifiers).
  - Remind insurers of best practices in reporting based on evaluation findings.
- Implement regular top-down stress tests and sensitivity analysis for the insurance sector.
  - First step: cover insurers that use the standard formula for calculating the solvency capital requirement.
  - Subsequently roll out stress testing to (full) internal model users, building on internal model supervision and insurers’ Own Risk and Solvency Assessments (ORSA).
  - As an intermediate step, approximate sensitivity analyses based on the assumption of standard formula use, leveraging new supervisory reporting under Solvency II available from 2024 onwards.

### Insurance Liquidity Analysis — Approach and Scope
- Elevated lapse rates in life insurance in 2022-23, particularly in guaranteed business:
  - Monthly lapse rates typically hovered between 0.5 and 1 percent and peaked at 3.5 percent in late 2022 and early 2023.
  - Since Q3 2023, lapse rates have started to normalize.
  - Unit-linked business lapse rates more stable since 2020; only during a few months exceeded 1 percent.
  - Unit-linked redemptions are not significantly correlated to gross redemptions in Euro area investment funds.
- Liquidity risk mitigants:
  - Typical payout to policyholders about one month after notification of intention to lapse.
  - Dedicated funds (about 80 percent of total market) allow insurers to redeem underlying assets in kind to policyholders.
  - Some large life insurers have fully reinsured guaranteed life business including lapse risk; the CAA requires full collateralization of such reinsurance arrangements.
- Data notes:
  - FSAP tested eight large life insurers in a scenario where monthly lapse rates in guaranteed business increase by an additional 50 percent compared to the highest level observed—individually for each insurer—since 2020.
  - Due to limited Solvency II QRT lapse data, detailed lapse data were directly collected from life insurers.
  - Redemption coverage ratio (RCR) relates outflow to available total liquidity sources; haircuts applied to HQLA in line with Basel Committee (2013) Liquidity Coverage Ratio.

### Insurance Liquidity Analysis — Results
- Aggregate simulated outflow and liquidity:
  - Simulated outflow for the full sample amounts to EUR 2.4bn, about 7.6 percent of the statutory reserves in guaranteed business as of June 2023.
  - Total sources of liquidity amount to EUR 32.5bn, comprising:
    - Cash holdings of EUR 1.2bn.
    - High-quality liquid assets (after haircuts) of EUR 9.3bn.
    - Reinsurance recoverables of EUR 22.0bn.
  - Aggregated redemption coverage ratio (RCR) amounts to 13.6.
- Distributional findings:
  - Even for the least liquid life insurer, the RCR amounts to 9.6.
  - Half of the sample firms rely on reinsurance arrangements (each with a different reinsurer); the others hold more cash and/or HQLA.
- Derivatives and margin-call risk:
  - Total notional of derivatives held by Luxembourgish insurers as of September 2023 amounts to EUR 3.6bn (market value EUR 88m).
  - Of that: EUR 1.4bn are interest rate swaps and EUR 1.3bn are foreign-exchange forward rate agreements.
  - Use of derivatives is limited to a very small number of insurers; margin call risk is very limited.
- Source: IMF staff calculations based on company submissions.

### Risk Assessment Matrix (high-level excerpts)
- A sharp and prolonged rise in inflation expectations:
  - Relative likelihood: Medium.
  - Impact if realized: High/Medium.
  - Channels: Higher inflation through wages/indexation, increased fiscal cost, tighter financial conditions, higher credit risk, severe effects on banks and non-banks.
- Commodity price volatility:
  - Relative likelihood: Medium.
- Abrupt global slowdown or recession:
  - Relative likelihood: Medium.
  - Impact if realized: Medium.
  - Europe-specific channels include energy crisis, supply disruptions, higher market interest rates, with potential weakening of Luxembourg export demand.
- Systemic financial instability:
  - Relative likelihood: Medium: High.
  - Impact if realized: High/Medium.
  - Channels: Sharp swings in real interest rates, risk premia, asset repricing, insolvencies, investment fund redemptions and liquidity pressures.
- Cyber-attacks:
  - Relative likelihood: Medium.
  - Time horizon: ST/MT.
  - Impact if realized: High.
- Sharp correction in house prices:
  - Relative likelihood: Low.
  - Time horizon: ST/MT.
  - Impact if realized: Low.
- Possible changes in international corporate and personal taxation:
  - Relative likelihood: Medium.
  - Time horizon: ST/MT.
  - Impact if realized: Medium.

