## cr17309 ## Source details **Canonical URL:** [cr17309](https://www.imf.org/-/media/files/publications/cr/2017/cr17309.pdf) ## Other formats - [Markdown version](/-/media/files/publications/cr/2017/cr17309.pdf.md) - [Structured JSON version](/-/media/files/publications/cr/2017/cr17309.pdf.json) --- ### Executive summary — scope and coverage - Stress tests covered three major segments of the domestic financial sector, with emphasis on the largest financial institutions. - Banking: - The four largest banks together constitute over 75 percent of the banking system by assets. - Insurance: - Four largest life insurers and three non-life insurers hold 78 and 53 percent market share in terms of gross written premiums, respectively. - Investment funds: - Around 850 investment funds representing 95 percent of the industry’s total net assets were included. - Non-financial corporations were stress tested to validate corporate inputs used in the banking stress tests. ### Methodology and scenarios - Solvency stress tests implemented by FSAP team (TD), banks (BU) and authorities (Riksbank and FI). - BU tests followed EBA methodology; TD approaches included: - FSAP team: balance sheet approach, considered future changes to corporate risk weights. - Riksbank: simple solvency framework based on projection of credit losses. - Liquidity tests used LCR, NSFR and maturity ladder analysis, on aggregate basis and by currencies. - Contagion analysis covered domestic interbank exposures (implemented by the Riksbank). - Scenarios: - Baseline: IMF’s April 2016 World Economic Outlook projections for Sweden. - Stress: Sweden-specific risks and spillovers from a recession in the Nordic and Baltic region; mirrored EBA’s 2016 adverse scenario for first three years but included more severe interest rate and exchange rate shocks. - Single factor shocks: severe interest, market and concentration shocks and a shutdown of the foreign exchange swaps market. ### Key solvency stress-test findings (banks) - General result: banks would be resilient to a severe economic distress, but with differences across approaches. - IMF top-down (TD) results: - One bank would fall below the supervisory hurdle rate and no bank would breach the regulatory hurdle rate over the stress testing horizon. - System-wide CET1 ratio would fall by 2.5 percentage points in 2016 relative to the base year (or 2.5 percentage points relative to the baseline scenario in 2016) and by additional 1.5 percentage points in 2017–2018. - System-wide “all-in” leverage ratio would fall by 0.3 percent point in the downturn period (from 4.1 percent in 2015 to 3.8 percent in 2018). - Provisions for credit losses increased seven times by 2018 (comparing to four times in EBA test and twenty times in the authorities’ test). - Bottom-up (BU) results: - One bank would fall below the supervisory hurdle rate. - System-wide CET1 would fall by 2.3 percentage points during the first three years of the stress horizon. - Authorities’ TD results: - All banks would fall below the supervisory thresholds. - System-wide CET1 would fall by 4.9 percentage points during the period of downturn. - Drivers of capital deterioration (2016–2018): - Lower net interest income, mainly due to sharp increases in funding costs. - Higher provisions for credit losses, particularly on corporate exposures. - Increases in RWAs largely from higher PDs and depreciation of SEK. - Trading income losses on fixed income investments due to higher yields played a modest role. - For most banks, losses on corporate loans were larger than losses on household loans. ### Sensitivities and single-factor tests - Single-factor shocks: - Losses from individual severe interest, market and concentration shocks would be manageable. - Shutdown of FX swaps market leading to increases in RWAs would entail higher capital needs. - Quantified single-factor outcomes (examples): - Interest rate shock in the banking book: aggregate loss ~3.5 percent of 2015 aggregate CET1 capital. - Market risk shock on trading book and AFS: aggregate loss rate 6.3 percent of portfolio, amounting to 6 percent of CET1 capital. - Shutdown of FX swaps market: increase capital needs by almost 20 percent and lower 2015 CET1 ratio by 2.3 percentage points. - Sensitivity using authorities’ credit loss model: - Credit losses would double compared to benchmark; system-wide CET1 capital ratio would fall by 8.7 percentage points over 2016-2018. - All banks would fall below the supervisory threshold and one bank would breach the regulatory threshold. - Recapitalization peak in 2018 would be 7 percent of nominal GDP. ### Effect of increased corporate risk weights - Incorporation of supervisory initiative raising corporate risk weights: - Higher corporate risk weights reduced the system-wide CET1 ratio in both baseline and stress. - Under IMF approach, this pushed three additional banks below the supervisory threshold in the stress scenario and one bank in the baseline scenario. - Even with higher corporate risk weights no bank would breach the regulatory threshold per IMF TD. ### Liquidity stress tests and contagion (banks) - Liquidity: - LCR and NSFR tests (by currency) and maturity ladder analysis performed. - LCR results: - Under baseline: aggregate LCR (all currencies) above 100 percent for all banks. - LCR in SEK for three banks is below 100 percent, but shortfall compensated by buffers in convertible currencies. - When off-balance sheet items not included in Swedish LCR are included, aggregate LCR of one bank falls below 