## 1mltea2019009 ## Source details **Canonical URL:** [1mltea2019009](https://www.imf.org/-/media/files/publications/cr/2019/1mltea2019009.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2019/1mltea2019009.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2019/1mltea2019009.pdf.json) --- ### INTRODUCTION: scope and context - Malta FSAP stress testing exercise followed the IMF’s 2018 Euro Area FSAP and ran concurrently with the 2018 stress test of the European Banking Authority (EBA). - Solvency stress test covered 11 banks representing 93 percent of the banking sector assets (excluding foreign branches). - Exercise components: - Solvency stress tests (top-down, scenario-based, sensitivity analysis). - Liquidity stress tests using multiple approaches (LCR-based, NSFR-based, cash flow-based). - Interconnectedness analyses (cross-border and domestic networks, market-based spillovers). - Household stress test using micro evidence and the same macro stress scenario. ### Financial system structure and size - Banking sector total assets: 4⅓ times GDP; banks: €48 billion in assets, 4.3 times GDP (end-2017). - Banks accounted for two-thirds of financial sector assets (excluding captive financial institutions and money lenders) at end-2017. - Core domestic banks: €23 billion total assets at end-2017; account for 81 percent of banking sector assets (excluding foreign branches) and 99 percent of mortgages to residents. - Three largest core domestic banks are Significant Institutions (SIs) and hold 87 percent of this category’s assets (42 percent of total bank assets). - Three core domestic banks are foreign-owned. - Non-core domestic banks: €2.2 billion total assets. - International banks: €23 billion total assets; two non-EU branches account for €19 billion (83 percent of this category’s assets) and have negligible exposure to the Maltese economy. - Foreign bank branches: 1.7 times GDP. - Insurance: 1 time GDP; Investment funds: 1.1 times GDP. - Private pension market: negligible. ### Key metrics, strengths, and vulnerabilities - Capital and liquidity: - Total capital adequacy ratio (2017): 21.2 percent of RWA. - Tier 1 Capital ratio (2017): 19.0 percent. - Core domestic banks’ capital ratio improved from 16.8 percent in 2017Q4 to 17.2 percent in 2018Q1. - Liquidity: ample; high claims on the Eurosystem for some banks. - Profitability: supported by stable net interest margins and operating costs; risks to sustainability include contraction of corporate loan books, weakening property market, large exposures to low-yield bonds, increased regulatory compliance costs, higher loan loss provisioning requirements, and implementation of EBA guidelines on Minimum Required Eligible Liabilities. - Asset quality and NPLs (end-2017, excluding non-EU branches): - Total NPL ratio: 4.1 percent of total loans (down from 6.6 percent in 2014Q4). - Core domestic banks: 4.1 percent; Non-core domestic banks: 2.3 percent; International banks: 5.3 percent. - NFC NPLs: declined from 11.8 percent in 2014Q4 to 9 percent in 2017. - Sectoral NPLs: Construction 27.8 percent; Corporate real estate 13.9 percent. - Loan loss provisioning: core domestic and international banks ≈ 34 percent of NPLs; non-core domestic banks 57 percent of NPLs. - Property and household indicators: - Resident mortgage lending growth: 8.5 percent annually since 2013. - Property-based lending to residents accounts for 62 percent of core domestic banks’ total loans to residents. - Core domestic banks’ share of property lending to total assets: 25 percent; all banks in stress test sample: 22 percent. - Residential housing prices rose by about 33 percent (2010Q4 to 2017Q4). - Household debt at 108 percent of gross disposable income in 2017. - Home ownership ratio: 82 percent (2017). - Share of homeowners with debt: about 20 percent. - LTV of new mortgages: averaging 77 percent in 2017Q4. - DSTI of new mortgages: averaging 21 percent in 2017Q4. - Debt-to-financial-wealth ratio: stable at 23 percent. ### SOLVENCY STRESS TEST: aggregate results and drivers - Coverage: 11 banks covering 93 percent of banking sector assets (excluding foreign branches); sample includes all 6 core domestic banks, two non-core domestic banks, and three international banks. - Capital framework: based on Pillar I Basel II minimums; banks allowed to deplete 2.5 percent capital conservation buffer; O-SII buffer phased in 2018–2020 (ranges 0.5 percent to 2.0 percent). - Macroeconomic adverse scenario (2018–2020): - GDP: -2.3 percent in 2018; -1.0 percent in 2019; 2.9 percent in 2020 (average decline 0.1 percent per annum; 15.4 percentage points cumulative decline relative to baseline; equivalent to a 2.07 standard deviation shock). - House prices: exogenous cumulative decline of 30 percent over 3 years. - Quasi-static approach (balance sheets grow with nominal GDP; no nominal contraction; off-balance sheet triggered portion: baseline 5 percent; adverse 10 percent): - Aggregate CAR: 19.9 percent (base) → 16.0 percent (adverse). - CET1 ratio: decline of 329 bps to 14.5 percent over three years. - Aggregate Tier 1 relative to RWA declines by 329 bps to 14.5 percent. - Aggregate leverage ratio declines by 129 bps from 8.9 to 7.7 percent; no bank below 3 percent leverage threshold, one bank close. - Core domestic banks: CET1 decline 270 bps to 12.4 percent. - Other banks: CET1 decline 564 bps to 24.3 percent. - Aggregate total capital adequacy ratio declined by 390 bps to 16.0 percent; core domestic decline 341 bps to 13.9 percent; other banks decline 580 bps to 26.0 percent. - Three banks could not meet regulatory minimum for one or more capital ratios; estimated recapitalization needs about 0.14 percent of GDP. - Main loss drivers (three-year horizon): - Credit risk (loan-loss provisioning): largest driver, accounts for 506 bps decline in CET1. - Additional RWA (newly classified NPLs, loan growth, triggered off-balance sheet): reduces CET1 by 186 bps. - Valuation losses on securities: reduce CET1 by 107 bps. - Pre-provision net revenue increases aggregate CET1 by 474 bps relative to starting point. - Static-balance-sheet approach (comparison): - CET1 and T1 ratios decline by 229 bps (from 17.8 percent to 15.5 percent). - Total capital ratio decline 286 bps (from 19.9 percent to 17.0 percent). - One bank falls below 8.0 percent total capital ratio; total recapitalization needs 0.02 percent of GDP. - Sensitivity and concentration tests: - Aggregate single-largest-exposure equals 15 percent of CET1. - Three banks have single-largest-exposure >25 percent of CET1 (net of mitigants). - Simultaneous default of five largest borrowers: - Zero-recovery: aggregate CET1 decline 7.1 percentage points (17.8 → 10.6); four banks below 4.5 percent CET1. - 65 percent provisioning: aggregate CET1 decline 4.5 percentage points (17.8 → 13.3); one bank below 4.5 percent CET1. - Parallel interest rate increase of 200 bps would increase aggregate CET1 by 0.6 percentage point. - Funding cost shock of 100 bps reduces aggregate CET1 by 0.9 percentage point; no bank breaches minimum CET1. ### LIQUIDITY STRESS TESTS: LCR, NSFR, cash-flow results - Starting funding structure: - System loan-to-deposits ratio: 41 percent; core domestic banks: 58 percent; other banks: 39 percent. - Excess liquidity placements in central bank and debt securities account for 46 percent of total deposits of sample banks. - Retail funding accounts for 50 percent of core domestic bank total funding; other banks retail deposits 19 percent. - Non-resident deposits: 24 percent of total retail and wholesale deposits; 85 percent of these in current and savings accounts with open maturities. - LCR-based test (aggregate LCR = 190 percent baseline): - Four LCR scenarios: Baseline (B), S1 (higher run-off retail & wholesale), S2 (S1 + market value decline), S3 (S2 + non-resident deposit withdrawals). - Baseline: aggregate LCR 190 percent; all sample banks meet 80 percent threshold. - S1: aggregate LCR 146 percent. - S2: aggregate LCR 142 percent; all sample banks meet 80 percent but three approach threshold. - S3: aggregate LCR 121 percent; three banks fail; total liquidity shortfall for three banks: 0.7 percent of GDP. - Note: with 100 percent threshold, shortfall would be 1.6 percent of GDP. - Foreign currency LCR: - U.S. dollar: aggregate USD-LCR declines from 84.1 percent to 56.5 percent under adverse scenario with non-resident dollar outflows; out of seven banks, three struggle in all scenarios including baseline. - British pound: aggregate GBP-LCR declines from 156 percent to 105 percent; one of four banks struggles to meet 80 percent minimum in all scenarios. - NSFR-based test: - Baseline (end-2017): aggregate available stable funding €19.1 billion; aggregate required stable funding €11.9 billion; aggregate NSFR 160 percent; one core domestic bank NSFR below 100 percent with funding gap €681 million. - Adverse: aggregate NSFR declines to 130 percent; one additional bank below 100 percent; total stable funding gap for two small banks €992 million (€311 million additional to baseline); shortfall equivalent to 8.9 percent of GDP or 3.8 percent of total sample bank assets. - Cash flow–based test (maturity ladder; end-March 2018 data; time horizons 1 month, 3 months, 5 days): - 1-month, Scenario 1: two banks fail; total shortfall 0.7 percent of total sample bank assets, or 1.8 percent of GDP. - 1-month, Scenario 2 (excludes committed lines): total shortfall 0.8 percent of total sample bank assets, or 1.9 percent of GDP. - 3-month, Scenario 1: two banks fail; total shortfall 1.1 percent of total sample bank assets, or 2.7 percent of GDP. - 3-month, Scenario 2: total shortfall 1.2 percent of total sample bank assets, or 2.8 percent of GDP. - 5-day, Scenario 1: three banks fail; total shortfall 0.4 percent of total sample bank assets, or 0.9 percent of GDP. - 5-day, Scenario 2: total shortfall 0.4 percent of total funding assets, or 1.0 percent of GDP. - Funding concentration sensitivity: - Withdrawal of top five depositors: aggregate LCR declines from 190 percent to 146 percent; three small banks fall below 80 percent LCR; total liquidity shortfall of three banks: 3 percent of GDP. - 60 percent withdrawal from key sectors (arts, entertainment, recreation including gaming; accommodation and food service; financial sector): aggregate LCR declines from 190 percent to 115 percent; three banks below threshold; shortfalls total 0.9 percent of GDP. - Summary liquidity vulnerabilities: - Aggregate resilience strong due to excess liquidity, but some small banks vulnerable to combined wholesale and non-resident deposit runs, funding concentration, and foreign-currency liquidity pressures. ### INTERCONNECTEDNESS AND CONTAGION ANALYSIS: cross-border and domestic - Frameworks: - Balance sheet-based network analysis using CoMap (Covi, Gorpe, and Kok 2019), extending Espinosa-Vega and Sole (2011). - Market-based analysis using Diebold-Yilmaz (2014) approach; market-based covered four listed Maltese banks. - Cross-border interbank network: - Sample: 21 Maltese banks and 58 non-Maltese banks (full network 79 banks). - Other banks report large exposures equal to about 83 percent of their total assets; core banks’ large exposures ≈ 43 percent of total assets. - Interbank exposures: 21 percent of total large exposures for core banks; 24 percent for other banks. - Contagion findings: - Of top 20 most contagious banks, 18 are non-Maltese. - Average contagion index ≈ 2.6. - Hypothetical default of the most contagious bank (Bank 1) yields average losses to Maltese banks close to 5 percent of their capital buffers; Bank 1 causes contagion to six Maltese banks with individual losses 20–140 percent of capital buffers; two of the six experience default. - Total contagion defaults in these simulations amount to 12 defaults in the default zone. - Credit shocks dominate funding shocks; contagion highly concentrated and largely direct. - Domestic cross-sectoral network (24 banks, 8 insurers, 8 funds; MFSA dataset June 2017): - Insurers have the largest exposure within this network (almost half the share of total bilateral claims). - Core banks hold two-thirds share of total obligations. - Funds and insurers incorporated via security-by-security reporting; dataset yields 40-by-40 matrices by instrument. - Contagion appraisal: - Core banks induce normalized average losses of 55 percent to insurers and 25 percent to funds. - Industry average contagion index value: 2.1. - Insurers vulnerability index average: 8 percent of their capital buffer. - Funds vulnerability average: 7.8 percent. - Funds drive 8 out of 18 total defaults across simulations; insurers and other banks absorb remainder with a 6-to-4 split. - Cascade potential exists; amplification average ratio about 1-to-6 (losses in subsequent rounds versus initial round) in some simulations. - Market-based spillovers: - Three of four listed Maltese banks peripheral in global network; one more central. - Euro area and other European banks account for a large portion of net inward spillovers to Maltese banks. - Robustness and caveats: - Sensitivity tests (less-conservative thresholds) reduce contagion and vulnerability indexes (contagion/vulnerability averages decline about 60 percent in domestic less-conservative scenario); number of contagion defaults declines from 18 to 7. - Modeling limitations: single solvency threshold assumption, point-in-time data (2017Q2 and 2017Q4), limited counterparty-level funding data, limited market coverage (four listed banks). ### HOUSEHOLD STRESS TEST: micro-based findings - Data: 2017 Eurosystem Household Finance and Consumption Survey sample for Malta (1,004 households; data based on 2016); projections to 2018–20 under adverse scenario. - Adverse scenario parameters (2018–20, percent): - Output Growth: 2018 = -2.3, 2019 = -1.0, 2020 = 2.9. - Inflation: 2018 = -0.1, 2019 = -0.6, 2020 = -0.2. - Growth in expenditure: 2018 = 1.3, 2019 = 1.3, 2020 = 2.3. - Growth in wages: 2018 = -2.8, 2019 = -4.7, 2020 = -2.9. - Changes in Mortgage rate (ppt): 2018 = 1.9, 2019 = 0.4, 2020 = -0.1. - Growth in house prices: 2018 = -18.2, 2019 = -14.1, 2020 = -0.4. - Growth in equities: 2018 = -28.6, 2019 = -1.8, 2020 = 8.3. - Pre-shock household indicators: - 28 percent of households had debt; of these, about 59 percent receive some social transfers; 9 percent have negative extended financial margin. - Pre-shock prevalence of household distress: 3.7 percent of households. - Stress-test outcomes: - Average implied PD for household loans increases from 3.7 percent to 8.9 percent over the stress period (multiplier 2.4). - Average PD in 2018 increases to 5.8 percent. - Second-year PD average: 10.4 percent; final-year PD average: 10.5 percent. - More than 91 percent of indebted households can cover at least 