### Stress Test Matrix (STeM) — Banking Sector excerpts relevant to approach
- Banking Sector Solvency Stress Test (Top-down by IMF):
  - Institutional perimeter:
    - Institutions included: 39 banks; 19 banks subcategorized as SIs.
    - One bank domestically owned; 24 subsidiaries of euro area banks; 14 subsidiaries of non-euro area banks.
    - Market share covered: 90 percent of banking sector’s assets; 100 percent of residential mortgage loans; at least 80 percent of different business models.
  - Data and baseline date:
    - Supervisory data including FINREP and COREP, IRRBB, liquidity risk and market risk sensitivities; market and public data.
    - Baseline date: October 2023.
  - Channels of risk propagation and methodology:
    - FSAP team satellite models and methodologies; balance-sheet regulatory approach; market data-based approaches.
    - Losses for securities portfolios based on duration approach.
    - Provisioning for IRB and STA modeled using IFRS9 transition matrix approach.
  - Stress test horizon: 3-years (2024-2026).
  - Scenario analysis:
    - Two scenarios: a baseline (October 2023 WEO projections) and an adverse scenario capturing key RAM risks.
    - Luxembourg-specific Bayesian VAR estimated for domestic variables to complement GFM projections.
  - Sensitivity analysis:
    - Evaluated interest rate shocks, exchange rate shocks, decline in sovereign bond prices, and failure of largest to 10 largest corporate exposures.
  - Risks and buffers:
    - Risks covered: credit, market, and interest rate risk (IRRBB) on the banking book; solvency-liquidity interactions through funding costs.
    - Behavioral adjustments: quasi-static balance sheet growth in line with nominal GDP with 0 percent floor; no accrual of interest income from nonperforming loans; dividends paid by banks remaining adequately capitalized.
- Source: IMF staff materials included in the chapter.

*Source: IMF staff calculations and analysis based on CAA data, company submissions, ECB data, supervisory data, and IMF FSAP procedures as presented in the chapter.*

### 5. Regulatory

### 5. Regulatory

### Regulatory and Market-Based Standards and Parameters (Credit Risk and Liquidity)
- Calibration of risk parameters:
  - Through the cycle and Point-in-time for credit risk parameters or proxies.
- Regulatory/accounting and market-based standards:
  - National regulatory framework of Basel III regulatory minima: CET1 ratio of 4.5 percent and leverage ratio of 3 percent.
  - The hurdle rate for the CET1 ratio is equal to the Overall Capital Requirement, including the minimum requirement, the Pillar II requirements, the Capital Conservation Buffer, the Countercyclical Buffer as well as the O-SII Buffer.
- Liquidity regulatory standards referenced:
  - Basel III standards (revision as of January 2013).
  - European Commission Delegated Act.

### Reporting Format for Results (General)
- Output presentation items across tests:
  - System-wide capital shortfall.
  - Number of banks and percentage of banking assets in the system that fall below regulatory minima.
  - Outputs include information on impact of different result drivers, including profit components.
  - Liquidity outputs: liquidity gap by bank (and aggregated); survival period in days by bank; number of banks that can still meet their obligations.
  - Investment funds outputs: impact on value of assets and liabilities for all funds in the sample; for MMFs, deviations between constant NAV and shadow NAV and number/share crossing the 20-basis points threshold.
  - Liquidity test outputs for funds: number of funds with a redemption coverage ratio below one; total net assets of funds with RCR below one as a percentage of aggregate total net assets; liquidity shortfall amount for individual funds after redemptions.
  - Insurance outputs: impact on value of assets and liabilities; impact on solvency ratio (SCR coverage); aggregated capital shortfall; dispersion across companies; contribution of individual shocks.

---

### Banking Sector: Liquidity Risk (Top-Down by IMF)

### Institutional perimeter and data
- Institutions included: 39 banks.
- Market share: 90 percent of banking sector’s assets.
- Latest data: June 2023 for LCR, NSFR and cash flow analysis.
- Source: supervisory data (COREP, ST exercise).
- Scope of consolidation: perimeter of individual banks.

### Methodology and channels of risk propagation
- Methodology:
  - Basel III-LCR and NSFR type proxies.
  - Cash-flow based liquidity stress test using maturity buckets by banks.
  - Liquidity test in foreign currencies.
- Risks:
  - Funding liquidity (liquidity outflows, instantaneous shocks).
  - Market liquidity (price shocks, instantaneous shocks).
- Buffers:
  - Counterbalancing capacity.
  - Central bank facilities.

### Tail shocks and stress calibration
- Size of the shock:
  - Run-off rates calculated following historical events, or IMF expert judgment and LCR/NSFR rates.
  - Bank run and dry up of wholesale funding markets, taking into account haircuts to liquid assets.

---

### Investment Funds Sector: Change in NAVs / Solvency Risk (Top-Down by IMF)

### Institutional perimeter and data
- Institutions included:
  - 1,135 largest open-ended funds.
  - Largest UCITS Investment Funds covering EM bond funds, HY bond funds, Mixed funds, Equity funds and MMFs with total net assets over 1 billion euros, totaling close to 2.7 trillion euros in total net assets.
  - Largest open-ended AIFs with total net assets over 1 billion euros subject to same reporting standards as UCITS, totaling close to 400 million euros in total net assets.
- Market share:
  - 80 percent of total net assets of UCITS within full reporting scope.
  - Roughly 50 percent of total net assets of open-ended AIFs subject to same reporting standards as UCITS in full reporting scope.
- Data and baseline date:
  - Source: Supervisory data.
  - Baseline data: March 2023.