100 percent. - Under adverse scenario (larger run-offs/haircuts), three banks would fall below 100 percent. - NSFR results: - Proxy NSFR suggests most banks would fall below 100 percent in stress scenario. - Basel III NSFR numbers for all banks are almost 100 percent in baseline. - Maturity ladder and cash-flow: - Two banks would have medium-term funding problems if faced with severe and long-lasting liquidity and market shocks (1 month–12 months). - Large maturity mismatches for some banks reflect reliance on wholesale funding in foreign exchange to finance longer-term loans. - Contagion analysis: - Spillover risk between four largest banks appears limited in baseline because interbank exposures were significantly reduced by collateralization. - Baseline network results (LGD/funding/fire-sale discount = 25 percent; regulatory hurdle): - No contagion chains among the four banks. - Capital losses per bank ranged between 1 percent and 9.5 percent of CET1 capital. - Sensitivity (parameters = 50 percent): losses increased to 3.5–22 percent of CET1; second-round defaults not triggered unless supervisory hurdle used. - Caveats: exercise likely underestimated spillovers because not all interbank exposures were captured; did not capture default risk outside banks (including abroad); did not capture liquidity risks related to illiquid collateral of covered bonds or simultaneous investor flight. ### Insurance sector stress tests — key findings - Aggregate position: Swedish insurance companies are, on aggregate, sufficiently capitalized to withstand severe shocks on financial markets. - Life insurers: - Pre-stress SCR coverage (aggregate): 196 percent. - Post-stress Scenario 2 SCR coverage: 132 percent (BU) and 135 percent (TD). - Face large decline in own funds due to high sensitivity to market risks but start from high capitalization levels. - Available own funds decline by 35 percent on aggregate under Scenario 2 (TD). - Equity shock (-25.4 percent in Swedish stock prices) reduces life insurers’ available own funds by 22 percent on aggregate. - Non-life insurers: - Pre-stress SCR coverage (aggregate): 152 percent. - Post-stress Scenario 2 SCR coverage: 140 percent (BU) and 154 percent (TD). - Aggregate own funds decline by 33 percent under Scenario 2. - Capital shortfalls: - Two companies did not meet SCR after stress, with small capital shortfalls. - Aggregate capital needed: SEK 17.7 bn in scenario 1 and SEK 4.0 bn in scenario 2 (both < 1 percent of Swedish GDP). - Re-investment risk: - Nearly three quarters (73 percent) of fixed-income assets will expire in the next five years, with average coupon rates of more than 3 percent (1st five-year cohort average coupon 3.4 percent; next 5-year cohort 2.7 percent). - Insurers may need to lower guaranteed interest rates or change asset allocation. ### Investment funds liquidity analysis - Funds’ total net assets increased by 70 percent since end-2011. - Asset allocation (shares of total assets): domestic equity 36 percent; foreign equity markets 36 percent; domestic mortgage bonds 8 percent. - IMF TD liquidity stress test design: - One-time tail redemption shock equal to the 1st percentile of net flow rate distribution for funds of same style over 2005Q1–2015Q4. - Coverage: funds accounting for about 95% of asset management industry; data cut-off 2015Q4. - Two asset-sale approaches: - Pro rata: sell assets pro-rata to keep portfolio weights intact. - Waterfall: sell assets in descending order of liquidity based on LCR haircut hierarchy. - Key quantitative result: - Under pro-rata, volume of corporate bonds sold by funds under severe stress could be 50 percent higher than the average turnover of corporate bonds. - For other bond types and equities, volumes sold under tail event shock smaller than average turnover; domestic stock market turnover much larger than equity sold by funds. - Caveat: analysis did not account for liquidity risk materialization in other sectors that hold same assets; exercise likely overestimates markets’ capacity to absorb assets sold. ### Key scenario quantitative levels (selected exact entries from stress scenario table) - Stress scenario levels (billions of 2014 national currency units): - Real GDP: 2015 = 4078.6; 2016 = 4050.0; 2017 = 3912.2; 2018 = 3798.9; 2019 = 3874.3; 2020 = 3943.7 - Investment: 2015 = 988.9; 2016 = 984.8; 2017 = 879.1; 2018 = 819.9; 2019 = 818.1; 2020 = 827.9 - Unemployment rate: 2015 = 7.4; 2016 = 8.2; 2017 = 10.1; 2018 = 12.6; 2019 = 13.5; 2020 = 13.5 - Real house prices (1981=100): 2015 = 129.2; 2016 = 109.0; 2017 = 95.9; 2018 = 92.0; 2019 = 90.8; 2020 = 94.6 - Short term interest rate (percent): 2015 = -0.2; 2016 = -0.5; 2017 = -0.5; 2018 = -0.5; 2019 = -0.4; 2020 = 0.1 - Equity prices: 2015 = 1562.5; 2016 = 1165.1; 2017 = 1210.3; 2018 = 1415.8; 2019 = 1550.8; 2020 = 1682.4 - Stress scenario percent changes: - Real GDP growth (percent): 2016 = -0.7; 2017 = -3.4; 2018 = -2.9; 2019 = 2.0; 2020 = 1.8 - Real house prices (yoy change): 2016 = -15.6; 2017 = -12.0; 2018 = -4.1; 2019 = -1.3; 2020 = 4.2 - Equity prices (yoy change): 2016 = -25.4; 2017 = 3.9; 2018 = 17.0; 2019 = 9.5; 2020 = 8.5 ### Identified data and methodological gaps - Authorities’ stress testing framework can be improved in areas including: - Liquidity and solvency stress testing frameworks. - Data gaps: collect all interbank exposures and conduct regular network analysis including market-price based approaches. - Design and perform regular liquidity stress tests for investment funds. - Collect