12 months of income shortages using liquid assets. - Debt burden of bottom four deciles accounts for about 6.4 percent of total mortgage debt, or less than 6.3 percent of total household debt in the sample. - IMF staff conclusion: households overall resilient; low-income households more vulnerable; credit losses from distressed households likely manageable given low share of total debt held by most vulnerable deciles. ### POLICY RECOMMENDATIONS AND OPERATIONAL ACTIONS - Solvency — CBM, near term (1–2 years): - Enhance data management for credit risk analysis to run stress tests timely and accurately. (¶59) - Move toward expected loss-based credit risk models for stress testing, particularly with IFRS 9. (¶59) - Conduct regular sensitivity tests for credit risk, including asset concentration at borrower and sector levels. (¶59) - Liquidity — CBM; CBM, MFSA; near term: - Strengthen liquidity stress-testing by differentiating resident and non-resident deposit shocks; conduct currency-specific and longer-horizon liquidity tests. (¶93) - Conduct regular sensitivity analysis on funding concentration. (¶93) - Enhance data management for liquidity stress-testing, cross-check consistency, and prepare for NSFR implementation. (¶93) - Interconnectedness — CBM, MFSA; medium term (3–5 years): - Expand interbank analysis based on large exposures and funding concentration to capture cross-border bilateral exposures. (¶116) - Enhance network analysis of interlinkages to fully incorporate multiple transmission channels. (¶116) - Integrate contagion risks into macro stress-testing frameworks to inform calibration of macroprudential tools. (¶116) - Supervisory implications: - Strong monitoring of bank NPL resolution plans, particularly for some small banks. - Consider concentration risk when setting Pillar 2 capital buffers during the Supervisory Review and Evaluation Process. - Liquidity operational steps: - Regular sensitivity tests for reliance on top depositors and sectoral funding concentrations. - Regular verification and cross-checking of resident vs non-resident deposit reporting; readiness checks for NSFR implementation. - Contagion monitoring: - Periodic monitoring of cross-border linkages and enhancement of inter-sectoral linkages analysis for early warning. *Source: INTRODUCTION section and related chapters of the Malta FSAP stress testing chapter (IMF).* ### INTRODUCTION ________________________________________________________________________________9 ### INTRODUCTION ### Scope and Context - The Malta FSAP stress testing exercise followed the IMF’s 2018 Euro Area FSAP and ran concurrently with the 2018 stress test of the European Banking Authority (EBA). - A comprehensive set of stress tests and interconnectedness analyses were conducted, tailored to the Maltese financial system. - The solvency stress test covered 11 banks representing 93 percent of the banking sector assets (excluding foreign branches). - The exercise included: - Solvency stress tests (top-down, scenario-based, sensitivity analysis). - Liquidity stress tests using multiple approaches (LCR-based, NSFR-based, cash flow-based). - Interconnectedness analyses (cross-border and domestic networks, market-based spillovers). - A household stress test using micro evidence and the same macro stress scenario. ### Financial System Structure and Size - Total assets of the banking sector equal 4⅓ times GDP. - At end-2017, banks accounted for two-thirds of financial sector assets (excluding captive financial institutions and money lenders). - Core domestic banks account for 81 percent of banking sector assets, excluding foreign branches. - Foreign bank branches equal 1.7 times GDP; their Maltese exposure is negligible. - Business models and market orientations vary across banks; many other banks are primarily funded by wholesale and non-resident deposits. ### Key Metrics, Strengths, and Vulnerabilities - Banks are generally well-capitalized, liquidity is ample, and profitability has been healthy. - Risks identified: - High exposure to key advanced and emerging economies. - Growing exposure of core domestic banks to property-related loans amid rapidly rising residential housing prices with some indicators of overvaluation. - Pockets of distressed corporate loans affecting some bank balance sheets. - High share of non-resident deposits in international and non-core domestic banks, increasing vulnerability to changes in Malta’s comparative tax advantage and reputational risk. - Cyclical indicators: - The credit-to-GDP gap has been negative for several years, mainly due to a large decrease in the share of non-financial corporation (NFC) loans to GDP. - Recent strong economic growth has improved asset quality and supported capital above minimum requirements. ### Solvency Stress Test Findings - Aggregate resilience: - Banking sector CAR at end-2017 would decline from 19.9 percent to 16 percent of total risk-weighted assets under the adverse scenario. - CET1 ratio would decline by 329 bps to 14.5 percent over three years (through end-2017). - Main contributors to losses: loan-loss provisioning, additional risk-weighted assets, and valuation losses from debt securities. - Distributional impacts: - Three small banks would be under pressure; estimated total recapitalization needs about 0.14 percent of GDP. - Risk drivers: - Credit risk from corporate and mortgage loans is the main contributor to declining capital ratios. - Concentration risk is high; default of largest borrowers would significantly impact capital of several banks (small and large). - For some core domestic banks, large borrowers are concentrated in construction and real estate. - Policy implication highlighted: - Strong monitoring of bank NPL resolution plans, particularly for some small banks. - Consider concentration risk when setting Pillar 2 capital buffers during the Supervisory Review and Evaluation Process. ### Liquidity Stress Test Findings - LCR-based test: - A combined shock (withdrawal of wholesale and retail deposits, plus withdrawals of non-resident deposits) would reduce system LCR from 190 percent to 121 percent. - Most banks resilient; some small banks come under pressure. - Cash flow-based test: - Total liquidity shortfall for the 3-month horizon would be about 2.8 percent of GDP. - NSFR-based test: - Most banks are positioned to meet a level of 100 percent NSFR, but a few small banks face structural longer-term refinancing and funding problems. - Additional liquidity concerns: - Some banks are reliant on funding from a few large depositors (funding concentration). - On foreign currency liquidity, some banks showed insufficient liquid assets to cover withdrawals in certain major foreign currencies even under the baseline scenario. - Policy implications: - Strengthen liquidity stress-testing by differentiating resident vs. non-resident deposit shocks, conducting currency-specific tests, and extending time horizons. - Regular sensitivity analysis on funding concentration. - Improve data management for liquidity stress testing and prepare for NSFR implementation. ### Interconnectedness and Contagion Analysis - Cross-border vs domestic contagion: - Contagion risk through domestic interlinkages is currently higher and more widespread than contagion risk through cross-border interbank exposures. - The “contagion index” produces mild readings for cross-border networks but points to concentrated distress in smaller banks. - Expanded domestic analysis: - Including funds and insurers indicates insurers are vulnerable to contagion risk from core banks. - Policy implication: - Address structural challenges related to concentration of large exposures to reduce domestic and spillover vulnerabilities. - Monitor cross-border linkages periodically and enhance inter-sectoral linkages analysis for early warning. ### Household Stress Test and Complementary Analysis - A household stress-test incorporating micro evidence was performed to complement solvency assessment using the same macro stress scenario. - Household-level findings contribute to understanding of solvency risk transmission to banks (see results and figures in detailed sections). ### Key Quantitative Results and Indicators - Coverage of solvency test: 11 banks representing 93 percent of banking sector assets (excluding foreign branches). - Banking sector total assets: 4⅓ times GDP. - Foreign bank branches: 1.7 times GDP. - Aggregate CAR: 19.9 percent (base) → 16 percent (adverse). - CET1 decline: 329 bps to 14.5 percent over three years. - Estimated recapitalization needs for pressured small banks: about 0.14 percent of GDP. - LCR: 190 percent (base) → 121 percent (combined shock). - Cash-flow liquidity shortfall (3-month horizon): about 2.8 percent of GDP. ### Risk Analysis Recommendations (Table 1 summary) - Solvency (CBM, near term): - Enhance data management for credit risk analysis to run stress tests timely and accurately. (¶59) - Move toward expected loss-based credit risk models for stress testing, particularly with IFRS 9. (¶59) - Conduct regular sensitivity tests for credit risk, including asset concentration at borrower and sector levels. (¶59) - Liquidity (CBM; CBM, MFSA; near term): - Strengthen liquidity stress-testing by differentiating resident and non-resident deposit shocks; conduct currency-specific and longer-horizon liquidity tests. (¶93) - Conduct regular sensitivity analysis on funding concentration. (¶93) - Enhance data management for liquidity stress-testing, cross-check consistency, and prepare for NSFR implementation. (¶93) - Interconnectedness (CBM, MFSA; medium term): - Expand interbank analysis based on large exposures and funding concentration to capture cross-border bilateral exposures. (¶116) - Enhance network analysis of interlinkages to fully incorporate multiple transmission channels. (¶116) - Integrate contagion risks into macro stress-testing frameworks to inform tool calibration. (¶116) *NT (near term) = 1–2 years; MT (medium term) = 3–5 years.* *Source: INTRODUCTION section of the Malta FSAP stress testing chapter (IMF).* ### 1. Malta’s financial system is relatively large and heavily connected with the world. The ### 1mltea2019009 - 1. Malta’s financial system is relatively large and heavily connected with the world. The ### Financial system size and composition - Main components of financial system assets (end-2017): - Banks: €48 billion in assets, 4.3 times GDP - Insurance: 1 time GDP - Investment funds: 1.1 times GDP - Private pension market: negligible in size. - Large residual category of “other financial institutions” (OFIs) comparable to other euro-area cross-border intermediators (e.g., Ireland, Luxembourg, Netherlands). - Cross-sectoral linkages: part of the banking sector, insurers and OFIs hold assets and liabilities mostly vis-à-vis the rest of the world. ### Bank groupings and business models - Authorities classify banks into three categories based on resident exposures and financial-stability impact: - Core domestic banks: - Total assets: €23 billion at end-2017 - Lend mostly domestically; attract household and corporate deposits - Account for 99 percent of mortgages to residents - Three largest core domestic banks are Significant Institutions (SIs) and hold 87 percent of this category’s assets (42 percent of total bank assets) - Three core domestic banks are foreign-owned - Debt securities (foreign and Maltese government) account for a quarter of their total assets - High claims on the Eurosystem reflecting excess liquidity - Two core domestic banks each own a large domestic insurer - Non-core domestic banks: - Total assets: €2.2 billion - Small, foreign-owned, funded primarily from wholesale markets and non-resident deposits; some exposure to residents - Business models vary: syndication, factoring and finance, private banking, conventional lending - Two banks account for about 80 percent of this category’s assets - International banks: - Total assets: €23 billion - Foreign-owned with insignificant domestic exposures - Rely mostly on wholesale (including intra-group) funding of relatively long maturities - Focus on group activities (custodian services, trade finance, investment banking) - Account for over 77 percent of the banking system’s total non-resident deposits and about 78 percent of lending to non-residents - Two non-EU branches account for 83 percent of this category’s assets (€19 billion) and have negligible exposure to the Maltese economy ### Capitalization, liquidity, profitability - Bank capital and liquidity: - Total capital adequacy ratio (2017): 21.2 percent of risk-weighted-assets (RWA) - Tier 1 Capital ratio (2017): 19.0 percent - Core domestic banks’ capital ratio improved from 16.8 percent in 2017Q4 to 17.2 percent in 2018Q1 (note applies to core banks) - Liquidity: ample; claims on the Eurosystem are high for some banks - Profitability: - Aggregate profitability supported by stable net interest margins and operating costs - For core domestic banks, margins are supported by increasing interest rate spreads, high by euro-area comparison - Risks to sustainability of profitability: continued contraction of corporate loan books, weakening property market, large exposures to low-yield bonds, increased regulatory compliance costs, higher loan loss provisioning requirements, and implementation of EBA guidelines on Minimum Required Eligible Liabilities which will likely raise funding costs - Implementation of IFRS 9 expected (per EBA) to reduce average CET1 for LSIs by 45 bps on average; Maltese authorities are assessing effects with auditing assistance ### Asset quality and nonperforming loans (NPLs) - NPL ratios (end-2017, excluding non-EU branches): - Total NPL ratio: 4.1 percent of total loans (down from 6.6 percent in 2014Q4) - Core domestic banks: 4.1 percent - Non-core domestic banks: 2.3 percent - International banks: 5.3 percent - NPLs of non-financial corporations (NFCs) excluding non-EU branches: declined from 11.8 percent in 2014Q4 to 9 percent in 2017 - Sectoral NPLs remain high: - Construction: 27.8 percent - Corporate real estate: 13.9 percent - Loan loss provisioning for total loans: - Core domestic and international banks: about 34 percent of NPLs - Non-core domestic banks: 57 percent of NPLs - Regulatory monitoring: authorities are monitoring NPL reduction plans in line with a 2016 regulation mandating all banks reduce their NPL ratio below 6 percent over five years. ### Funding structure and correspondent banking - Funding and liquidity characteristics: - Loan-to-deposit ratios generally low; liquidity high - Growing shares of non-resident deposits (especially in smaller banks) and high share of sight deposits raise concerns about funding stability under adverse shocks - Correspondent banking relationships: - Some banks’ correspondent