### Methodology and scenarios
- Methodology:
  - IMF’s Financial shock simulator.
  - Market data-based approaches.
  - For MMFs: increase in risk free rates and credit spreads that would result in deviations between shadow NAV and Constant NAV.
- Scenarios:
  - Adverse scenario (same as for other sectors) capturing key risks in the IMF’s RAM. External assumptions given by GFM.
  - GFC scenario based on changes in asset prices during September and October 2008.
- Sensitivity analysis:
  - Reverse stress test for MMFs based on different levels of exogenous shocks to interest rates and corporate spreads.

### Risks, buffers and regulatory thresholds
- Risks assessed:
  - Market risk (valuation impact through repricing and credit spread risk) and interest rate risk on the investment fund portfolio.
  - For MMFs: inability to maintain Constant NAV due to interest rate and corporate spread shocks.
- Buffers:
  - For AIFs and UCITS, there is no buffer requirement; change in NAVs affects outflows.
- Regulatory/market-based standards:
  - According to the IMMFA Code of Practice and ERSB Regulation, escalation procedures should exist for deviation between the published price and the shadow NAV above 20 basis points.

---

### Investment Funds Sector: Liquidity Risk (Top-Down by IMF)

### Institutional perimeter and data
- Institutions included:
  - 1,085 largest open-ended funds.
  - Largest UCITS Investment Funds with total net assets over 1 billion euros, totaling close to 2.3 trillion euros in total net assets.
  - Largest open-ended AIFs with total net assets over 1 billion euros subject to same reporting standards as UCITS, totaling close to 400 million euros in total net assets.
- Market share:
  - 70 percent of total net assets of UCITS within full reporting scope.
  - Roughly 50 percent of total net assets of open-ended AIFs subject to same reporting standards as UCITS in full reporting scope.
- Data and baseline date:
  - Source: Supervisory data.
  - Latest data: March 2023.

### Methodology and channels
- Methodology:
  - Liquidity measure based on cash and high-quality liquid assets.
  - Flow-performance Model to integrate impact of macro shock on redemptions.
  - Models of market depth to integrate second round effect from asset sales, accounting for illiquidity.
  - Incorporation of intersectoral linkages, especially with Luxembourg banks, to assess liquidity access capacity.

### Risks, shocks and reporting
- Risks:
  - Severe redemption shock following asset devaluations.
  - Funding liquidity and inability to sell assets to cope with redemptions.
  - Market liquidity leading to second round effects.
- Buffers:
  - Stock of highly liquid assets (HQLA).
- Tail shocks:
  - Initial shock from scenarios on Total Net Assets and redemption shock estimated from a model relating funds flows to macrofinancial variables.
  - Second round effects from price effects due to asset sales.
  - Exogenous monthly redemption shock equal to the first percentile of historical net flows.
- Sensitivity analysis:
  - Reverse stress test based on exogenous levels of redemptions (as a percentage of NAV) applied homogeneously to all funds in the test sample.
- Reporting format:
  - Number of funds with a redemption coverage ratio below one.
  - Total net assets of funds with RCR below one, as a percentage of aggregate total net assets.
  - Liquidity shortfall amount for individual funds after redemptions.

---

### Insurance Sector: Solvency and Liquidity Risk (Top-Down by IMF)

### Solvency risk — institutional perimeter and data
- Number of institutions:
  - 10 life insurers.
  - 12 non-life insurers and reinsurers.
- Market share:
  - Life: ~83 percent of investment assets.
  - Non-life: 81 percent of gross written premiums.
- Data: Supervisory reporting (Solvency II Quantitative Reporting Templates).
- Reference date: 30 June 2023.

### Solvency methodology, scenarios and sensitivities
- Methodology:
  - Investment assets: market value changes after price shocks.
  - Liabilities: valuation change due to interest rate shock.
  - Impact on available capital (net assets = stressed assets minus liabilities).
  - Recalculation of the solvency capital requirement.
- Time horizon: Instantaneous shock.
- Scenario analysis:
  - Adverse scenario aligned with investment fund risk analysis, with additional granularity on market and interest rate risks.
  - Single-factor sensitivities:
    - EUR risk-free rate term structure +/-200bps.
    - EUR external value +/-20 percent.
    - Equity prices shock: -50 percent.
    - Domestic sovereign shock: +500bps.
    - Default of largest and the three largest banking counterparties.
- Risk factors:
  - Market risks (equity, property), interest rate risks, credit risks (credit spread risk, default of largest counterparty), currency risks.
- Buffers and mitigating factors:
  - Eligible own funds.
  - Loss-absorbing capacity of deferred taxes.
  - No management actions assumed.
- Regulatory/accounting standards:
  - Solvency II.
  - National GAAP.
- Reporting format:
  - Impact on value of assets and liabilities.
  - Impact on solvency ratio (SCR coverage).
  - Aggregated capital shortfall and dispersion across companies.
  - Contribution of individual shocks.