granular data on trading and net interest income (including derivatives), and on available and required stable funding consistent with Basel III NSFR for all significant currencies. ### Stress-testing recommendations (selected, with timing) - 1. Conduct regular (e.g. on annual basis), comprehensive stress tests for all major financial sectors and the non-financial sector that capture macro-financial factors, cross-country spillover, market-price based network analysis and feedback effects between institutions and between solvency and liquidity (FI, Riksbank). Timing: NT. - 2. Improve scenario-based solvency stress tests; map FI household stress test results into credit losses on mortgage exposures; start running corporate sector stress tests and continuously monitor corporate sector; link liquidity and solvency risks (FI, Riksbank). Timing: I. - 3. Start collecting granular data on available and required stable funding consistent with Basel III NSFR for all significant currencies to monitor the Basel III NSFR on a regular basis (FI). Timing: I. - 4. Improve the cash flow liquidity stress testing framework (FI). Timing: I. - 5. Collect all interbank exposure data, and run network analysis on a regular basis; supplement with market-price based approaches (Riksbank). Timing: I. - 6. Develop and perform macroprudential stress tests for the insurance sector complementing Solvency II Standard Formula and ORSA; require additional capital where risk profile deviates from Standard Formula (FI). Timing: NT. - 7. Follow up inaccurate or incomplete reporting at an early stage to avoid embedding misreporting in analytical work (FI). Timing: I. - 8. Develop and perform liquidity risk analysis for the mutual fund industry on a regular basis (FI). Timing: I. - Timing legend: C = continuous; I (immediate) = within one year; NT (near term) = 1-3 years; MT (medium term) = 3-5 years. Italic: Source: EXECUTIVE SUMMARY, IMF FSAP Sweden (cr17309) ### EXECUTIVE SUMMARY __________________________________________________________________________ 6 ### EXECUTIVE SUMMARY ### Overview and Scope - Stress tests covered three major segments of the domestic financial sector, with an emphasis on the largest financial institutions. - Banking: the four largest banks together constitute over 75 percent of the banking system by assets. - Insurance: the four largest life insurance providers and three non-life insurers hold 78 and 53 percent market share in terms of gross written premiums, respectively. - Investment funds: around 850 investment funds representing 95 percent of the industry’s total net assets were included. - Non-financial corporations were stress tested to validate corporate inputs used in the banking stress tests. ### Methodology and Scenarios - Solvency stress tests were implemented by the FSAP team (TD), the banks (BU) and the authorities (Riksbank and FI). - BU tests followed the methodology used in the last EU-wide (EBA) stress testing exercise. - TD approaches: - FSAP team: balance sheet approach, also considered future changes to corporate risk weights. - Riksbank: simple solvency framework based on projection of credit losses. - Liquidity tests used the liquidity coverage ratio (LCR), NSFR and maturity ladder analysis, on aggregate basis and by currencies. - Contagion analysis covered domestic interbank exposures (implemented by the Riksbank). - Scenarios: - Baseline: IMF’s April 2016 World Economic Outlook projections for Sweden. - Stress: Sweden-specific risks and spillovers from a recession in the Nordic and Baltic region; mirrored EBA’s 2016 adverse scenario for the first three years but included more severe interest rate and exchange rate shocks. - Single factor shocks: severe interest, market and concentration shocks and a shutdown of the foreign exchange swaps market. ### Key Solvency Stress Test Findings - General result: banks would be resilient to a severe economic distress, but with differences across approaches. - IMF top-down (TD) results: - One bank would fall below the supervisory hurdle rate and no bank would breach the regulatory hurdle rate over the stress testing horizon. - System-wide CET1 ratio would fall by 2.5 percentage points in 2016 relative to the base year (or 2.5 percentage points relative to the baseline scenario in 2016) and by additional 1.5 percentage points in 2017–2018. - System-wide “all-in” leverage ratio would fall by 0.3 percent point in the downturn period (from 4.1 percent in 2015 to 3.8 percent in 2018). - Provisions for credit losses increased seven times by 2018 (comparing to four times in EBA test and twenty times in the authorities’ test). - Bottom-up (BU) results: - One bank would fall below the supervisory hurdle rate. - System-wide CET1 would fall by 2.3 percentage points during the first three years of the stress horizon. - Authorities’ TD results: - All banks would fall below the supervisory thresholds. - System-wide CET1 would fall by 4.9 percentage points during the period of downturn. - Drivers of capital deterioration in downturn years (2016–2018): - Lower net interest income, mainly due to sharp increases in funding costs. - Higher provisions for credit losses, particularly on corporate exposures. - Increases in risk weighted assets (RWAs) largely from higher probability of defaults (PDs) and depreciation of SEK. - Trading income losses on fixed income investments due to higher yields played a modest role. - For most banks, losses on corporate loans were larger than losses on household loans. ### Sensitivities and