banking relationships face pressures and restrictions, particularly when providing correspondent services and channeling flows from high-risk jurisdictions or dealing with high-risk clients (nonresidents, e-gaming, virtual-asset operators, participants in the Individual Investor Program (IIP), politically exposed persons) - Pressures stem from profitability (low transaction volumes, high compliance costs), risk appetite, and reputational risk ### Sovereign exposure - Banks’ holdings of government debt (2017): - Banks held 29 percent of total government debt - This equates to 3.3 percent in terms of assets share - Over 90 percent of these holdings are concentrated in core banks (6.6 percent of their assets) ### Credit developments and cyclical indicators - Credit-to-GDP gap: negative for several years, reflecting broad-based slowdown in credit growth after the Global Financial Crisis - Main contributor to negative gap: large decrease in share of NFC loans to GDP - Recent strong economic growth has helped improve bank asset quality and maintain banking sector capital above minimum requirements ### Intercompany lending and non-bank intermediation - Shift from bank intermediation to intercompany lending for NFCs driven by: - Higher bank lending rate in Malta compared to euro area (by 160 basis points on average) - Highly collateralized bank loans in Malta, making bank financing costly for corporations - Strong firm profitability and formation of larger groups enabling internal financing (intra-group and related-group lending) - Structural shift toward less capital-intensive sectors (services) reducing demand for bank loans - Systemic risk assessment: - Intercompany lending currently does not appear to pose systemic risk - Reliance on internal cash buffers may expose firms to business-cycle risks, but impact on banking sector is manageable because of declining bank exposure to NFCs - Limited use of collateral for intercompany lending limits spillovers to banks through fire sales ### Leverage, property lending, and housing market - Firm leverage: - NFC leverage remains high compared to pre-crisis average and European peers - Contributing factors: significant presence of SMEs, construction and real estate sectors with highest leverage, possible statistical issues from cross-border intra-group financing - Mitigating factors: relatively high profitability of Maltese firms, high degree of collateralization of bank loans, reliance on intercompany lending - Property-related lending: - Resident mortgage lending growth: 8.5 percent annually since 2013 - Property-based lending to residents (including credit to construction and real estate) accounts for 62 percent of core domestic banks’ total loans to residents - Core domestic banks’ share of property lending to total assets: 25 percent; all banks in stress test sample: 22 percent - Risk: susceptibility to a potential sharp decline in house prices - Residential housing prices (2010Q4 to 2017Q4): - Prices rose by about 33 percent - Drivers: strong employment growth, rising disposable income, influx of foreign workers, portfolio rebalancing toward property in a low interest rate environment, buoyant tourism, lower withholding tax on rental income, and the IIP - Construction investment has picked up in response - House price indices have gone beyond long-term trends and conventional price indicators have increased - House price-to-income ratio: around historical average as of end-2017 - House price-to-rent ratio: higher as of end-2017 - Econometric analyses suggest some overvaluation, with estimated size varying by house price index; authorities’ models point to small overvaluations at this juncture ### Household balance sheets and resilience - Household indebtedness: - Household debt at 108 percent of gross disposable income in 2017 (slightly above euro-area average) - Home ownership ratio: 82 percent (2017) versus euro-area average 66 percent (2016) - Share of homeowners with debt: about 20 percent - Mortgage characteristics: - Mortgages mainly at variable interest rates; loan-to-value (LTV) ratio of new mortgages: averaging 77 percent in 2017Q4 - Debt-service-to-income (DSTI) ratio of new mortgages: averaging 21 percent in 2017Q4 - Ratios became more prudent in 2017 for secondary residence and buy-to-let loans - Financial wealth: - Household financial wealth has increased; debt-to-financial-wealth ratio stable at 23 percent - Stress considerations: - An adverse scenario (falling incomes, rising interest rates) could rapidly deteriorate household debt service capacity, particularly for low-income households - Overall analysis indicates household balance sheets are resilient against adverse economic shocks (per the note’s last section) *Prepared by IMF staff as part of the FSAP mission (note reflects discussions and recommendations of the FSAP mission as of September 2018).* ### 12. Reputational risks, including from money laundering and terrorism financing, and ### 1mltea2019009 - 12. Reputational risks, including from money laundering and terrorism financing, and ### Reputational and tax-competitiveness risks - Recent high-profile money laundering and terrorism financing–related incidents in Malta, including one linked to an LSI. - Large remote gaming sector and Government aspiration to host fintech-related industries increase challenges for banks entering relationships with these sectors: - Containing regulatory compliance costs and financial integrity risks is crucial to maintaining correspondent banking relationships. - Shocks to Malta’s reputation, potential changes in the country’s comparative tax advantage, and high exposure to non-resident deposits of international and non-core domestic banks could result in: - Capital flight and relocation of companies away from Malta. ### Brexit impact (sectoral implications) - Impact of Brexit is highly uncertain but assumed to mostly affect insurers and investment funds with foreign-oriented business models. - Under a “hard” Brexit scenario: - About 10 insurers who write business in the U.K. exclusively and two insurers with more than 50 percent of their business in the U.K. may decide to cease operations in Malta and establish an entity in the U.K. - U.K. insurers have been writing a substantial volume of health insurance policies in Malta via the EU’s passporting of services; if the U.K. is considered a third country, these insurers would either need to operate in Malta via a third country branch or cease carrying on business in Malta, which may affect consumer choices. - Banking sector exposure to the U.K.: - Core domestic banks have about 8 percent of their total assets in the U.K., mostly in loans to corporates. - Two core banks have relatively high share of assets, accounting for more than 15 percent of their total assets; these two are not expected to be affected by Brexit because they are part of financial groups based in the U.K. - On the funding side, the core banks have minimal amounts of deposits from the U.K. ### Key macrofinancial risks (RAM summary) - The Risk Assessment Matrix (RAM) identifies key macrofinancial risks from global and Malta-specific factors: - Weak growth in advanced economies. - Tighter and more volatile global financial conditions. - Policy and geo-political uncertainties. - A possible sharp correction in house prices. - Changes in international corporate taxation. - Growing reputational risks. - Rising regulatory compliance costs. ### Purpose and nature of FSAP risk and vulnerability analysis - Objective: assess capacity of the financial system to withstand severe but plausible macroeconomic shocks. - Components: - Solvency and liquidity stress tests for the banking system based on macro scenarios and sensitivity analysis. - Interconnectedness analysis in domestic and cross-border markets. - Caveats: - FSAP stress tests are macroprudential (system-wide) and differ from supervisory (microprudential) stress tests; results are not intended to lead to supervisory actions at the level of individual banks. - Credit loss estimates and solvency projections in the “adverse scenario” are subject to data and methodological limitations. - Adverse scenarios are not macroeconomic “forecasts” but represent tail events. ### Stress test coverage and methodologies - Stress test types: - Solvency tests (scenario-based and single-factor sensitivity) over a three-year horizon from 2017Q4 to 2020Q4. - Liquidity tests using regulatory approaches (LCR and NSFR) and a cash-flow based approach with maturity ladder and supervisory data. - Contagion analysis covering cross-border interbank exposures and domestic cross-sectoral interlinkages between banks, insurers, and funds. - Data sources and reference dates: - Main sources: EBA ITS templates (FINREP and COREP) with end-2017 as a starting point. - Cash-flow based liquidity stress test maturity profiles based on end-March 2018. - Supplementary CBM historical data from Banking Rule 06 for time-to-repricing gap buckets and resident vs non-resident deposits. - Public data sources: Bloomberg, Fitch, Haver Analytics, IMF’s World Economic Outlook database, and Moody’s KMV. - Collaboration: - FSAP team worked closely with CBM and MFSA and accessed supervisory data in an ECB secure data room. - Differences vs other institutions: - FSAP used different credit risk models, a different sample of banks, and different macro scenarios and parameters compared with EBA and ECB, though tests were carried out in close cooperation with the ECB. ### Solvency stress tests — sample and regulatory assumptions - Sample: - Top-down solvency stress tests focused on 11 sampled banks, covering 93 percent of banking sector assets (excluding foreign branches). - Sample includes all 6 core domestic banks (including all three SIs), two non-core domestic banks, and three international banks. - Sample accounts for 93 percent of banking system total assets (excluding foreign branches), 99 percent of credit to residents, and 99 percent of resident deposits (excluding foreign branches). - Foreign branches excluded due to negligible exposure to Maltese economy, though reputational channel remains. - Capital framework and hurdle rates: - Tests based on minimum capital ratios under Pillar I of Basel II; Pillar II requirements not taken into account. - Banks allowed to deplete the capital conservation buffer of 2.5 percent under the adverse scenario. - For SIs, the hurdle rate includes the Other Systemically Important Institutions (O-SII) buffer under baseline and adverse scenarios; the O-SII buffer ranges from 0.5 percent to 2.0 in 2020, with phase-in during 2018 to 2020 taken into account. - Hurdle rates (Adverse scenario): - Total capital ratio (total capital to RWAs): 8 + O-SII buffer for O-SII; 8 for Other banks (Non-O-SII). - Tier I capital ratio (Tier 1 capital to RWAs): 6 + O-SII buffer for O-SII; 6 for Other banks. - Common Equity Tier I capital ratio (CET1 capital to RWAs): 4.5 + O-SII buffer for O-SII; 4.5 for Other banks. - Leverage ratio (Tier 1 capital to total assets): 3 for both O-SII and Other banks. ### Macroeconomic scenarios used in solvency testing - Time horizon: three-year period from 2018 to 2020. - Baseline scenario: - Aligned with April 2018 World Economic Outlook projections. - Assumes slightly declining GDP growth converging to potential, unemployment below 5 percent, and modest pickup in inflation. - Adverse scenario: - Developed using IMF Flexible System of Global Models and calibrated with CBM and ECB. - Features a V-shaped GDP profile reflecting Malta’s high trade diversification and flexible labor markets. - Scenario drivers: combination of external and domestic shocks leading to decline in exports, capital outflows, shrinkage in internationally-oriented financial services and remote gaming sectors, declining asset prices, and sharp GDP contraction. - Resulting average decline in GDP of 0.1 percent per annum over the 3-year horizon, specified as: - -2.3 percent in 2018, - -1.0 percent in 2019, - 2.9 percent in 2020. - Leads to 15.4 percentage points cumulative decline of GDP relative to the baseline (equivalent to a 2.07 standard deviation shock). - Adverse scenario reflects shocks plausible but notably more severe compared to past crisis periods in Malta. - Malta-specific shocks embedded in the adverse scenario: - Possible changes in international corporate taxation and/or growing reputational risks and regulatory compliance costs could weaken Malta’s attractiveness as a financial and business location, leading to shrinkage of the international sector and erosion of the tax base. - Financial sector (including ancillary professions) generates an estimated 10 percent gross value added to the economy (MFSA 2017). - Gaming sector’s share in economic value-added stands at 11 percent of GDP (Malta Gaming Authority 2018). - Sharp correction in historically high housing prices: - Exogenous assumption: house prices decline by 30 percent in cumulative terms over the 3-year stress horizon. - This triggers wealth effects, negative feedback to domestic demand and deflation, and further deterioration of NPLs. ### Adverse scenario transmission to financial system - Recession and tighter financial conditions expected to: - Spike interest rates and bond yields. - Increase debt service and refinancing risks for households and corporates, leading to higher NPLs. - Cause large credit and market losses for banks given impaired borrower repayment capacity and holdings of debt securities. - Erode bank profits and capital adequacy ratios, potentially causing wide-spread distress and tightening lending conditions. - Materialize liquidity risk with deposit funding outflows of retail and wholesale deposits, particularly from non-residents. ### Solvency stress test methodological scope - Risks captured: credit risk, sovereign risk, market risk (including foreign exchange and interest rate risk), and interest rate risk in the banking book. - Accounting basis: solvency stress test based on International Accounting Standard 39 principles, particularly for provisioning approach. - Projection and transmission: balance sheet and income projections used to quantify impacts under baseline and adverse scenarios. *Source: IMF staff technical note excerpt (Malta FSAP). ### 29. The balance sheet projection followed a quasi-static approach, which is the standard ### 29. The balance sheet projection followed a quasi-static approach, which is the standard ### Methodology: balance sheet and RWA projections - Balance sheets grow in line with nominal GDP growth under baseline and adverse scenarios; the size of the banking system in terms of assets remained constant relative to the size of the economy. - Balance sheet growth measured for each individual bank using the weighted average GDP