### Liquidity risk — institutional perimeter and methodology
- Number of institutions: 8 life insurers.
- Market share: Life: ~90 percent of investment assets in non-unit-linked business.
- Data: Supervisory reporting (Solvency II QRTs) and additional data request to life insurers.
- Reference date: 30 June 2023.
- Methodology: Outflow through surrenders of guaranteed life insurance policies.
- Time horizon: One month.
- Scenario: Simulated monthly outflow which exceeds highest historical outflow by 50 percent.
- Buffers and mitigating factors:
  - Holdings of highly liquid assets.
  - Surrender payouts to policyholders according to contractually allowed periods.
  - No management actions assumed.
- Reporting format:
  - Cash in- and outflows.
  - Coverage of net outflows by liquid assets.
  - Distribution across companies.

---

### Adverse Scenario Calibration and Key Macro Assumptions (Appendix IV)

### Benchmarks and GaR
- Two benchmarks used for calibrating severity:
  - First benchmark: application of at least a 2 standard deviation shock in terms of Luxembourg 2-year GDP growth (1990-2022 benchmark):
    - Cumulative decline of GDP relative to the baseline over two years would reach 8.7 percentage points.
    - Cumulative drop in GDP level compared to 2023 would reach 4.7 percent.
  - Second benchmark: 5th percentile of the Growth at Risk framework using quantile regression.
- The two-year GaR for Luxembourg at the 5 percent threshold is estimated at -12.8 percent.

### Luxembourg-specific Bayesian VAR projections
- Bayesian VAR model (Minnesota prior) included endogenous variables: Luxembourg year-on-year GDP growth, HICP inflation rate, growth in Luxembourg-domiciled investment funds net assets. Explanatory variables included euro area GDP growth, US GDP growth, euro area inflation, Brent oil price inflation, the 10-year average long-term rate for euro area countries and the VIX index.
- Sample: 2000Q2–2022Q4; included observations: 91 after adjustments.
- Projected Luxembourg GDP growth path under the scenario:
  - 2024: -4.3 percent
  - 2025: -1.7 percent
  - 2026: +3.1 percent
- The projected path is equivalent to a 2.4 standard deviation shock with a cumulative decline in GDP over two years by 6.3 percent.

### Variables projected outside the model (conservative assumptions)
- Unemployment rate:
  - Assumed to peak at 12 percent of the labor force during the second year of the scenario (corresponds to a 2 standard deviation shock to the annual change in unemployment, 1990-2022 benchmark).
- House prices:
  - Assumed to decline by 30 percent in cumulative terms over the three-year stress horizon (bringing house prices back to their 2008 level).
- 10-year Luxembourg government bond yield:
  - Projections based on assumption that Luxembourg sovereign spread would converge to the euro area average in the adverse scenario.

### Selected external scenario projections (adverse vs baseline, percent unless noted)
- Euro area GDP growth rate:
  - Baseline: 5.6, 3.3, 0.7, 1.2, 1.8, 1.7 (Est.2021, 2022, Y0=2023, Y1=2024, Y2=2025, Y3=2026)
  - Adverse: 5.6, 3.3, 0.7, -3.0, 0.2, 1.7
- Euro area inflation rate:
  - Baseline: 2.6, 8.4, 5.6, 3.3, 2.2, 2.0
  - Adverse: 2.6, 8.4, 5.6, 6.3, 3.3, 1.6
- Euro area unemployment rate:
  - Baseline: 7.7, 6.7, 6.6, 6.5, 6.4, 6.3
  - Adverse: 7.7, 6.7, 6.6, 7.0, 8.0, 8.2
- Euro area ST rate:
  - Baseline: -0.4, 0.3, 3.3, 3.4, 3.0, 2.7
  - Adverse: -0.4, 0.3, 3.3, 6.0, 4.7, 3.3
- VIX index (historical benchmarking):
  - Baseline: 19.6, 25.6, 20.0, 20.0, 20.0, 20.0
  - Adverse: 19.6, 25.6, 20.0, 48.3, 35.0, 20.0

*Source: IMF staff calculations.*

### Appendix IV. Table

### Appendix IV. Table

### Luxembourg: Scenario Projections (Concluded) — Domestic variables (in percent)
- Table header timeline: Est. 2021 2022 Y0=2023 Y1=2024 Y2=2025 Y3=2026

- GDP growth rate
  - Baseline: 7.2 1.4 -0.4 1.5 2.4 2.5
  - Adverse: 7.2 1.4 -0.4 -4.3 -1.7 3.2

- Inflation rate (HICP-based)
  - Baseline: 3.5 8.1 3.2 3.3 2.2 2.0
  - Adverse: 3.5 8.1 3.2 5.5 3.6 2.0

- Inflation rate (CPI-based)
  - Baseline: 2.5 6.3 3.9 3.1 2.5 1.9
  - Adverse: 2.5 6.3 3.9 4.7 5.1 2.6

- Energy inflation (CPI-based)
  - Baseline: 18.9 32.9 -7.0 2.2 12.0 -2.4
  - Adverse: 18.9 32.9 -7.0 7.0 18.0 0.6

- Food inflation (CPI-based)
  - Baseline: 0.9 7.0 9.9 2.0 2.3 1.7
  - Adverse: 0.9 7.0 9.9 3.0 3.7 2.5

- Inflation excl food and energy (CPI-based)
  - Baseline: 1.6 3.6 4.0 3.4 1.7 2.3
  - Adverse: 1.6 3.6 4.0 4.8 4.1 2.8