Single Factor Tests - Single factor shocks: - Losses from individual severe interest, market and concentration shocks would be manageable. - A shutdown of the FX swaps market leading to increases in RWAs would entail higher capital needs. - Risk around credit loss projections is large and likely led to underestimation of credit losses in EBA and IMF stress tests. ### Effect of Increased Corporate Risk Weights - Incorporation of the recent supervisory initiative that will increase corporate risk weights has significant impact: - Higher corporate risk weights reduced the system-wide CET1 ratio in both the baseline and stress scenario. - Under the IMF approach, this pushed three additional banks below the supervisory threshold in the stress scenario and one bank in the baseline scenario. ### Liquidity Stress Tests and Contagion - Liquidity stress test findings: - Banks are broadly resilient to shocks characterized by withdrawal of funds and haircuts on liquid assets similar to Swedish LCR and Basel III NSFR standards. - Some banks would face greater short term and medium term liquidity pressures when subjected to additional increases in parameters. - Maturity ladder exercise indicates large maturity mismatches for some banks, reflecting heavy reliance on wholesale funding in foreign exchange to finance longer term loans. - Contagion analysis: - Spillover risk between the four largest banks appears limited, mainly because interbank exposures were significantly reduced by collateralization. - Important caveats: exercise likely underestimated spillovers because not all interbank exposures were captured; did not capture default risk outside banks (including abroad); did not capture liquidity risks related to illiquid collateral of covered bonds or simultaneous investor flight in extreme distress. ### Insurance Sector Results - Aggregate position: Swedish insurance companies are, on aggregate, sufficiently capitalized to withstand severe shocks on financial markets. - Life insurers: - Face a large decline in own funds due to high sensitivity to market risks but start from high capitalization levels before stress. - Non-life insurers: - Immediate impact of stress is less pronounced. - Two companies did not meet the solvency capital requirement after stress, with only a small capital shortfall. - Identified vulnerabilities: - Large holdings of equity. - Strong interconnectedness with the domestic banking sector. - Profitability expected to recover quickly after stresses. - Underwriting risks: - Higher longevity, pandemics and catastrophic events are unlikely to create major solvency issues if they occur in isolation. - Re-investment risk: - Nearly three quarters of fixed-income assets will expire in the next five years, with average coupon rates of more than 3 percent. - Insurance companies may need to adjust products (effectively lowering guaranteed interest rates) or change asset allocation. ### Investment Funds Liquidity Analysis - Liquidity analysis focused on whether markets could absorb severe redemption pressures. - Findings: - Corporate bond markets would be under stress in the tail event of severe redemptions. - Funds invested in corporate bonds may be forced to sell assets at fire-sale discounts to meet redemptions. - Fire sale risks from funds meeting redemptions in other bonds’ and equity markets appear small. - Caveat: the analysis did not account for materialization of liquidity risks in other sectors that hold the same assets; hence the exercise likely overestimated the markets’ capacity to absorb assets sold during times of stress. ### Overall Assessment and Areas for Improvement - The overall stress testing exercise suggests room for improvement in authorities’ stress testing framework. - Identified areas for enhancement include: - Improving liquidity and solvency stress testing frameworks. - Addressing data gaps by collecting all interbank exposures and conducting network analysis on a regular basis, including market-price based approaches. - Designing and performing regular liquidity stress tests for investment funds. ### Stress Testing Recommendations (summarized) - 1. For macroprudential purposes, conduct regular (e.g. on annual basis), comprehensive stress tests for all major financial sectors and the non-financial sector that capture the impact of macro-financial factors, cross-country spillover, including by using market-price based network analysis and stress testing and feedback effects between financial institutions and interactions between solvency and liquidity (FI, Riksbank). Timing: NT. - 2. Improve scenario-based solvency stress tests; map the results of the FI household stress test into credit losses on mortgage exposures under the solvency stress test scenario; start running corporate sector stress tests and continuously monitor the health of the corporate sector; link liquidity and solvency risks (FI, Riksbank). Timing: I. - 3. Start collecting granular data on available and required stable funding consistent with the Basel III NSFR for all significant currencies to monitor the Basel III NSFR on a regular basis (FI). Timing: I. - 4. Improve the cash flow liquidity stress testing framework (FI). Timing: I. - 5. Collect all interbank exposure data, and run network analysis on a regular basis. Supplement the spillover analysis with approaches that extract interconnectedness risks from market prices (Riksbank). Timing: I. - 6. Develop and perform macroprudential stress tests for the insurance sector which complement the Solvency