growth of all countries where the bank has significant exposure of assets. - Other factors affecting balance sheet growth: revaluation of foreign currency exposures in accordance with applicable foreign exchange movement and triggered off-balance sheet items (i.e., credit lines and guarantees). - The structure and composition of assets and liabilities assumed unchanged across the projection period. - Additional funding could be needed to cover the gap between assets and equity. - Model caps bank deleveraging at zero percent under adverse scenario, i.e., balance sheets do not contract in nominal terms. - Triggered portion of off-balance sheet items: 5 percent for the baseline and 10 percent for the adverse scenario. - RWA projections accounted for: (i) new provisions for credit losses and exchange rate movements for foreign currency exposures; (ii) additional risk weights on new defaulted loans; and (iii) a triggered portion of off-balance sheet items (undisbursed credit lines and guarantees). - Risk weight for new defaulted loans topped up to 120 percent of net amount (net of specific provisions), based on historical RWAs of defaulted exposures; consistent with Basel II standardized approach where defaulted exposures are risk-weighted between 100 and 150 percent. - Income (profit and loss) projection: - Most non-interest income items projected to follow nominal GDP growth (includes net fee and commission income, operational and administrative expenses). - Under the adverse scenario a floor of zero percent set for growth in non-interest income and expenses. - Extraordinary income and losses assumed not to recur. - Income tax reflected using historical tax rate in the system with a floor set at 28 percent (the median tax rate of the sampled banks). - Interest income projected using base and interest rate risk components incorporating interest rate risk in the banking book. - Loss provision projected using the credit satellite model; valuation losses of debt securities (trading and available-for-sale) calculated using modified duration approaches. - Loan loss provision is a final loss, identical to a write-off. - Distribution of after-tax profit assumed only if bank net profit positive; dividend payout ratio set at historical dividend payment of each bank. - Maturing Tier 2 capital instruments during the three-year horizon assumed not to be renewed. - Static balance sheet approach used for comparison: assumes bank balance sheets remain at end-2017 levels; maturing assets replaced by exposures of same type and risk; income projection follows same risk factors including accrued income for non-defaulted assets. ### Credit risk: composition, modelling, and provisioning - Credit risk accounts for 90 percent of total RWA in the sample banks. - Asset composition (percent): Loans, net of provision 71.2; Debt securities 23.4; Non-interest earning assets 5.4; Other assets 0.1. - Loan composition by sector (percent): Central banks 21.9; General governments 0.3; Credit institutions 13.6; Other financial institutions 5.2; Non-financial corporations 29.9; Residential mortgage loans 23.7; Credit for consumption 3.3; Other household loans 2.1. - Loans by currencies (percent): Domestic 90.0; Foreign 10.0. Loans in foreign currencies dominated by U.S. dollar (6 percent) and British pounds (3 percent). - Variety of bank business models led to different loan exposures: mortgages about 99 percent provided by core domestic banks; placements with central bank about 23 percent of total loans for both core domestic and other banks. - Satellite models used to estimate credit losses: - NPL ratios projected using fixed-effect panel regression techniques for nonfinancial corporation and mortgages exposure classes. - Loans to financial institutions proxied using Moody’s Expected Default Frequency (EDF) series and projected via fixed-effect panel regression. - Consumer loans (non-mortgages) credit risk estimated by benchmark PDs and LGDs provided by the ECB due to unavailability of consistent NPL data. - Sovereign exposures and HTM securities stressed by migrating exposures by three notches down in the adverse scenario and applying corresponding long-term PDs and LGDs from Moody’s. - Provisioning ratios for the NPL-based approach calculated based on collateral coverage for each type of loan, with a floor of provisioning coverage ratio set at 65 percent; under the adverse scenario, a minimum of 65 percent coverage applied. - For PD-model approach provision expense calculated as percentage of average balance of performing loans (provision expense = PD × LGD × average balance). For loans to financial institutions, a minimum 50 percent LGD floor imposed. - Starting points for default rates constructed from historical ratio of stock of exposures in default to total exposures; flow of new defaults calculated from changes in stock incorporating write-offs and expert judgment. ### Market risk and valuation - Market risks assessed from interest rate and exchange rate movements; transmission via net open positions in foreign currencies and valuation losses for debt securities (held for trading and available-for-sale). - Direct impact from foreign exchange movements minimal: median of absolute net open positions to total capital was 0.12 percent. - Market valuation losses for debt securities estimated using modified duration approach: - Sovereign yield curves constructed by linear interpolation of short- and long-term interest rates as specified in macro scenarios. - Average duration of debt portfolio calculated for each bank. - Losses = size of each bond portfolio × average duration × changes in yields. - For non-sovereign securities yield change included credit spread along three-year horizon. - Credit spread determined using ITraxx data (EUR 5 yr, iTraxx cross-over, European Senior Financial Index); average credit spread 190 bps. - For debt securities issued in Malta, adverse scenario shock assumed of 120 basis points. ### Interest rate risks in the banking book - Assessed using time-to-repricing gap analysis with buckets: <1 month, 1–2 months, 2–3 months, 3–6 months, and 6–12 months. - Impact on net interest income estimated by measuring gap between interest sensitive assets and liabilities in each bucket. - Effective liability (funding) interest rates estimated using satellite models with fixed-effect panel regressions; effective interest rate of liabilities sensitive to 3-month Euribor and U.S. dollar Libor. - Under adverse scenario the relationship between funding and lending rates estimated using quantile regression implying tightening net interest margin. ### Results of the scenario-based solvency stress test (quasi-static approach) - Under baseline scenario (quasi-static): aggregate total capital ratio stabilized at about 19.4 percent and leverage ratio (Tier 1 to total assets) at 9.1 percent during the three-year stress horizon. No bank faced difficulties fulfilling minimum capital and leverage requirements. - Under adverse scenario (quasi-static): - Aggregate CET1 ratio would decline by 329 basis points to 14.5 percent in 2020. - Core domestic banks: CET1 decline 270 bps to 12.4 percent. - Other banks: CET1 decline 564 bps resulting in 24.3 percent. - One small bank would see its CET1 ratio fall below minimum regulatory requirement of 4.5 percent in 2020, by an amount equal to 0.01 percent of GDP. - Aggregate Tier 1 relative to RWA declines by 329 bps to 14.5 percent; core domestic decline 271 bps; other banks decline 564 bps. - Two small banks would see Tier 1 ratio decline below regulatory threshold of 6 percent; total recapitalization needs 0.13 percent of GDP. - Aggregate total capital adequacy ratio (CAR) declined by 390 bps to 16.0 percent; core domestic decline 341 bps to 13.9 percent; other banks decline 580 bps to 26.0 percent. - Two small banks would see CAR decline below regulatory threshold of 8 percent; total recapitalization needs 0.14 percent of GDP. - Three banks could not meet the regulatory minimum for one or more of the three capital ratios; total recapitalization needs relative to GDP still manageable. - Aggregate leverage ratio would decline by 129 bps from 8.9 to 7.7 percent. No banks below 3 percent leverage threshold; one bank very close to threshold. - Comparison: static approach (balances at end-2017) results reported separately (table omitted here) — e.g., average CET1 ratio in 2020 under static baseline 21.4 percent; under static adverse 15.5 percent. - Drivers of adverse scenario impact on capital: - Credit risk (loan loss provisioning) largest driver: accounts for a 506 bps decline in the CET1 ratio over the 3-year horizon. - Corporate loan NPL increases mainly explained by GDP growth in respective geographical areas of exposures; oil price declines and home price declines used to represent country-specific shocks for some banks. - For five banks concentrated domestically, GDP decline and assumed decline in housing prices were significant contributors to corporate NPL increases. - Mortgage loan credit risk sensitive to housing prices, unemployment rate, and wage growth. - Additional RWA from newly classified NPLs, loan growth, and triggered off-balance sheet items reduce CET1 by 186 bps during the three-year horizon. - Valuation losses on securities portfolios reduce CET1 by 107 bps during the stress horizon. - Pre-provision net revenue (aggregate NII, non-interest income, and non-interest expenses) increases aggregate CET1 by 474 bps relative to the starting point. *Source: IMF staff calculations as presented in the IMF chapter content.* ### 50. Under the static-balance sheet approach, the solvency stress test results improved, with ### 1mltea2019009 - 50. Under the static-balance sheet approach, the solvency stress test results improved, with ### Solvency stress test (static-balance sheet approach) - Aggregate capital ratio changes: - CET1 and T1 capital ratios declined by 229 bps (from 17.8 percent to 15.5 percent). - Total capital ratio declined by 286 bps (from 19.9 percent to 17.0 percent). - Bank-level outcomes: - One bank would see its total capital ratio fall below the minimum regulatory requirement of 8.0 percent. - Total recapitalization needs at 0.02 percent of GDP. - Leverage: - Leverage ratio of the banking system would decline from 8.9 percent to 8.2 percent, with no bank breaching the threshold. - Explanation of difference with quasi-static approach: - Difference mostly due to lower credit provisions, valuation losses, and RWA under the static assumption of gross loan and debt securities. ### Sensitivity analysis (single-factor shocks and concentration) - Scope: - Sensitivity tests covered concentration of assets and interest rate shocks; run separately from scenario-based stress test; results refer to banks' starting position as of end 2017. - Concentration single-largest-exposure test: - At the aggregate level, the single largest exposure accounts for 15 percent of CET1. - In three sample banks, the single largest exposure accounts for more than 25 percent of CET1 – net of credit risk mitigants and exemptions. - Simultaneous default of five largest borrowers (net exposures): - Zero-recovery rate: - Aggregate CET1 ratio decline by 7.1 percentage points – from 17.8 percent to 10.6 percent. - Four banks would see CET1 decrease below the 4.5 percent threshold. - Provisioning rate of 65 percent: - Aggregate CET1 ratio decline by 4.5 percentage points – from 17.8 percent to 13.3 percent. - One bank would see CET1 decrease below the 4.5 percent threshold. - Sector concentration: - For some core domestic banks, largest borrowers concentrated in construction and real estate. - Three core domestic banks have large borrowers in construction and real estate. - Total exposure to these large borrowers ranges from 28 percent to 60 percent of the bank CET1. - Interest rate “parallel yield shock”: - Parallel interest rate increase by 200 bps would increase aggregate CET1 ratio by 0.6 percentage point. - Four banks would see CET1 ratio decrease (impact minimal). - Average positive direct impact due to positive repricing gap reflecting dominance of loans with variable rate, including mortgages. - Funding shock (single-factor): - Funding cost shock of 100 bps translating to increased funding cost for liabilities with repricing profiles less than one year: - At aggregate level, CET1 ratio would decline by 0.9 percentage point. - No sample bank would breach the minimum CET1 ratio threshold. ### Summary of solvency stress-test results and vulnerabilities - Under the adverse scenario: - Aggregate total capital ratio would decline by 390 basis points – from 19.9 percent to 16.0 percent. - CET1 ratio would decline by 330 bps to 14.5 percent. - Main contributors: additional loan-loss provisioning and RWA; valuation losses on debt securities. - Bank-level vulnerabilities: - Three small banks would not meet at least one minimum regulatory capital requirement. - Recapitalization needs for those banks: 0.14 percent of GDP. - Vulnerabilities partly due to weaker starting point capital ratios and lower quality corporate loan portfolios, particularly high exposure to property lending. - Asset concentration risk: - Default of the five largest exposures would result in failure of three to five banks, depending on provisioning ratio used. - Some core domestic banks have borrowers concentrated in construction and real estate, sectors with historically relatively high shares of NPLs. ### Policy recommendations (mission) - Conduct regular sensitivity tests for assets concentration: - Monitor name concentration and sector concentration periodically, including sensitivity tests. - Consider concentration risk when determining Pillar 2 capital buffers during the Supervisory Review and Evaluation Process. - Enhance data management for credit risk analysis: - Strengthen data management to run stress tests timely and accurately, including credit risk data. - Note: recent establishment of a Central Credit Register is welcome. - Enhance CBM’s macroprudential stress tests: - Current coverage is appropriately based on financial stability considerations and includes all core domestic and non-core domestic banks. - Develop expected loss-based credit risk models to improve stress testing framework, particularly with the introduction of IFRS 9. ### Liquidity stress tests — introduction and main findings - Approaches used: - LCR, NSFR, and cash-flow based approaches. - LCR measures bank ability to meet short-term liquidity needs in a 30-day stress scenario using HQLA. - NSFR gauges structural longer-term refinancing and funding risks. - Cash-flow approach uses maturity buckets to analyze counterbalancing capacity to cover net-cash outflows. - Data sources and thresholds: - LCR and NSFR based on COREP reports for