- House price index growth
  - Baseline: 13.9 9.6 -2.3 -0.4 2.9 1.9
  - Adverse: 13.9 9.6 -2.3 -17.0 -15.0 -1.0

- Unemployment rate
  - Baseline: 5.7 4.8 5.2 5.8 5.9 5.8
  - Adverse: 5.7 4.8 5.2 8.7 12.1 11.1

- 10-yr gov.-bond yield
  - Baseline: -0.4 1.5 1.7 2.9 3.1 3.2
  - Adverse: -0.4 1.5 1.7 6.8 5.7 3.4

- Credit growth
  - Baseline: 4.5 6.6 2.4 4.7 5.4 4.9
  - Adverse: 4.5 6.6 2.4 0.0 0.0 0.0

- Inv. funds' net asset growth
  - Baseline: 17.8 -14.2 0.0 0.0 0.0 0.0
  - Adverse: 17.8 -14.2 0.0 -14.1 -7.5 4.5

- Wage (comp. per empl.) growth
  - Baseline: 5.1 5.9 5.5 3.5 2.9 2.3
  - Adverse: 5.1 5.9 5.5 3.6 3.7 3.4

*Source: IMF team calculations.*

### Appendix V — Methodological assumptions for banks’ balance sheet and profit projections
- General approach
  - Baseline balance sheet growth: projected to grow in line with nominal GDP.
  - Adverse scenario balance sheet: quasi-static balance sheet assumption (no growth in line with nominal GDP; banks not allowed to deleverage).
  - Note: In adverse scenarios, growth of net assets and exposures at default is usually lower than growth of total assets because provisions are higher.

- Projection of risk-weighted assets (RWAs)
  - For seven banks under Basel II Internal Rating-Based (IRB) approach: risk weights projected using Basel II credit risk formula depending on PD, LGD and asset correlation (risk weights typically rise in stress).
  - For seven banks under Basel II standardized approach: RWAs projected based on new defaulted loan projections and reduced proportionally to new provisions after provisioning rate assumptions; past due loans risk-weighted between 100 and 150 percent for unsecured portion net of specific provisions.
  - Exposures are risk-weighted net of specific provisions.

- Evolution of profits and income statement items
  - Most non-interest profit items projected to grow in line with nominal GDP in the baseline and remain flat in the adverse scenario.
  - Non-interest expenses assumed to grow in line with the inflation rate (wage indexation scheme in Luxembourg).
  - Income from extraordinary items assumed not to recur during 2024-2026 in both scenarios.
  - Non-performing loans assumed to provide no accrued income.

- Satellite econometric models (three income statement items)
  - Lending rates (ratio of interest income to net loans): panel model (annual change) with bank fixed effects over 2005–2022; explanatory variables: annual change in 3-month Euribor and annual change in VIX.
    - Result: model applied to new loans or loans with variable lending rates.
  - Funding costs (ratio of interest expenses to total liabilities excluding own funds): panel model with fixed effects over 2005–2022; same explanatory variables as lending rate model.
    - Custodian banks: higher R-squared (0.49) and negative (statistically insignificant) VIX coefficient (custodian banks may benefit from higher investor risk aversion via lower funding costs).
  - Net fee and commission income growth: panel model with fixed effects; explanatory variables: annual change in 3-month Euribor and annual growth rate of Luxembourg real GDP.
    - Custodian banks: higher R-squared (0.32) and a much larger significant effect of Euribor changes (a 1 percentage point increase in Euribor lowers annual growth of custodian banks’ net fee and commission income by 7 percentage points).

- Dividend distribution rules
  - Banks with total CET1 ratio below the Overall Capital Requirements (OCR) in any year of a scenario: not allowed to distribute dividends.
  - Banks that pass OCR distribute dividends at a rate of 50 percent.
  - Note: Pillar II requirements are not displayed as this is not public information for every bank.

### Appendix V — Selected estimation results (income statement satellite models)
- Table highlights (items and model diagnostics)
  - Change in lending rate and funding costs presented for Non-custodian banks and Custodian banks across Baseline and Adverse.
  - Change in 3-month Euribor rate coefficients (examples):
    - Change in lending rate: 0.374* (Non-custodian), 0.384 (Custodian), 0.324** (Non-custodian adverse), -1.971 (Custodian adverse), -7.082*** (Custodian funding cost specification)
    - Change in VIX index: 0.055** (Non-custodian lending), 0.021 (Custodian lending), -0.008 (funding cost)
  - Bank fixed effects: Yes for all models.
  - R-square examples: 0.14, 0.03, 0.49, 0.15, 0.32
  - # of banks: 3, 2, 15, 3, 21 (as displayed)
  - # of observations: 280, 145, 411, 749 (as displayed)
  - Significance notation: * at 10 percent; ** at 5 percent; *** at 1 percent.