II Standard Formula and the Own Risk and Solvency Assessment, and require insurance undertakings to meet additional capital requirements where the risk profile deviates significantly from the Standard Formula (FI). Timing: NT. - 7. Follow up inaccurate or incomplete reporting at an early stage to ensure that any significant misreporting is not embedded in analytical work (FI). Timing: I. - 8. Develop and perform liquidity risk analysis for the mutual fund industry on a regular basis (FI). Timing: I. -Timing legend: C = continuous; I (immediate) = within one year; NT (near term) = 1-3 years; MT (medium term) = 3-5 years. *Source: EXECUTIVE SUMMARY, IMF FSAP Sweden (cr17309)* ### INTRODUCTION ### INTRODUCTION ### Scope and Methodology of FSAP Stress Tests - The 2016 Financial Sector Assessment Program (FSAP) stress tests covered banks, insurance companies, and mutual funds using supervisory data provided by the authorities and the banks. - Types of tests performed: - Banks: solvency, liquidity and contagion stress tests. - Solvency stress tests implemented by FSAP team (TD), banks (BU) and the authorities (Riksbank and FI). - Liquidity stress tests performed by the IMF based on supervisory data. - Contagion stress test implemented by the Riksbank. - Banks’ BU stress tests followed the EBA methodology; authorities’ TD solvency framework projected credit losses. - Insurance companies: IMF TD solvency stress tests; companies provided BU solvency stress tests. One BU scenario aligned with EIOPA EU-wide insurance sector stress test. - Investment funds: liquidity stress tests undertaken by the FSAP team. - Coverage and data: - Stress tests covered the four largest banks which account for about 75 percent of the banking sector’s assets. - Cut-off date of the data was December 30, 2015. - Minimum capital requirements used as hurdle rates were consistent with Swedish capital regulatory standards reflecting Basel III, Sweden-specific buffers and pillar 2 add-ons. - Approach: - Solvency tests used two macroeconomic scenarios (baseline and stress) over a five-year risk horizon (2016–2020). - Liquidity tests used one-off shocks calibrated to regulation (banks) or historical evidence (investment funds), performed separately from solvency tests. - Institution-specific risks not well captured by scenarios were addressed via single factor sensitivity tests. ### Scenarios: Baseline and Stress - Two common scenarios for solvency tests over a five-year horizon (2016–2020): - Baseline scenario: followed IMF April 2016 WEO projections for Sweden. - Stress (adverse) scenario: mirrored the EBA 2016 adverse scenario for Sweden but with more severe interest rate and exchange rate shocks; accounted for Sweden-specific risks and Nordic/Baltic spillovers. - Key qualitative features of the stress scenario: - A deep recession with an “L-shaped” recovery driven by global, regional and domestic risks. - Swedish exports, investment, and employment are hit by a sharp but temporary global recession; Real GDP declines by 6.9 percent three years after the initial shock. - Investment falls by about 17 percent three years after the shock. - Unemployment rate rises by about 5 percentage points in the scenario. - House prices decline by about 32 percent in three years. - Market sentiment shifts increase financing costs for the Swedish banking system and domestic credit conditions. - IMF stress scenario considers a slightly larger spread of the Swedish long term bond yields over German bunds (by about 50bps on average for the first two years) relative to EBA, implying some krona depreciation. - Riksbank response in the scenario: policy rate kept at -0.5 percent until mid-2019 before gradually raising it. - Sensitivity analysis (“low for long”): assumed policy rate 100bps lower than in the stress scenario. - Recovery: growth recovers somewhat three years after the shock but remains subdued in years 4 and 5 as households repair balance sheets. - Nordic and Baltic supplement: baseline and stress scenarios for the Nordic and Baltic region; an assumed resurgence of Russia/Ukraine geopolitical conflict depresses Baltic growth; declining oil prices hinder Norway and Denmark. ### Key Scenario Quantitative Levels and Changes (as presented in Table 3) - Stress scenario levels (billions of 2014 national currency units, unless stated otherwise): - Real GDP: 2015 = 4078.6; 2016 = 4050.0; 2017 = 3912.2; 2018 = 3798.9; 2019 = 3874.3; 2020 = 3943.7 - Investment: 2015 = 988.9; 2016 = 984.8; 2017 = 879.1; 2018 = 819.9; 2019 = 818.1; 2020 = 827.9 - Unemployment rate: 2015 = 7.4; 2016 = 8.2; 2017 = 10.1; 2018 = 12.6; 2019 = 13.5; 2020 = 13.5 - HICP index: 2015 = 100.7; 2016 = 96.8; 2017 = 96.5; 2018 = 97.1; 2019 = 97.7; 2020 = 98.5 - Real house prices (1981=100): 2015 = 129.2; 2016 = 109.0; 2017 = 95.9; 2018 = 92.0; 2019 = 90.8; 2020 = 94.6 - Nominal house prices: 2015 = 135.4; 2016 = 109.4; 2017 = 95.5; 2018 = 92.1; 2019 = 91.5; 2020 = 96.1 - Short term interest rate (percent): 2015 = -0.2; 2016 = -0.5; 2017 = -0.5; 2018 = -0.5; 2019 = -0.4; 2020 = 0.1 - Long term interest rate (percent): 2015 = 0.7; 2016 = 2.4; 2017 = 2.0; 2018 = 1.9; 2019 = 2.0; 2020 = 2.5 - 2-year covered bond yield (percent): 2015 = 0.2; 2016 = 1.5; 2017 = 2.5; 2018 = 3.0; 2019 = 3.5; 2020 = 3.9 - 5-year covered bond yield (percent): 2015 = 0.9; 2016 = 2.3; 2017 = 3.4; 2018 = 4.0; 2019 = 4.6; 2020 = 4.9 - Equity prices: 2015 = 1562.5; 2016 = 1165.1; 2017 = 1210.3; 2018 = 1415.8; 2019 = 1550.8; 2020 = 1682.4 - KIX: 2015 = 112.6; 2016 = 115.6; 