end-2017; cash-flow based on COREP for end-March 2018. - LCR-based stress test threshold: 80 percent (minimum regulatory LCR as of end-2017). - NSFR-based stress test threshold: 100 percent (forward-looking exercise). - Cash-flow test threshold: counterbalancing capacity (negative amounts indicate bank failure). - Funding and liquidity structure starting point: - System-wide loan-to-deposits ratio: 41 percent. - Core domestic banks loan-to-deposits: 58 percent; other banks: 39 percent. - Excess liquidity placements in central bank and debt securities account for 46 percent of total deposits of sample banks. - Funding dominated by household (retail) deposits; most retail deposits reported as deposits with maturity less than 30 days. - Increasing trend of non-resident deposits over last five years. - HQLA and LCR observations: - Retail funding accounts for 50 percent of core domestic bank total funding; for other banks retail deposits account for 19 percent. - Other banks group-wide LCR at 201 percent; core domestic banks at 188 percent. - HQLA level composition: HQLA level 1 comprised 88 percent of total HQLA. - Placement in the central bank and holdings of sovereign securities account for 46 percent and 32 percent of total HQLA, respectively. - LCR-based stress test scenarios (four): - Baseline (B): standard LCR CRR parameters at aggregate currency level. - Scenario S1: higher run-off rates for retail and wholesale deposits; banks can use liquid assets with CRR haircuts. - Scenario S2: S1 plus declining market value in liquid assets (market value decline derived from solvency stress tests and ECB valuation haircut for repo use). - Scenario S3: S2 plus higher withdrawals of non-resident deposits (less stable); additional haircuts from market value declines and ECB haircut. - LCR-based stress test results: - Baseline: all sample banks meet the 80 percent threshold at end-2017. - Aggregate LCR stood at 190 percent; median around same number. - One bank has LCR below 100 percent but still higher than 80 percent. - Adverse scenarios: - S1 (retail & wholesale outflows): aggregate LCR declines to 146 percent. - S2 (plus market value decline): aggregate LCR declines to 142 percent. All sample banks meet 80 percent but three approach threshold. - S3 (plus non-resident outflows): aggregate LCR declines to 121 percent; three banks fail the test. - Total liquidity shortfall for the three banks: 0.7 percent of GDP. - Note: Using a minimum threshold of 100 percent (effective January 2018), the shortfall would be 1.6 percent of GDP. - Foreign currency LCR stress tests: - U.S. dollar: - Aggregate U.S. dollar-LCR declines from 84.1 percent to 56.5 percent in the adverse scenario with additional non-resident dollar outflows. - Out of seven banks in the sample, three banks struggle to meet foreign cash flow in all stress scenarios, including baseline. - British pound: - Aggregate British pound-LCR declines from 156 percent to 105 percent. - One of four banks in the sample struggles to meet the 80 percent minimum threshold for all scenarios, including baseline; funding in pounds amounts to 14 percent of total funding for this bank. *Source: IMF staff calculations and analysis as presented in the provided content.* ### 73. Two scenarios were considered in the NSFR-based test. The scenarios are consistent with ### 1mltea2019009 - 73. Two scenarios were considered in the NSFR-based test. The scenarios are consistent with ### NSFR-based test: scenarios and baseline results - Two scenarios considered: (i) baseline scenario, which assumes regulatory parameters; and (ii) an adverse scenario consistent with that in the LCR-based tests and solvency stress tests. The adverse scenario assesses bank capacity to maintain stable funding during a stressed macro financial setup. - Baseline scenario results (end-2017, sample banks, aggregate): - Aggregate available stable funding: €19.1 billion - Aggregate required stable funding: €11.9 billion - Aggregate NSFR: 160 percent (minimum requirement: 100 percent) - One core domestic bank had NSFR below threshold; total funding gap for this bank: €681 million - Funding composition under baseline: - Retail deposits and tier 1 capital account for 72 percent of total available stable funding - On the required stable funding side, loans account for 58 percent and HQLA assets account for 35 percent ### NSFR-based test: adverse scenario results and assumptions - Adverse scenario assumptions: 1. Higher rates for required stable funding for retail and wholesale deposits (aligned with higher run-off rates in LCR-based test) 2. Higher rates for required stable funding for loans due to increased credit risk 3. Market pressure reducing market value of liquid assets, implying need for more stable funding - Adverse scenario results (aggregate): - Aggregate NSFR declines to 130 percent - One additional bank would have NSFR below 100 percent threshold - Total stable funding gap for the two small banks: €992 million ( €311 million in addition to the baseline shortfall) - This shortfall is equivalent to 8.9 percent of GDP or 3.8 percent of total sample bank assets ### Cash Flow-Based Liquidity Stress Test: scope and data - Purpose: assess bank resilience to liquidity risk based on net cash balances after funding outflow shocks; net cash balance = cash positions + counterbalancing capacity + cash inflows - Data source and structure: - Based on maturity ladder in the COREP report of March 2018 - COREP template: contractual cash outflows and inflows in 21 maturity buckets (overnight to >5 years), classified by secured/unsecured and type of counterbalancing capacity - Key data observations: - About 71 percent of total outflows is below 30 days - Open maturity deposits hold about 66 percent of total outflows - Retail deposits are the largest contributor for short-term funding - For inflows, 74 percent of total inflows is concentrated in buckets longer than 30 days - More than 82 percent of total corporate and retail loans have residual maturities more than 12 months - Among total retail and wholesale deposits of the sample banks, 24 percent are non-resident deposits, mostly (85 percent) in current and savings accounts with open maturities - Counterbalancing capacity: - Includes bank liquid assets and committed lines from other counterparties - Banks can obtain cash by selling liquid assets or via repo operations with the central bank - ECB haircuts applied in addition to market price declines from solvency stress test - Total cash reserves in the central bank and marketable securities account for about 80 percent of total counterbalancing capacity ### Cash Flow-Based Liquidity Stress Test: parameters and historical precedents - General principles for outflow/inflow parameters: - Wholesale funding sources have higher run-off rates than retail - Secured funding sources have lower run-off rates than unsecured - Different run-offs for non-resident deposits to capture liquidity vulnerability - Inflows assumed at 100 percent of contractual inflows except for loans to retail and corporate customers (50 percent) - Historical extreme runs cited: - 1 week: 7.5 percent withdrawal from IndyMac (June 2008); 8 percent from Spanish banking system (1994); 11 percent from Saudi Arabian banking system (August 1990) - 10 days: 8.5 percent withdrawal from Washington Mutual (September 2008) - 12 days: 30 percent withdrawal from DSB Bank (2009) - 3 months: 25 percent withdrawal from Parex Bank (2008) - 9 months: 30 percent withdrawal from Argentinian banking system (2001) - 1 year: 57 percent withdrawal from Northern Rock (2007) - Market value decline assumption consistent with solvency-based stress testing and LCR-based haircut ### Cash Flow-Based Liquidity Stress Test: scenarios, horizons, and results - Time horizons used: one month and three months; additional 5-day severe test - Difference between one-month and three-month horizons: run-off of non-resident deposits - Scenario definitions: - Scenario 1: large withdrawals of retail and wholesale deposits; additional run-off for non-resident deposits; decline in market value of liquid assets; ECB haircut on top of market value decline - Scenario 2: same as Scenario 1 but excludes committed lines from intragroup or other counterparties from counterbalancing capacity (only cash, reserves in central bank, and debt securities included) - Results summary: - 1-month horizon, Scenario 1: - All banks except two would meet cumulative funding gaps - Total liquidity shortfall of the two banks: 0.7 percent of total sample bank assets, or 1.8 percent of GDP - 1-month horizon, Scenario 2: - No additional banks would fail; total shortfall of the two banks: 0.8 percent of total sample bank assets, or 1.9 percent of GDP - 3-month horizon, Scenario 1: - All banks except two would meet cumulative funding gaps - Total liquidity shortfall of the two banks: 1.1 percent of total sample bank assets, or 2.7 percent of total GDP - 3-month horizon, Scenario 2: - The same two banks fail; total liquidity shortfall increases to 1.2 percent of total sample bank assets, or 2.8 percent of GDP - 5-day horizon, Scenario 1: - Three banks fail the test; total shortfall of the three banks: 0.4 percent of total sample bank assets, or 0.9 percent of GDP - 5-day horizon, Scenario 2: - Total shortfall of the three banks: 0.4 percent of total funding assets, or 1.0 percent of GDP - Consistency with LCR-based test: - Cash flow-based test confirmed LCR-based stress test results: vulnerabilities arise from substantial shares of short-term deposits, including deposits with open maturity; banks relying on wholesale funding and non-resident deposits are vulnerable - Data quality caveats: - Reporting of resident vs non-resident deposits based on residency of depositors; domestic legal entities owned by non-resident counterparts may be classified as resident and hence excluded from non-resident deposits, potentially understating vulnerability - Authorities should regularly verify quality of bank reporting, cross-check with other reporting templates; NSFR implementation and maturity ladder reporting (since March 2018) increase need for verification ### Funding concentration sensitivity tests - Sensitivity to withdrawals of top five depositors (household and nonfinancial corporates); excludes deposits from credit institutions and intragroup funding - Aggregate LCR decline: from 190 percent to 146 percent due to withdrawal of the top five depositors - Three small banks see LCR fall below regulatory minimum of 80 percent - Total liquidity shortfall of the three banks: 3 percent of GDP - Sensitivity to 60 percent withdrawals of deposits from key sectors (arts, entertainment, and recreation including gaming; accommodation and food service; financial sector) - Aggregate LCR declines from 190 percent to 115 percent (a decline of 75 percentage points) but remains above 80 percent threshold - Three banks would have LCR below threshold, with liquidity shortfalls totaling 0.9 percent of GDP ### Summary findings - Most sample banks are resilient to short-term liquidity pressures at the aggregate level due to excess liquidity in the banking system - Vulnerabilities identified: - Reliance of some banks on wholesale deposits of residents and on retail and wholesale deposits of non-residents - Some small banks would fail LCR and cash flow-based stress tests under combined adverse events involving large withdrawals of wholesale and non-resident deposits - LCR-based stress test for foreign currency shows some banks fail even under the baseline scenario - NSFR-based test indicates most sample banks do not have structural longer-term liquidity risk, due to relatively high capital and share of retail deposits; high share of HQLA lowers required stable funding - A few small banks would struggle to meet NSFR due to high shares of mortgage and long-term corporate loans requiring high stable funding - Funding concentration risks: - Heavy reliance on funding from a few large depositors leads to high funding concentration risk; withdrawal of top five depositors would cause three banks to fall below 80 percent LCR; sectoral shocks similarly cause failures in three banks ### Policy recommendations - Strengthen liquidity stress test framework by incorporating additional features: - Different run-off rates for resident and non-resident deposits - Liquidity tests by currencies - Liquidity tests for longer time horizons (e.g., using maturity ladders) - Conduct regular sensitivity tests for vulnerabilities, including reliance on large depositors, to detect shifts in systemic risk and early identification of increases in deposit concentration risk - Enhance data management for liquidity stress-testing and prepare for NSFR implementation: - Regular checks on bank liquidity reports (LCR, NSFR, maturity ladder), including cross-checking with FINREP, COREP, and MFSA Banking Rule 06 - Authorities should ensure bank readiness for upcoming NSFR implementation, particularly for banks identified as vulnerable in NSFR-based stress test *Source: IMF staff calculations and analysis contained in the referenced chapter.