### Appendix VI — Satellite models for credit risk: solvency stress testing methodology
- Key challenges and data approach
  - Short time series and limited crisis experience in Luxembourg limit statistical estimation of credit risk parameters.
  - FSAP team applied international benchmarks (experience from countries with severe banking crises, e.g., Ireland 2008-2013) and calibrated shocks based on credit loss sensitivities under extreme severity.
  - PD projections used quarterly data over 2005Q1–2022Q4; different data sources by exposure class:
    - Corporate, financial institutions and sovereign exposures: Moody’s/CreditEdge Expected Default Frequencies dataset on seven euro area advanced countries starting in 2002.
    - Retail (mortgage and non-mortgage): coefficients/elasticities from 2023 Ireland FSAP and cross-checked with 2024 Luxembourg FSAP household stress test projections.

- IFRS9-consistent provisioning and LGD assumptions
  - Calculations consistent with IFRS9: depend on loan exposures, stage transition matrices (guided by stressed PiT PDs) and PiT LGDs under stress.
  - Provisioning for IRB and STA modeled using IFRS9 transition matrix approach.
  - LGD assumed to increase by 30 percent in the adverse scenario compared to the initial level.

- Data access and usage
  - Supervisory data available in a physical data room at individual bank level, on solo, consolidated and sub-consolidated bases.
  - Publicly available data used: commercial providers, EBA Transparency exercise, banks’ annual reports.
  - Appendix VI Table 1 summary of FSAP access to supervisory data:
    - Data at the individual bank level: Fully available
    - Data aggregated along groups of banks: Partially available
    - Data aggregated at the banking system level: Not available

- PD projection methodology
  - Five asset classes PDs projected via panel-based approach.
  - TTC PDs for RWA calculation; one-year PiT PDs for expected credit losses.
  - TTC PDs smoothed by applying a cyclical parameter of 0.1 to annual change in projected PiT PDs.
  - Starting-point PiT PDs:
    - Banks reporting IFRS9 staging: initial PIT PDs derived from sum of quarterly inflows into Stage 3 over one year on a year-on-year basis (formula provided).
    - Banks not reporting IFRS9 staging: initial PIT PDs derived from sum of quarterly inflows into the nonperforming loan stock over one year on a year-on-year basis (formula provided).

- Statistical model specification
  - Logit transformation applied to PDs to constrain predictions between 0 and 1: Y = ln(PD/(1-PD)).
  - Logit-transformed PDs modeled as linear function of exogenous macrofinancial regressors and lagged dependent variable:
    - Y_{PDi,t} = δ Y_{PDi,t-1} + α + β X_{t-s} + ε_{PDi,t}
  - Determinants by exposure class (examples):
    - Household mortgage loans: unemployment rate (percent), Luxembourg 10-year bond yield, 3-month Euribor, y-o-y growth in Luxembourg stock price index, y-o-y growth in Luxembourg house price index.
    - Other retail: unemployment rate, Luxembourg 10-year bond yield, 3-month Euribor.
    - Non-financial corporate loans: y-o-y euro area real GDP growth rate, proxy euro area 10-year bond yield, lagged PD.
    - Loans to financial institutions: y-o-y euro area real GDP growth rate, spread between euro area 10-year bond yield and 3-month Euribor.
    - Sovereign exposures: y-o-y real national GDP growth rate, 3-month Euribor, national government bond yield (lagged by one period), lagged PD.
  - PDs/NPLs under stress computed as inverse logit function:
    - PD_{t,stress} = 1 / [1 + exp(−(α + β X_{t-1}))] * 100

- Appendix VI Table 2 — Selected estimation results (logit transform of one-year PD or EDF)
  - Dependent variable coefficients (examples across asset classes HH Mortgages, Other retail, Corporate, Institutions, Sovereigns):
    - Dependent variable (lagged by 1 period): -- 0.817*** 0.923*** (t-statistics in parentheses (17.43)(40.45))
    - Real euro area GDP growth rate: --- 0.026** -0.022** -0.005
    - Unemployment rate: 0.159* 0.086* --
    - 10-year government bond yield: 0.015 0.062* 0.059*** -0.001
    - 3-month Euribor rate: 0.024* 0.123* --- 0.009*** (lagged by 1 period)
    - Spread between euro area 10-year bond yield and 3-month Euribor: ---- 0.214*** -
    - Stock price yoy change: -0.006* -----
    - House price growth: -0.011* -----
    - Constant examples: -5.201*** -5.505*** -1.078*** -5.828*** -0.688***
    - R-square examples: 0.97 0.99 0.92 0.46 0.86
    - # of observations: 828 342 5 (as displayed)
  - Note: Significance notation as in tables; an increase in the exchange-rate variable denotes an appreciation of the euro against the dollar where used.

### Appendix VII — Investment fund stress testing methodology (liquidity stress test)
- Scope and sample
  - Largest open-ended investment funds included: all open-ended firms with total net assets over 1 billion euros in March 2023.
  - Sample AUM: 3,100 billion euros.
  - Sample composition: mostly UCITS; AIFs account for roughly 13 percent of the AUM of funds in the sample.

- Building blocks of the liquidity stress test
  - Transmission of macro shocks in the severe scenario to investment funds’ balance sheets.
  - Transmission from impact on NAV to the redemption shock faced by funds.
  - Assumptions regarding liquidation strategies for funds facing net redemptions.