2017 = 126.1; 2018 = 124.0; 2019 = 120.3; 2020 = 113.8 - Stress scenario percent changes: - Real GDP growth (percent): 2016 = -0.7; 2017 = -3.4; 2018 = -2.9; 2019 = 2.0; 2020 = 1.8 - Investment growth: 2016 = -0.4; 2017 = -10.7; 2018 = -6.7; 2019 = -0.2; 2020 = 1.2 - Inflation (HICP): 2016 = -3.9; 2017 = -0.3; 2018 = 0.6; 2019 = 0.7; 2020 = 0.8 - Real house prices (yoy change): 2016 = -15.6; 2017 = -12.0; 2018 = -4.1; 2019 = -1.3; 2020 = 4.2 - Nominal house prices (yoy change): 2016 = -19.2; 2017 = -12.7; 2018 = -3.5; 2019 = -0.7; 2020 = 5.0 - Equity prices (yoy change): 2016 = -25.4; 2017 = 3.9; 2018 = 17.0; 2019 = 9.5; 2020 = 8.5 - KIX (yoy change): 2016 = 2.7; 2017 = 9.1; 2018 = -1.6; 2019 = -3.0; 2020 = -5.4 - Baseline (WEO) highlights (levels and growth rates presented in Table 3): - Real GDP: 2015 = 4078.6; 2016 = 4228.2; 2017 = 4347.0; 2018 = 4456.0; 2019 = 4559.9; 2020 = 4658.0 - Real GDP growth (percent): 2016 = 3.7; 2017 = 2.8; 2018 = 2.5; 2019 = 2.3; 2020 = 2.2 - Unemployment rate (percent): 2016 = 6.8; 2017 = 7.0; 2018 = 7.2; 2019 = 7.2; 2020 = 7.2 - Short term interest rate (percent): 2016 = -0.2; 2017 = 0.4; 2018 = 1.0; 2019 = 1.6; 2020 = 2.2 ### IMF’s Top-Down Solvency Stress Tests of Banks: Design and Implementation - Overview of TD solvency tests: - Covered: data and scope; state of the four largest banks and vulnerabilities; capital definitions and standards used; stress test methodology using satellite models mapping macro scenarios to credit losses, income projections, balance sheet and RWA dynamics; behavioral assumptions on capital actions; and results of IMF TD exercise. - Approach: balance sheet-based approach assessing solvency of individual banks through changes in net income and RWAs; resembled BU test; used supervisory consolidated data provided by authorities and banks. - Data and scope: - Coverage: four largest banks, about 75 percent of sector assets. - Minimum capital hurdles consistent with Swedish Basel III-based regulatory standards and Sweden-specific buffers and pillar 2 add-ons. - Capital and reporting: - Minimum capital requirements used as hurdle rates were consistent with Swedish capital regulatory standards reflecting Basel III requirements and Sweden-specific buffers and pillar 2 add-ons. ### Four Largest Banks: Structure, Performance and Vulnerabilities - Size and asset composition: - Total assets of the four largest banks at end-2015 represent around 300 percent of nominal GDP. - Loans represent 56 percent of total assets. - Securities represent 15 percent of total assets. - Both loans and securities increased by 5 percentage points since end-2011. - Mortgage share increased by 5 percentage points to reach 50 percent of the loan portfolio at end-2015. - Corporate loans share decreased from 47 percent to 42 percent. - Almost half of exposures of the largest banks were assets in foreign subsidiaries and branches, predominantly in other Nordic countries. - Liquidity and funding structure: - Liquid assets largely comprised covered bonds and sovereign bonds; covered bonds represent 30 percent of liquid assets. - Around 60 percent of total funding is in foreign exchange. - Customer deposits represent 30 percent of liabilities. - Long-term wholesale funding (mostly covered bonds) increased by 5 percentage points and reached 25 percent of total liabilities. - Other short-term bonds, both in SEK and foreign currency, represented 30 percent of total funding. - Profitability and margins: - Banks performed well in negative rates environment with solid profitability underpinned by strong loan re-pricing power. - High return on equity driven by low credit losses and NPLs, high fees, and cost efficiency. - Net interest margin remained very stable in negative rates environment. - About 70 percent of mortgages have floating rate contracts enabling re-pricing every three months. - Vulnerabilities highlighted for stress testing: - Credit risks from residential mortgages (sensitive to interest rate and unemployment shocks). - Credit risks from corporate exposures (slower growth, higher financing costs). - Effects of higher funding costs under stress. - Roll-over risks from funding illiquid loans with shorter-term/wholesale instruments, including in foreign currency. - Counterparty credit risk from swaps counterparties failing to meet obligations. - Contagion risks from high interconnectedness via interbank and swaps markets and holdings of each other’s debt. - Leverage and capital observations: - Capital and capital ratios of largest banks greatly increased in recent years. - Leverage ratio increased at lower speed, suggesting increases in capital ratios were primarily driven by lower risk weights rather than higher capital. ### Structure of the Note and Next Sections - The note next: - Reviews assumptions of the two solvency scenarios. - Presents three different approaches of the FSAP’s solvency stress test of the banking sector, analyzes results and reconciles findings. - Presents liquidity stress testing exercise results. - Presents contagion analysis. - Follows with insurance sector stress-test analysis and finishes with liquidity stress test of investment funds. *Source: INTRODUCTION, 2016 Sweden FSAP stress-testing note (cr17309).