* ### 94. Spillover and contagion risks are examined from both a cross-sectoral and cross- ### 1mltea2019009 - 94. Spillover and contagion risks are examined from both a cross-sectoral and cross-border perspective, based on supervisory and market information. ### Methodology and analytical framework - Network analysis uses the CoMap framework (Covi, Gorpe, and Kok 2019), extending Espinosa-Vega and Sole (2011) with granular supervisory data, bank- and exposure-specific parameters, and liquidity dynamics. - Complementary market-based analysis employs the Diebold-Yilmaz (2014) methodology deriving pair-wise net spillovers on equity returns; the market-based analysis covers only four Maltese banks listed on the stock exchange and may be affected by low market activity. - Balance sheet-based network analysis focuses on bilateral exposures, combining stylized credit shock simulations (one credit counterparty defaults at a time) and funding shock simulations (counterparty default inducing liquidity shortfalls). - Liquidity dynamics are incorporated by considering each entity’s liquidity surplus, liquidity constraint, and quality of assets to determine “fire sale” discount rates. - Counterparty-level exposure data: detailed supervisory data reported by 21 Maltese banks (6 core and 15 other) as of December 2018, allowing coverage of total 78 banks, of which 58 are significant banking counterparts domiciled outside Malta. - Cross-sectoral network adaptation: insurers and funds incorporated using a dataset provided by MFSA (as of June 2017); analysis focused on domestically-relevant insurers (8) and investment funds (8) and the entire banking sector (24), including foreign branches but excluding one bank with no exposures to financial institutions. ### Stylized facts — Cross-border interbank exposures - Other banks report large exposures equal to about 83 percent of their total assets; core banks’ large exposures constitute about 43 percent of total assets. - Interbank exposures represent 21 percent of total large exposures for core banks and 24 percent for other banks. - For core banks, the majority interbank credit exposures is domestic (10 percent); for other banks, exposures to non-Maltese euro area banks dominate (16 percent), followed by non-euro area banks (7 percent). - On the liabilities side, large funding sources constitute 10 percent of liabilities for core banks and 42 percent for other banks. - Of funding exposures, interbank linkages make up 45 percent for core banks and 44 percent for other banks. - Largest funding sources: core domestic banks receive 39 percent from within the euro area; other banks mostly rely on non-euro area banks (31 percent). - Aggregate interpretation: - Other banks have significantly higher concentration of large exposures and are likely to face more contagion risk than core banks. - Credit shocks dominate funding shocks as the main transmission channel of potential contagion. - Cross-border exposures are expected to play a critical role for other banks. - Total assets of other banks are about one-fourth of the total assets of core banks, moderating systemic importance. ### Appraisal — Cross-border contagion mapping and results - The topography network comprises 21 Maltese banks (red nodes) and 58 non-Maltese banks (blue nodes); node size measures degree of vulnerability and line thickness is proportional to relative size of large exposures with respect to capital. - Of the top 20 most contagious banks, 18 are non-Maltese, underscoring the importance of cross-border exposures. - Individual bank contagion index values are evenly dispersed around an average index of 2.6. - Hypothetical default of the most contagious bank (Bank 1) results in average losses to Maltese banks close to 5 percent of their capital buffers. - Bank 1 causes contagion to six Maltese banks, with individual losses ranging between 20 and 140 percent of their capital buffers; two of those six experience default leading to undercapitalization. - Total contagion defaults in these simulations amount to 12 defaults associated with losses in the default zone. - Vulnerability ranking of Maltese banks: - The most vulnerable bank (also labeled Bank 1) incurs average losses of about 13 percent of its capital buffer. - Of the 13 percent losses for this bank, nearly 9 percentage points are attributable to losses caused by euro area banks, with the remainder from other European banks. - The top three most vulnerable banks are exposed only to cross-border contagion. - Of the 12 contagion defaults, 9 are associated with the top three banks (which includes one core bank). - Remaining core banks are in the bottom half of the vulnerability distribution. - Risk channel dominance and dynamics: - Most contagion can be attributed to credit risk; liquidity positions generally help absorb funding risk for most banks. - Four banks have limited liquidity space and face distress due to funding risk while remaining sufficiently capitalized. - Amplification effects are limited: contagion defaults, when they occur, take place in the immediate round with no cascade defaults. - Overall cross-border conclusion: - Diffusion of contagion is highly concentrated and largely direct. - Most contagion is attributed to credit risk from non-Maltese euro area banks with no single bank acting as the key transmitter. ### Stylized facts — Domestic cross-sectoral network (banks, insurers, funds) - Total cross-sectoral network constructed from MFSA dataset (June 2017) covering 40 financial entities: 24 banks, 8 insurers, 8 funds. - Breakdown by instrument type shows an equally dominant role of equity holdings and deposits, indicating cross-ownership as a possible source of interconnectedness. - Insurers as a group have the largest exposure within this network, with almost half the share of total bilateral claims. - Core banks hold the overwhelming majority of total obligations, with a two-thirds share. - Network topography: nodes arranged with banks in outer circle and funds/insurers in inner circle; node size indicates vulnerability and line thickness indicates exposures normalized by loss absorption capacity (excess capital). - The highest concentration of contagion is on a fund with exposures mainly to two banks and an insurer. - Banks are the main source of contagion (red lines dominate in number and strength). - Funds sector shows significant but fewer large connections to banks and insurers (green lines). - Insurers are well interconnected mainly as receivers driven by exposures. - Majority of banks either have connections only within the banking sector or remain sparsely connected to the domestic system. ### Appraisal — Cross-sectoral contagion and vulnerability - Core banks, on average, cause contagion losses equaling almost 10 percent of the capital buffer of the domestic financial system. - Industry average contagion index value is 2.1. - Distribution among core banks is highly skewed, with a single bank as the key transmitter of contagion risk. - Funds and insurers have contagion indexes of 2.1 and 1.5, respectively; both sectors exhibit skewed distributions indicating concentration. - Only 6 entities (a mix of core banks, insurers, and funds) are above the industry average; the top three among them cause 14 out of the total 18 contagion defaults. - Insurers are the most vulnerable group: average “vulnerability index” of 8 percent in relation to their capital buffer, with individual indices evenly dispersed around the mean. - Funds have an average vulnerability of 7.8 percent, driven mainly by a single entity with a high index value. - Potential for cascade defaults exists: under two simulations an initial shock leads to further second-round contagion defaults. - Amplification observed: average ratio of about 1-to-6 (losses in subsequent rounds versus in the initial round). *Source: IMF staff summary of relevant sections of the provided chapter content.* ### 108. A deeper dive reveals potential for cascade effects and spillovers mainly from core ### 108. A deeper dive reveals potential for cascade effects and spillovers mainly from core ### Distress matrices and contagion findings - Shares of system-wide losses (in percent of total losses) across simulations: - The highest share, 38 percent, of total losses can be attributed to those caused by core banks toward insurers. - Losses from core banks toward funds account for 18 percent of total losses. - Insurers and other banks are exposed to contagion from funds to a lesser degree. - Insurers can induce losses to core banks as well as to other insurers. - Normalized losses (percent of capital buffers, average): - Core banks induce normalized losses of 55 percent to insurers, on average. - Core banks induce normalized losses of 25 percent to funds, on average. - Losses experienced by core and other banks can be absorbed by their capital buffers, resulting in very low average figures. - Default frequency (number of entities, across simulations): - Funds lead insurers, accounting for 8 out of 18 total defaults across simulations. - Insurers and other banks absorb the rest with a 6-to-4 split. - Other banks cause no contagion to the domestic system, both in terms of losses and defaults. - Overall assessment: - Insurers are the most affected group primarily because of their deposits and equity exposures, making them vulnerable to large-exposure counterparties through multiple channels. - High capital can help alleviate contagion risks to some financial sector players, but the domestic focus of insurers implies increased fragility towards domestic-born risks if tail events were to materialize. *Sources: ECB, MFSA, and IMF staff calculations.* ### Market-based interconnectedness analysis (Diebold-Yilmaz approach) - Methodology: - Uses daily equity returns for a selected set of listed banks and a vector autoregressive model (VAR) to estimate a spillover metric directionally defined between each pair in the network. - Metric is based on how much of the return variability for a given bank can be attributed to shocks to returns of every other bank in the network. - Key findings: - Three of the four Maltese banks appear on the periphery of the global network because of their relatively lower degree of connectivity (light and thin lines). - One Maltese bank is more centrally located, reflecting significant common underlying factors with other banks as captured in market pricing of its equities. - Euro area and other European banks account for a large portion of net inward spillovers to Maltese banks, consistent with the balance sheet-based analysis. *Sources: Bloomberg and IMF staff estimates.* ### Robustness analysis and sensitivity tests - Cross-border interbank network sensitivity (alternative default threshold using minimum CET1 requirement): - Change reduces average losses somewhat and is largely linear with limited change in ranking of banks and number of defaults. - Contagion indexes decline in the 15–20 percent range. - Vulnerability indexes decline in the 5–30 percent range with ranking changes limited to moving up/down by one place. - Overall, there are three less insolvency defaults across all simulations. - The number of illiquidity-driven defaults remains at three (unchanged). - Cross-sectoral domestic financial network sensitivity (less-conservative alternative assumptions): - Alternative assumptions used: - (i) excess capital based on minimum CET1 ratio for banks; - (ii) tier 1 capital for insurers; - (iii) a redemption shock of 20 percent as the default trigger for funds; - (iv) liquidity surplus for funds incorporates sovereign holdings and short-term debt securities (both with 50 percent risk weight). - Both contagion and vulnerability index averages decline by about 60 percent in the alternative scenario with less conservative assumptions. - Wide variation across individual entity indexes leads to several changes in ranking of entities. - Number of contagion defaults declines from 18 to 7. - Despite degree declines, main patterns at the group level remain the same: insurers remain vulnerable to contagion risk and core banks play a key role in perpetuating these risks. *Sources: Maltese authorities’ data and IMF staff estimates.* ### Caveats and limitations - Modeling limitations: - The CoMap framework incorporates liquidity dynamics and bank-specific parameter calibration but fails to account for market perceptions to exposures (e.g., additional losses due to common exposures). - Assuming a single solvency threshold that triggers an outright default may miss dynamics related to the ability of banks to react to depleting capital bases; such dynamics could decrease or increase contagion risk depending on bank characteristics and network topology. - Data limitations: - Analysis is based on a specific point in time (2017Q2 and 2017Q4). - Limited counterparty-level data on funding sources and lack of information on foreign counterpart balance sheets allow only partial analysis of outward spillovers from Maltese entities. - Adaptation of the framework for the domestic financial system, including insurers and funds, is subject to misestimations due to several assumptions made on non-bank financial entities. - Market-based analysis covered only four banks listed in the Malta Stock Exchange; limited coverage and low market activity imply significant data limitations and individual results should be interpreted with caution. - Mitigation: - Robustness tests were performed with less-conservative assumptions. ### Summary and policy implications - Synthesis of results: - Risk of contagion through interlinkages within the financial sector is currently higher and more widespread relative to cross-border interbank exposures. - Within the cross-border network, contagion index produces mild readings but evidence points to concentrated distresses in smaller banks. - Expanding analysis from interbank networks to include funds and insurers has implications for the health of financial institutions. - Analysis points to the vulnerability of insurers to contagion risk and the key role of core banks in perpetuating these risks. - Market-based analysis confirms Malta is a net recipient of cross-border spillovers, particularly from euro area and other European banks. - Recommendations: - Expand the scope of the interbank analysis based on large exposures and funding concentration to capture cross-border bilateral exposures. - Continue to enhance network analysis of interlinkages within the financial sector to fully incorporate risks due to various transmission channels. - Integrate contagion risks into the overall macroeconomic stress-testing framework driven by macro scenarios to inform the calibration of macroprudential tools. ### Household stress tests — key trends, data, and methodology - Key trends: - Mortgage lending has been growing at nearly 10 percent annually in recent years, resulting in a rising concentration of mortgages in bank loan portfolios. - Indebtedness of Maltese households stabilized at about 110 percent of gross disposable income in recent years, around 10 percentage points above the euro area average. - Residential property prices have been on an upward trend, exceeding the euro area average in recent years. - Strong employment growth, higher disposable incomes, a large influx of foreign workers, and portfolio rebalancing toward property investments have supported property prices and rents. - Combined with relatively high home ownership, Maltese household balance sheets have significant exposure to real estate market developments. - Data and methodology: - Maltese banks do not have internal models to estimate probabilities of default and loss given default in the housing loan segment. - The household stress test uses micro data from the 2017 Eurosystem Household Finance and Consumption Survey sample for Malta (1,004 households; data based on 2016) and updates income and expenditure using latest figures on wages, prices, and national accounts. - Projections for the stress period (2018–20) are based on the adverse scenario developed by the FSAP team, including estimates for mortgage lending rates and household consumption growth in the stress period. - Descriptive statistics and scenario parameters: - Initial data: 28 percent of households had debt; of these, about 59 percent receive some social transfers; 9 percent have negative extended financial margin. - Stress test scenario dynamics (2018–20, percent): - Output Growth: 2018 = -2.3, 2019 = -1.0, 2020 = 2.9 - Inflation: 2018 = -0.1, 2019 = -0.6, 2020 = -0.2 - Growth in expenditure: 2018 = 1.3, 2019 = 1.3, 2020 = 2.3 - Growth in wages: 2018 = -2.8, 2019 = -4.7, 2020 = -2.9 - Changes in Mortgage rate (in ppts): 2018 = 1.9, 2019 = 0.4, 2020 = -0.1 - Growth in house prices: 2018 = -18.2, 2019 = -14.1, 2020 = -0.4 - Growth in equities: 2018 = -28.6, 2019 = -1.8, 2020 = 8.3 - Definition of financial distress for households: - Extended financial margin (EFM) calculated with specified formula; households with negative EFM and negative net assets (NA) are deemed at risk of financial distress. - Net assets under stress: NA_t = real estate holdings_{t-1} × (1 + μ_t) + EFM_t − other liabilities (as per detailed formula). - Probability of default (PD_t) is set to 1 if EFM_t < 0 and NA_t < 0; otherwise PD_t = 0. - Initial assessment: - Analysis suggests Maltese households are resilient against shocks under the adverse scenario and systemic risk in the sector is unlikely at this juncture. - Prolonged economic shocks could impact the solvency risk of low-income households, reflecting their relatively thin financial buffers, but the overall impact appears manageable in terms of outstanding debt held by distressed households and their share of the population. *Appraisal of Risks — IMF staff calculations and Maltese authorities’ data.