- Initial shock design
  - Assets aggregated into 42 asset groups, each subject to its own price shock.
  - Grouping splits assets across asset classes (equities, sovereign bonds, corporate bonds, deposits, investments in MMFs, other liquid assets, real estate, other assets) and geographical regions (Luxembourg, US, Germany, France, UK, Japan, Other Euro Area, Rest of the World).
  - This breakdown aims for granular impact while preserving tractability.
  - Figure 1 presents breakdown of total assets of funds in sample into main asset groups used in the stress test sample.

*Source: IMF team calculations.*

### Appendix VII. Figure 1. Luxembourg: F Holdings of Asset Groups as a Share of Total

### Appendix VII. Figure 1. Luxembourg: F Holdings of Asset Groups as a Share of Total

### Assets for the Stress Test Sample
- Source: BCL and IMF Staff calculations.
- Note: Data as of March 2023.
- Methodology for initial shock to investment funds:
  - The initial shock is based on application of estimated changes in asset prices using the financial simulator model (outlined in the Stress Test Strategy, Section B) to tradable assets.
  - Asset groups such as liquid short-term deposits are assumed to retain their full value.
  - Assets subject to market fluctuations were matched to market price indices and estimates for the change in price were made, consistent with the results of the adverse stress test scenario.
  - This methodology ensures the changes in prices are consistent with the historical co-movement of the price of these assets.
- Real estate shocks:
  - Based on IMF model of House Price at risk from Deghi and others (2021).
  - The model produces one-year ahead estimates for changes in commercial real estate asset prices based on the fifth percentile of the distribution.
  - The model is estimated using 2023Q2 data.
  - The yearly shock from the model is translated into a two-month shock and applied to the holdings of investment funds.

### Redemption Shock Calibration
- Calibration approach:
  - Constructed using a historical flow-performance relationship.
  - Model estimated for the sample of regulated funds using monthly data from 2016 until 2023.
  - The coefficient on returns establishes the statistical relationship between returns and net flows.
  - The model is run separately by fund investment strategy to capture different behavioural patterns.
- Model specification (as presented):
  - f_{P,t} = α + Σ_{k=1}^{12} β_k f_{P,t−k} + Σ_{h=1}^{12} γ_h reer_{P,t−h} + θ_P,t X_{P,t} + ε_{P,t}
  - X_{P,t} is a vector of fund characteristics including fund size (negatively related with redemption levels), investor concentration (positively related with redemption levels), and fund leverage (positively related with redemption levels).
  - The estimated percentage fall in NAV from applying asset price changes to funds’ balance sheets is then used as input in the corresponding strategy-specific model for each fund to predict redemptions.
- Complementary use of literature estimates:
  - Mirza and others (2020) Fama-Macbeth results used where the flow-performance model produces lower estimates.
  - For fund strategies where Mirza et al. parameters exceed CSSF-based estimates, Mirza et al. parameters are used to generate a more conservative distribution of net outflows.
  - Funds with estimated coefficients not statistically different from zero are assumed not to experience redemptions following changes in NAV; their liquidity risk is tested via scenarios based on historical exogenous redemptions.
- Historical GFC redemption scenario:
  - Uses the same asset groups as the main scenario but applies changes in asset prices during September and October of 2008 based on aggregate market indices.
  - These changes are applied to the balance sheet of investment funds in the stress test sample in the same way as the main severe scenario.
- Exogenous redemption shocks (historical net flows):
  - Calibrated as an instantaneous shock on the distribution of net flows.
  - Two calibration assumptions:
    - Homogeneity: each fund within the same strategy faces the same redemption shock based on the distribution of all individual fund net flows in that category.
    - Heterogeneity: the redemption shock is calibrated separately for each fund based on its own historical data.
  - The exercise uses the first, third and fifth percentile of the distribution of net flows.

### Flow-Performance Elasticities (Appendix VII. Table 1 excerpts)
- Source: Mirza and other (2020) and IMF staff calculations.
- Strategy — Coefficient in ECB paper — Parameter in ST
  - Bond funds — 0.40 — 0.40
  - Equity funds — 0.20 — 0.23
  - Mixed funds — 0.29 — 0.29
  - Hedge funds — 0.14 — 0.14
  - MMF — 0.99 — 0.99
  - Real estate funds — -0.01 — 0.00
  - Commodities — 0.29 — 0.29
  - Fund of funds — 0.29 — 0.29
  - Other funds — -0.03 — 0.00