* ### 13. Hurdle rates were defined in terms of common equity tier 1 (CET1) ratios. The capital ### 13. Hurdle rates were defined in terms of common equity tier 1 (CET1) ratios. The capital ### Hurdle rate definitions and methodology - Hurdle rates were defined in terms of common equity tier 1 (CET1) ratios. - The capital definition applied in the stress test corresponded to that required by Swedish regulation. - Solvency stress tests assessed the potential impact of negative shocks on capital requirement metrics over a five-year risk horizon against two sets of hurdle rates. - The hurdle rates were “individualized”, i.e. calculated for each bank as a function of projected exposures and different mortgage floors in Sweden and Norway. - Regulatory hurdle rate: - Defined as the CET1 ratio that does not include any buffers and therefore includes: - a minimum Basel III capital requirement, - Pillar 2 own-fund requirements associated with pension risk, concentration risk and interest rate risk in the banking book, - the ”microprudential” capital requirements for Swedish and Norway mortgage (mortgage floors, 15 percent). - Supervisory hurdle rate: - Took into account all capital buffers (in addition to capital elements of the regulatory hurdle rate): - “macroprudential” mortgage floors (10 percent), - capital conservation buffer, - countercyclical capital buffer, - the common equity systemic risk surcharge of 5 percent. - Note: This supervisory threshold does not take into account that some buffers might be released in a crisis. ### Basel III leverage ratio and capital definition - The impact on the Basel III leverage ratio was considered: - It was assumed that the total exposure grows at the rate of total assets. - Regulatory adjustments were assumed to stay constant at the 2015 level. - Tier 1 capital was defined on the fully phased-in basis. - Basel III phase-ins and phase outs were incorporated in the definition of capital: - Consistent with the national regulatory arrangement, 100 percent of projected AOCI was incorporated into CET1 capital from 2016 onwards. - The nominal value of goodwill, intangible assets, and DTAs was held constant. - Since there are no phase-ins of deductions from CET1 and provision factors are equal to 100 percent there were no effects of deductions on CET1 ratios over the stress testing horizon. ### Sweden: Hurdle Rates (components and numeric entries) - Components listed (I. to IX.): - I. Minimum Common equity tier 1 (CET1) - II. Capital requirement in Pillar II, excl systemic risk and risk weight floor - III. "Microprudential" mortgage floors (15%) - IV. "Macroprudential" mortgage floors (10%) - V. Countercyclical capital buffer - VI. Capital conservation buffer - VII. Systemic risk buffer - VIII. Specific Pillar II risks - IX. Supervisory threshold (I. to VIII.) - Phase in of deductions from CET1: 100, 100, 100, 100, 100, 100 (as listed) - Phase out of existing AOCI capital adjustments: 100, 100, 100, 100, 100, 100 (as listed) - Calculated based on projection of exposures in Sweden and Norway and mortgage floors: Different for each bank - 2015-2020: 4.5 - Calculated based on projection of exposures in Sweden and Norway and respective CCB rates: 3.0 - 2.5 (listed without additional qualifier) - Held constant and equal to 2015q4: 2.0 - Calculated based on projection of exposures in Sweden and Norway and mortgage floors: Regulatory threshold (I. to III.) ### Projection of Risk-Weighted Assets (RWAs) - Three components of RWAs for each bank were projected: credit RWAs, market RWAs and operational RWAs. - Credit risk: - The IRB formulas were used to calculate and project credit RWAs for each asset class reported by the banks (Table 5). - Through-the-cycle PDs were updated to reflect the increase in point-in-time PDs. - Point in time PDs for different economic sectors over the stress testing horizon were calculated using projected credit losses for each economic sector. - LGDs (reported by banks in 2015Q4 for each economic sector) and projected exposure at default were used. - Exposures included both standardized and IRB exposures and their projection was consistent with projection of loans and total assets, taking into account the dynamics of the exchange rate and defaulted loans. - Credit RWAs for trading, equity, and securitization were taken from the bottom-up exercise. - Market and operational risk: - No meaningful way to project RWAs for operational and market risks; they were taken from the bottom up exercise. - Corporate exposure risk-weighting adjustment: - Credit RWAs for corporate exposure were adjusted to take into account the recent FI’s initiative to increase risk weights for corporate exposures. - Consistent with the new regulatory initiative, the average risk weights for exposures to corporate sector was set to 30 percent from the end of 2016 onwards. - For those banks whose risk weight on corporate exposure was already above 30 percent, their PDs were recalculated such that the average risk weight on corporate exposure increased by 2 percentage points. ### Models and behavioral assumptions - Historical quarterly data, simple panel regression, duration and maturity gap models were used to forecast each bank’s main components of balance sheets and income statement items (Table 6). - The models were intended to capture how the balance sheet, RWAs, and net income of each bank are affected by the macroeconomic and financial conditions described in the scenarios. - Projections process: - Step 1: Projections of balance sheets over the stress testing horizon were used for calculation of credit risk RWAs and income statement items. - Step 2: Projections of RWAs and net income, with assumptions on dividend distribution and AOCI, determined capital ratios over the stress testing horizon. - Asset disposals and