* ### 126. The distribution of household debt varies significantly across income deciles ### 126. The distribution of household debt varies significantly across income deciles ### Household debt distribution (pre-shock) - The share of households with debt is much lower among the three lowest income deciles (below 10 percent). - Beyond the 4th income decile, the proportion of households with debt broadly increases with income. - More than half of the households have debt in the top two income deciles. - In the sample, households in the top three income deciles did not default. ### Household distress and PD definition (pre-shock) - Pre-shock prevalence of household distress: 3.7 percent of households in distress (i.e., PD equaling 1). - PD (probability of distress) is defined in the text context as negative assets and negative financial margin. - Households with lower income have higher risks of financial distress; PDs are highest in the 1st, 2nd, and 3rd deciles. ### Household stress-test results and dynamics (adverse scenario) - Average implied PD for household loans increases from 3.7 to 8.9 percent over the stress period under the adverse scenario (reflecting a multiplier of 2.4). - Average PD ratio in 2018 (the first year of the stress period) increases to 5.8 percent. - More than 91 percent of indebted households have sufficient liquid assets to cover at least 12 months of income shortages (i.e., debt service payments and living expense larger than their income). - Second-year effect: noticeable jumps in PDs, averaging 10.4 percent (mostly driven by higher share of distressed households among low-income deciles). - As shocks dissipate in the final year of the stress period, PDs stabilize and average 10.5 percent for the whole sample. ### Credit-loss concentration and overall exposure - The debt burden of the bottom four deciles (households with highest PD and most affected by the adverse scenario) accounted for about 6.4 percent of total mortgage debt, or less than 6.3 percent of total household debt in the sample. - Implication: credit loss associated with distressed households would likely remain manageable given low share of total debt held by the most vulnerable deciles and strong initial household liquid buffers. ### Stress-test scope, methodology, and key assumptions (selected) - Stress-test horizon: 2017Q4–2020Q4 (3 years). - PD and distress measurement: PD equaling 1 corresponds to negative assets and negative financial margin. - Satellite models and proxies: - FSAP team’s own model and expert judgment for balance-sheet and credit growth, pre-impairment net income as sum of net interest income and non-interest income. - Credit risk for consumer loans (non-mortgage) estimated by benchmark PD and LGD provided by ECB due to unavailability of consistent NPL data series. - Behavioral/balance-sheet assumptions: - Quasi-static balance sheet: (i) balance sheet growth in line with nominal GDP growth with a floor at zero; (ii) RWA change due to newly non-performing loans, off-balance sheet items, and new loans; (iii) balance sheet composition remains constant; (iv) banks build capital only through retained earnings; (v) maturing capital instruments (AT1 and Tier 2) are not renewed. - For comparison, a static balance sheet assumption is also conducted. - Figure annotation: "Share of households with debt outstanding in 2016." and "Sources: IMF staff calculations." *Source: IMF staff calculations.* ### 5. Regulatory and ### 5. Regulatory and market-based standards and parameters ### Regulatory standards - Consistent with Basel III regulatory framework (LCR and NSFR). - Liquidity shortfall by bank. ### Reporting format for results (liquidity) - Output presentation: - Liquidity ratio or shortfall by groups of banks and aggregated (system wide). - Number of banks that can still meet their obligations. ### Financial Sector: Contagion Risk — Domain and assumptions - Exercise: Top-down by FSAP team. - Institutions included: - Cross-border: 21 Maltese banks and 58 non-Maltese banks. - Cross-sectoral (domestic): 24 banks, 8 insurers, and 8 funds. - Market-based global: 4 listed Maltese banks, 34 global banks. - Market share: - Cross-border: 99 percent of banking system (excluding branches). - Cross-sectoral: - 99 percent of total assets in the banking system (including foreign bank branches), - 35 percent of total assets of the insurance industry (100 percent of domestic insurers), - 17 percent of total net asset value of investment funds (100 percent of domestic funds). - Market-based global: 70 percent of banking system excluding branches (100 percent of market capitalization of listed banks). - Data and baseline date: - December 2017 for cross-border, June 2017 for cross-sectoral, 2005-2018 for market-based. - Sources: supervisory data (COREP: LE, AMM, LCR, FINREP: AE) for cross-border; proprietary MFSA dataset from various supervisory reports for cross-sectoral; Bloomberg for market-based. - Scope of consolidation: consolidated (sub-consolidated for subsidiaries) only within own sector. ### Channels of risk propagation and methodology - Cross-border and cross-sectoral: Espinosa-Vega and Sole Bank Network Model (2010) framework calibrated based on Covi, Gorpe, and Kok (2018) CoMap methodology. ### Tail shocks (contagion exercise) - Size of the shock: Pure contagion — hypothetical default of institutions. ### Reporting format for results (contagion) - Output presentation: - Number of undercapitalized institutions in distress. - Capital shortfall systemwide, by bank, and by group (contagion and vulnerability scores). - Amplification and cascade effects, and direction and size of spillovers within the network. - Net spillovers due to interconnectivity (market-based). ### Appendix III — Stress Test Macro Scenario Calibration (technical details and key calibrations) - Model: Flexible System of Global Models (FSGM), EUROMOD version calibrated to Malta’s macro conditions; Malta represented within “Other EA countries block.” - Country-specific shock to attractiveness as business and financial hub implemented as synchronized hits to financial and remote gaming sectors: - Contractions of -9.6 percent and -11.2 percent, respectively, corresponding to 95th percentile of yearly declines in their gross value added (1995-2017). - Combined impact on Malta’s GDP estimated at -2.4 percentage points (using sector-specific multipliers in Rapa (2017)). - Sovereign interest rate shock: - Deviation by 122 bps from the baseline in the second year (aligned between low-spread (core) and high-spread (periphery) euro area countries). - Corporate spread calibrations (exogenous, 95th percentile of historical annual changes using iTraxx data 2004-2017): - Investment grade corporate spreads widen by 80.8 bps relative to 2017 levels. - Sub-investment grade corporate spreads widen by 295.0 bps relative to 2017 levels. - Equity price projection: - MALTEX index 95th percentile annual decline (2004-2017) → 28.6 percent drop in the first year. - Temporal pattern: Prominent shock takes place in the first year, continues to worsen slightly into the second year, and eases somewhat in the third year. - House prices (authorities and FSAP adjustments): - Assumed cumulative decline of 30 percent over the 3-year stress horizon. - Calibration incorporated endogenous (model-driven) and exogenous components, resulting in a 1-to-4 proportion between exogenous correction and endogenous response. - Unemployment rate: - Malta-specific coefficients used to reflect weaker output–unemployment relationship. - Deviation of unemployment rate from baseline peaked at 1.5 percentage points in 2020 under the adverse scenario. ### Appendix IV — Solvency Stress Test: Credit Risk Estimations (methodology and main quantitative drivers) - Data: - Bank-by-bank historical data quarterly from 2004Q3 to 2017Q4 provided by CBM and MFSA. - Macroeconomic data from Bloomberg, Haver Analytics, IMF IFS, and WEO. - Corporate and mortgage loans: - NPL-based projections per bank using logistic (logit) transformation of NPL ratios to constrain predictions between 0 and 1. - Fixed-effect panel regression for baseline and adverse scenario: - Model: Y_k,j,t = α + υ_k + δ Y_k,j,t-1 + φ' X_t + ε_k,j,t (logit-transformed NPL as dependent variable). - For banks with significant exposures outside Malta, time-series regressions used (model: Y_t = α + δ Y_t-1 + φ' X_t + ε_t). - Projected logit NPLs transformed back to NPL ratios via NPL = 1/(1 + exp(-Y)). - Loans to credit and other financial institutions: - Moody’s Expected Default Frequencies (EDF) used for PD-based projections covering 10 countries. - PDs transformed with logit and projected via fixed-effect panel regressions; transformed back by PD = 1/(1 + exp(-Y)). - For Malta, absence of EDF for financial institutions: PD projection for loans to credit and financial institutions in Malta set as the median PD of euro area countries in the sample. - Main drivers and findings: - Corporate loans: NPL increases mainly driven by GDP growth in borrowers’ geographies; notable additional drivers in some cases: oil price (proxy for GDP) and foreign house prices where exposures were real-estate corporate lending. - For five domestically concentrated banks, decline in GDP growth and assumed decline in housing prices were significant contributors to corporate NPL increases. - Other contributors: interest rates (level or annual change) and sovereign spreads. - Mortgage loans: credit risk sensitive to housing prices, unemployment rate, and wage growth (income effects on household repayment capacity). - PDs for credit and financial institutions: sensitive to GDP growth in relevant geographies, interest rates (level or changes), and the U.S. stock market volatility index. ### Appendix IV — Empirical estimation outputs (selected summary statistics from tables) - Mortgage (logit NPL) estimation highlights (selected coefficients and stats): - L1 dependent variable coefficients: 0.7353***; 0.7966***; 0.5353***; 0.7969***; 0.6217*** (standard errors in parentheses as in source). - L5 Malta GDP growth: -1.0885*** (0.3312). - L1 USD Libor: 0.1054*** (0.0336). - Adjusted R-squared values: 0.94, 0.94, 0.84, 0.96, 0.90 across regressions. - Observations: 210, 48, 49, 44, 252 (per column). - PDs estimation (logit PD) highlights (selected coefficients and stats across geographical groups G1–G3): - L1 dependent variable coefficients: 0.8639***; 0.7704***; 0.6218***; 0.8619***; 0.7984***; 0.7182***. - Various lagged GDP growth coefficients significant (e.g., L3 G1 GDP growth -6.5896** (3.3092); L5 G2 GDP growth -4.6376** (1.8351); L0 G3 GDP growth (Contemporaneous) -5.0924** (2.1715)). - Volatility Index (Contemporaneous) coefficients: 0.0103***; 0.0122***; 0.0134*** (standard errors in parentheses as in source). - Adjusted R-squared values: 0.80, 0.75, 0.77, 0.84, 0.88, 0.77. - Observations: 196, 47, 47, 336, 55, 50 (per column). ### Appendix V — Solvency Stress Test: Funding and Lending Rate Estimation (methodology) - Bank funding costs: - Estimated using the effective funding rate per bank (ratio of interest expense to interest-bearing liabilities). - Data: quarterly from 2004Q3 to 2017Q4 provided by authorities. - Macroeconomic series for the adverse scenario follow the stress test scenario; macro data sourced from Bloomberg, Haver Analytics, IMF IFS, and WEO. *Source: 1mltea2019009 - 5. Regulatory and (IMF staff text and appendices as provided in the supplied content).