### Liquidation Strategies and Liquidity Weights
- Liquidity weights assigned by instrument rating and type (Appendix VII. Table 2):
  - Cash — 100%
  - Sovereign bonds:
    - AAA to AA- — 100%
    - A+ to A- — 85%
    - BBB+ to BBB- — 50%
    - High-Yield Below BBB- — 0%
  - Corporate bonds:
    - AAA to AA- — 85%
    - A+ to A- — 50%
    - BBB+ to BBB- — 50%
    - High-Yield Below BBB- — 0%
  - Securitized Products — (investment grade 100% shown in table header)
  - Equities — 0% for BBB+ to BBB- and below
- Note on corporate bond holdings:
  - Data on corporate bond holdings do not include the associated credit rating for each security.
  - Alternatively, the liquidity weight of corporate bond portfolio was calculated using average holdings of bond holdings by credit rating and investment strategy as shares for the liquidity weights of corporate bonds.
- Manager liquidation strategies:
  - Slicing (prorata) method: sell assets in line with portfolio weight.
  - Waterfall methods: prioritize sale of assets according to a pre-specified order (two variants used).
  - Mixed approach.
- Stress test uses three liquidation assumptions:
  - Slicing methodology.
  - Waterfall classical (cash-first) — managers first use cash buffers before liquidating securities.
  - Waterfall securities-first — managers empirically may liquidate most liquid securities first to sell before price falls.
- Implications of liquidation assumptions:
  - Securities-first waterfall → greater second-round effects on asset prices due to larger security sales; funds may not withdraw from deposits at banks.
  - Cash-first waterfall → greater impact on bank counterparties as funds withdraw deposits; in the adverse scenario, under cash-first funds would withdraw 25 percent of deposits.
  - Slicing yields results between the two waterfall extremes.

### Waterfall Order (Appendix VII. Table 3)
- Waterfall order — Securities first / Cash first
  1. Sovereign BBB or higher bonds / Cash/Deposits
  2. Corporate bonds / Sovereign BBB or higher bonds
  3. Sovereign below BBB / Corporate bonds
  4. Equities / Sovereign below BBB
  5. Cash/Deposits / Equities
  6. MMF / MMF
  7. Other liquid assets / Other liquid assets

### Price Impact of Asset Sales (Appendix VIII)
- Rationale:
  - Given the size of the financial sector, sale of a significant share of assets by investment funds, banks and insurance companies in Luxembourg is likely to move market prices of securities.
  - Changes in asset prices affect remaining balance-sheet values, generating additional solvency and liquidity risks.
  - Asset sales across sectors are aggregated into broad asset classes to capture impact originating in Luxembourgish financial sector; impacts incorporated into balance sheets for stage three of the stress test.
- Market impact function and market depth:
  - Market impact function referenced: V = −V0 φ(q; P, τ), where φ(q; D, τ) = q / D
  - q is the value being sold; P is the market depth.
  - D = c ADV / (σ √τ)
  - ADV is the average daily volume in the market, σ is daily volatility, τ is the time horizon for liquidation, c is a scaling parameter.
  - Market depth over time horizon τ is c times the ratio between average daily trading volumes and asset volatility, multiplied by the square root of τ.
  - Price impact is lower when the time horizon is longer.
- Market depth estimates:
  - Taken from past FSAPs and the literature.
  - US securities: estimates rely on US 2020 FSAP.
  - Sovereign debt of other major countries: Cont and Schaanning (2017) and Coen and others (2019).
  - Market depth for other securities estimated by ratios of equivalent securities or based on ratio of total outstanding value.
  - Main estimates rely on market depth under stress (using volatility during the global financial crisis) sold over a period of 7 days; other estimates used for sensitivity analysis.
- Incorporation into fund NAV and second-round redemptions:
  - For investment funds, price impact of asset sales is incorporated by re-estimating outflows following the fall in NAV caused by lower prices.
  - After estimating price impact, NAV of each fund is recalculated to reflect costs due to liquidation on remaining portfolio.
  - Negative returns lead to a second wave of redemptions, magnitude depending on the flow-performance relationship.

### Estimated Price Impact of 1 bn Asset Sales (Appendix VIII. Table 1 excerpts)
- Sources: US FSAP (2020), Cont and Schaaning (2017), Coen and others (2019) and IMF staff calculations.
- US FSAP (Asset sale period = 1 day / Asset sale period = 7 days; ADV (bn) column header present)
  - UST 54 50.13 0.30.05 0.1
  - Corp IG 213.5 77.7 1.35 2.9
  - Corp HY 126.4 622.3 2.4 48.4
  - US Equities 3200.88 2.80 0.33 1.1
  - EM Debt 812.50 42.5 4.72 16.1
- Schanning (Sovereign)
  - UK 34.1 1.76 3.460.67 1.31
  - Germany 18.9 3.17 6.241.20 2.36
  - France 10.2 5.88 11.55 2.22 4.37
- Estimated Sovereign — Other Euro Area 1.64 3.23 0.62 1.22
- Corporate (Estimated)
  - UK 9.21 18.10 3.48 6.84
  - Germany 16.62 32.65 6.28 12.34
  - France 30.80 60.50 11.64 22.87
- Equities (Estimated)
  - UK 11.99 23.55 4.53 8.90
  - Germany 21.63 42.48 8.17 16.06
  - France 40.07 78.72 15.15 29.75
- Note: Table presents impacts for Asset sale period = 1 day and Asset sale period = 7 days; ADV (bn) column appears in table header.

*Source: IMF staff compilation from Appendix VII and Appendix VIII of the provided content.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2024/english/1luxea2024008.pdf_