* ### 2. The funding rate estimation was conducted using a fixed-effect panel regression. The ### 2. The funding rate estimation was conducted using a fixed-effect panel regression. ### Funding rate estimation methodology - Specification: - Y_{k,t} = α + ϑ_k + δ Y_{k,t−1} + φ′ X_t + ε_{k,t} - Y_{k,t} captures the funding rate for bank k at time t. - X_t is a vector of macro-financial variables specified in the stress testing scenarios. - Y_{k,t−1} is the vector of lagged dependent variables. - ϑ_k represents the bank specific fixed-effect term. - ε_{k,t} represents the independently and identically distributed error term. - Sample: 11 sample banks projected using the fixed-effect panel regression. ### Empirical findings on funding rates (Table 1: Malta: Funding Rate Equation) - L1 dependent variable: 0.4385*** (0.099) - L2 dependent variable: 0.1388* (0.074) - L2 3-month Euribor: 0.110*** (0.047) - L1 USD Libor: 0.168*** (0.036) - Constant: 0.0048*** (0.0009) - Adjusted R-squared: 0.76 - Observations: 509 - Note: Robust standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 - Interpretation: Funding rates were associated with short-term market funding rate movements. The most statistically significant factors were 3-months Euribor and U.S. dollar Libor, which represent the deposits structure of sample banks. ### Lending rate projection methodology and results (Table 2: Malta: Quantile Regression Equation) - Method: Quantile regression (tau = 0.9) — projections based on the 90th percentile to capture imperfect pass-through and shrinking net interest margins under the adverse scenario. - Regression (dependent variable: Change in Effective Interest Rate on Assets): - Change in Effective Interest Rate on Liabilities: 0.7134*** (0.036) - Constant: 0.0086*** (0.001) - Adjusted R-squared: 0.25 - Observations: 487 - Note: Robust standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1 - Sample: Interest rate on assets and liabilities of each sample bank sourced from FINREP from 2015Q3 to 2017Q4 (adjusted data sample). Assets and liabilities were limited to interest-bearing assets and liabilities. - Interpretation: A 1 unit change in effective interest rate on liabilities is associated with a 0.7134 change in effective interest rate on assets at the 90th percentile (imperfect pass-through). ### Liquidity stress test parameters (LCR, NSFR, cash-flow) - Sources: EU’s Capital Requirements Regulation and IMF staff calculations. - LCR-Based Test Parameters (selected entries shown as percent): - Central government assets: 100 100 95 95 - Regional government / local authorities assets: 100 100 95 95 - Public Sector Entity assets: 100 100 95 95 - Qualifying CIU shares/units (underlying is coins/banknotes and/or central bank exposure): 100 100 95 95 - Deposits subject to higher outflows: category 1: 15 15 15; category 2: 20 20 20 - Stable deposits: 11 11 11 - Other retail deposits: 21 21 21 - Operational deposits covered by DGS: 8 8 8; not covered by DGS: 41 41 41 - Deposits by financial customers: 100 100 100 - Deposits by other customers covered by DGS: 25 25 25; not covered by DGS: 51 51 51 - Run-off of non-resident deposits: Current deposits 00046; Saving deposits 00036; Time deposits 00033 - Note: To avoid double counting of outflows, additional withdrawals of non-resident deposits is based on the difference between these rates and other retail deposits. - NSFR-Based Test Parameters (percent), Baseline / Adverse examples: - ASF Factor 100percent: Tier 1 capital instruments 100 100; Tier 2 capital instruments 100 100 - ASF Factor 95percent: Stable retail deposits 95 89 - ASF Factor 95percent: Less stable retail deposits 90 79 - ASF Factor 50percent: Liabilities from secured lending and capital market driven, collateralized by HQLA 50 40 - RSF Factor 0percent: Cash; Exposure to central bank - RSF Factor 5percent: Unencumbered level 1 assets - excluding cash and central bank reserves 5 10 - RSF Factor 15percent: corporate bonds - extremely high liquidity and credit quality assets 15 20; loan to borrower of which are financial customers other than credit institutions 15 25 - RSF Factor 25percent: corporate bonds - high liquidity and credit quality assets 25 30 - RSF Factor 35percent: corporate bonds - other assets 35 40 - RSF Factor 50percent: loan to SME that qualify for the retail exposures 50 70; loan to the borrowers other than financial customers 50 60; loan to borrower of which are financial customers other than credit institutions 50 60 - RFS Factor 65percent: Non-renewable loans collateralized by commercial real estate 65 80; residential real estate 65 80 - RFS Factor 100percent: All assets unencumbered for a period of one year or more 100 100; Derivative receivables 100 100 - Cash-flow-Based Test Parameters (percent), selected run-off and roll-off rates: - LT Unsecured issuance: 50 50 - Repo across all collateral classes: 30-50 30-50 - Stable retail deposits: 6 7 6 - Unstable retail deposits: 13 15 13 - Operational corporate deposits: 30 35 30 - Non-operational corporate deposits and other: 50 55 50 - Deposits FI: 33 35 33 - Other Outflows: 100 100 100 - FX-swap in/outflows: 100 100 100 - Derivative in/outflows: 100 100 100 - Retail/Corporate inflows: 50 50 95 - FI Inflows: 100 100 100 - Additional run-off for non-resident deposits: Current 25 37 10; Saving 15 23 7; Term 11 17 6 - Counterbalancing capacity (selected): - Withdrawable central bank reserves: 100 100 100 - Level 1 tradable assets: 90-100 90-100 90-100 - Level 2A tradable assets: 85 85 85 - Level 2B tradable assets: 50-75 50-75 50-75 ### Balance sheet–based interconnectedness analysis — methodology - Framework: CoMap framework (Covi, Gorpe, and Kok, 2019), extending Espinosa-Vega and Sole (2011). - Key components: - Credit shock: losses from a subset of banks defaulting; exposure-specific loss-given-default rates λ_{ij}^k applied to claims e_{ij}^k. - Funding shock and liquidity dynamics: funding shortfalls summed across defaulting banks using bank-specific funding shortfall rate ρ_i; banks can offset shortfalls with liquidity surplus γ_i (HQLA in excess of net liquidity outflows). - Fire-sale dynamics: remaining liquidity shortage can lead to asset sales subject to fire-sale discount rate δ_i and constrained by asset pool θ_i. - Failure condition: bank i fails if capital surplus is insufficient to cover combined credit and fire-sale losses; insolvency condition expressed as inequality using λ, δ, ρ, γ, and θ parameters. - Illiquidity condition: θ_i < (1/(1−δ_i)) * iwee[0, ρ_i e_{ji} − γ_i] (as specified). - Simulation design: - Initial default of each bank is triggered in turn to study cascade effects and contagion path. - Exercise moves to subsequent rounds if additional failures occur and stops when no further failures. - Outputs are generated for each triggered default. ### Network output metrics (customized for Malta) - Contagion index (CCI_i): captures each bank’s potential contagion by taking a weighted average of losses of all other banks in percent of their capital; uses N_{ji} losses experienced by bank j due to triggered default of bank i. - Vulnerability index (VVI_i): gauges each bank’s degree of vulnerability averaged across all individual default events with identical probability; uses N_{ij} losses experienced by bank i due to triggered default of bank j. - Contagion default: counts the number of banks that experience severe distress associated with the triggered default of bank i (binary pass/fail for undercapitalization). - Default frequency: tally of simulations under which bank i falls below the capital distress threshold. - Amplification ratio (AMP(C)_i): compares losses induced in the initial round versus all successive rounds to measure amplification through cascading defaults; ratio greater than 1 indicates cascade effects dominate direct losses. - Similar AMP(V)_i measures susceptibility components for bank i. ### Data calibration and implementation (cross-border interbank network) - Scope: - All non-branch banks (22) in Malta report large exposures; one bank has no large exposures, reducing main sample to 21. - Sample expanded to incorporate significant foreign counterparties (58), increasing full network size to 79 banks (풵풵 = 79). - Focus is Malta-centric; analysis captures inward spillovers from foreign entities toward Maltese banks as well as domestic spillovers. - Calibration inputs and sources: - Large credit exposures (e_{ij}^k): COREP Large Exposures templates (C.27 and C.28). - Large funding exposures (e_{ji}): COREP Concentration of Funding templates (C.67.00.a) — top ten largest counterparties where obtained funding exceeds 1 percent of the bank’s total liabilities. - Loss-given-default rate (λ_{ij}^k): calibrated at exposure level using ratio of net exposures to bank capital after accounting for credit risk mitigation, collaterals, and exemptions to gross exposures (C.28 and C.28). - Funding shortfall rate (ρ_i): calibrated at bank level using the share of obligations with maturities shorter than 30 days; estimated using maturity information of largest funding sources (C.67.00.a) and large credit exposures (C.30). - Liquidity surplus (γ_i): stock of HQLA above net funding outflows (NLO) over a 30-day liquidity distress scenario (Liquidity Coverage Ratio templates C.72.00-C.76.00). - Liquidity constraint (θ_i): total amount of unencumbered non-central bank eligible assets (Asset Encumbrance templates F.32.01). - Fire-sale discount rate (δ_i): estimated for each bank based on portfolio of unencumbered non-central bank eligible assets weighted by relevant haircuts; haircut rates based on ECB guidelines with a lower bound haircut equal to highest haircut used for central bank eligible instruments. - Default/distress threshold (i_i^e): based on CET1 minimum requirement plus additional buffers (capital conservation buffer, O-SII buffer for significant institutions, countercyclical buffer). Conservative threshold considers depletion enough to penetrate capital buffers; less conservative threshold uses minimum CET1 only for robustness. - Calibrated model parameter summary (Table 1: Summary Statistics of Calibrated Model Parameters): - Loss-given-default rate (λ_{ij}^k): Weighted average 0.65; Min 0; Max 1 - Funding shortfall rate (ρ_i): Weighted average 0.41; Min 0.18; Max 0.95 - Fire-sale discount rate (δ_i): Weighted average 0.55; Min 0.46; Max 0.68 ### Bank-specific default thresholds (Table 2 excerpt) - Default thresholds expressed in percent of RWA: - Minimum CET1: 4.5 - Capital conservation buffer: 1.875 - O-SII buffer: Ranging from 0.375 to 1.5 for significant institutions; 0 for other banks - Countercyclical buffer: 0 *Italic: Source: 1mltea2019009 - 2. The funding rate estimation was conducted using a fixed-effect panel regression.* ### 6.75 to 7.875 for ### 1mltea2019009 - 6.75 to 7.875 for ### Cross-Sectoral Domestic Financial Network - 6.75 to 7.875 for significant institutions; - 6.375 for other banks - Note: O-SII = other systemically important institution - The contagion analysis of the cross-sectoral linkages in the domestic financial system relied mostly on a unique dataset constructed by MFSA as of June 2017. - Scope: - The dataset encompassed the entire banking sector, including foreign branches (but excluded one bank with no exposures to financial institutions scope). - Included domestically-relevant entities from the insurance and investment fund sectors. - Full sample of 40 financial entities comprising 24 banks, 8 insurers, and 8 funds. - Bilateral credit and funding exposures: - Counterparty-level and by-instrument interfirm exposure data constructed by MFSA combines multiple supervisory reporting templates from different financial industries. - Templates include MFSA Banking Rule 06 reports for banks, Investment Fund reports, and Solvency II Quantitative Reporting templates for insurers. - Security-by-security reporting was fundamental to establish debt and equity interlinkages. - The FSAP team used COREP Large Exposures templates (C.27 and C.28) to fill in the interbank loan exposures. - Resulting dataset comprises a 40-by-40 matrix for each instrument: loans, deposits, debt, and equity. - Loss-given-default rate (휆휆푖푖푗푗푘푘) for interbank lending: - Calibrated as the ratio of net exposures to gross exposures vis-à-vis each counterparty (C.27 and C.28). - As risk mitigation measures and collateral information wasn’t available for other exposures, instrument-specific assumptions were made (Table x3). - Funding shortfall rate (휌휌푖푖) for interbank exposures: - Calibrated at bank level using the share of obligations with maturities shorter than 30 days using granular bank reporting templates. - As detailed information wasn’t available for other exposures, instrument-specific assumptions were made (Annex Table VII.3). - Bank-level data: - Bank-level model inputs (liquidity surplus, liquidity constraint, fire-sale discount rate, and default threshold) have been calibrated as described for the cross-border network analysis in the previous section. - Insurer-level data: - Insurer liquid assets were considered the liquidity surplus while total investment assets remaining were considered the liquidity constraint. - The fire-sale discount rate was assumed as 0.7 as no granular data was provided on their portfolio composition. - For the default threshold, the solvency capital requirement was used as the benchmark, while Tier 1 capital as the less conservative alternative was used for robustness. - Fund-level data: - Baseline for liquidity surplus was limited to the amount of cash holdings. - Less conservative calculation additionally considered government securities and short-term debt securities (1 year or less) with a risk weight of 50 percent. - Liquidity constraint is equivalent to the sum of non-cash assets. - Fire-sale discount rate was calibrated based on each fund’s portfolio composition using different haircut assumptions for asset classes, resulting in a sample average of 0.7. - As there is no capital requirement for funds, large redemptions were considered as a conceptual equivalent of the severe distress scenario for funds. - Baseline calibrated based on a redemption of 5 percent relative to total net assets. - Less conservative alternative based on a 20 percent redemption shock to trigger distress to the fund. ### Model Assumptions and Sample Averages - Table 3. Malta: Model Assumptions and Sample Averages (instrument-specific calibrated values): - Interbank loans - Loss-given default: Calibrated 0.3 - Funding shortfall: Calibrated 0.5 - Other loans - Loss-given default: 0.6 - Funding shortfall: 0.3 - Debt securities - Loss-given default: 1 - Funding shortfall: 1 - Deposits - Loss-given default: 1 - Funding shortfall: 0 - Equity - Loss-given default: 0.8 - Funding shortfall: 0.5 ### Appendix VIII. Market-Based Interconnectedness Analysis – Technical Details - Methodology: - Follows the Diebold-Yilmaz (2014) approach with enhancements adapted for the 2018 EA FSAP (IMF 2018b). - Relies on estimating a vector autoregressive (VAR) model based on time-series data. - Spillover metrics derived via a forecast error variance decomposition. - Uses the elastic net estimator (Zou and Hastie 2005) to estimate the high-dimensional VAR, blending shrinkage and selection (lasso and ridge regression). - For the error variance decomposition, the Generalized Variance Decomposition (GVD) is applied (Koop, Pesaran, and Potter 1996; Pesaran and Shin 1998), which is invariant to the ordering of variables. - Estimated error variance decomposition quantifies the proportion of return variability (contemporaneous and H-steps ahead) of a particular bank i attributable to shocks to returns of another bank j, denoted 퐶퐶푖푖←푗푗퐻퐻, proxied as the spillover from bank j to bank i. - Net spillover measure for bank i vis-à-vis j equals the difference: 푖푖↔푗푗 = 퐶퐶푖푖→푗푗퐻퐻 − 퐶퐶푖푖←푗푗퐻퐻. - Data: - Analysis conducted at the individual bank level to capture linkages between Maltese banks and a selected sample of foreign banking counterparts across relevant geographies. - Spillovers derived using daily returns from equity price data. - Sample comprises banks with the following geographical decomposition: - 4 from Malta - 13 from the euro area (7 global systemically important banks, 6 significant institutions) - 8 from the rest of Europe (7 global systemically important banks and 1 other) - 8 from other advanced economies (4 global systemically important banks) - 5 from emerging markets - Underlying equity price data for these 38 banks spans the period 2005–18. *International Monetary Fund: content from the referenced PDF chapter/section provided above.* --- _Source: https://www.imf.org/-/media/files/publications/cr/2019/1mltea2019009.pdf_