acquisitions over time were not considered. - Footnotes and modeling details: - All credit losses by economic sectors were assumed to grow at the same rate as total credit losses. - Panel regression models included models with fixed effects estimated over the period from 1989 to 2015. ### Balance sheet growth projections and stress scenario assumptions - Projections of banks’ balance sheet items were used in projecting income statement items and credit risk RWAs. - Stress scenario assumption for asset class growth: - In the stress scenario it was assumed that each asset class grows at the zero rate. - Rationale: forecasting exposures in the stress scenario would likely bring about negative growth rates of exposures which in turn would increase capital ratios. By setting the growth rate of assets to zero recapitalization of banks via deleveraging was avoided. - Additional dynamics: - Loan and total assets changed during the stress testing horizon reflecting the change in the exchange rate. - The securities portfolio reflected mark to market changes due to changes in interest rates. *Source: Finansinspektionen and staff calculations; IMF Staff.* ### 1. Credit RWAs ### 1. Credit RWAs ### IRB formula / From bottom up / Project CET1 and CET1 requirement - Independent variables from scenarios/approach included: GDP, Unemp. Rate, Int. rates, House prices. - Change in regulatory capital, CET1 calculated as (IX.-X.-XI.+XII.) with inputs from GDP, Int. rates, Equity prices. - Total assets and maturity gap assessed with duration approach. ### Projection methodology: balance sheet, income, and provisions - Loan growth in the baseline was modeled using a panel regression model with GDP growth, unemployment rate, and interest rates as exogenous variables. Growth rate of balance sheet assumed equal to growth rate of loan portfolio. - Provisions for loan losses: - Total provisions projected as total credit losses via a panel regression model with GDP growth rate, unemployment rate, interest rates and house price growth rate as exogenous variables. - GDP for cross-border spillovers defined as a weighted average of GDPs of countries where banks have exposure to (bank’s weight for a country = share of its exposure to that country in its total exposure). - Projected credit losses used to derive point-in-time (PIT) PDs to update through-the-cycle (TTC) PDs and calculate credit RWAs. - Provisions for own sovereign and corporate bonds held to maturity accounted for via haircuts due to changes in risk free rate and credit spreads. - Interest income/expense: - Maturity gap analysis by time-to-repricing; assumed no change in maturity profile over stress horizon and all interest earning assets and liabilities grow at projected loan growth rate. - Cash flow effects from increases in lending and funding costs considered. - EBA guidelines used to calculate lending rates and funding costs as function of change in sovereign spread, with pass-through parameters defined as weighted averages of EBA parameters reflecting each bank’s asset/liability structure. - Flow effects on net interest income from changes in interest rates on derivatives’ portfolio were not considered due to data constraints and reporting differences. - Non-interest income excluding trading income projected using panel regression with equity prices, GDP growth and interest rates as exogenous variables. - Trading income and realized losses on AFS/HTM securities assessed through a duration approach for interest rate and credit spread risks; losses on equity and other positions (including derivatives) modeled as functions of equity prices, interest rates and real GDP dynamics. - Non-interest expenses projected using a panel regression model with growth rate of total assets as the only independent variable. - Taxes set at the effective tax rate in 2015 for the whole stress horizon in case of positive net income and zero otherwise. - Extraordinary items and minority interest set to zero. - Accumulated other comprehensive income (AOCI): unrealized losses on AFS securities and foreign currency translation projected as part of AOCI; losses on AFS calculated using duration approach; FX gains/losses calculated based on net open position and exchange rate dynamics. Other AOCI items (actuarial gains/losses on defined contribution pension plans, unrealized gains and losses on qualifying cash flow hedges) held constant or set at zero. ### Capital action assumptions - Dividend rule: - Dividends payable out of current year’s profit using Basel III capital conservation rule. - Maximum allowed dividend payout equal to dividend payout ratio in 2015 (effective div. payout rate in 2015, EDPR). - If a bank fell below the supervisory threshold before dividend distribution it was capital constrained and followed a schedule of dividend payouts per Table 7. If a bank fell below the supervisory threshold because of dividend distribution, dividend payout limited to ensure threshold not breached. Applies only if bank earned positive net income; if net income negative, no dividend payout. - Banks assumed not to issue new shares or repurchase shares during stress test horizon. - Example dividend distribution schedule (where supervisory threshold is 15 percent; specific buffers also considered): - Effective div. payout rate in 2015 (EDPR) ranges and assumed dividend payout: - 12.5-13.125: 0% x Net income (t) - 13.125-13.75: 20% x EDPR x Net income (t) - 13.75-14.375: 40% x EDPR x Net income (t) - 14.375-15.0: 60% x EDPR x Net income (t) - >15: Effective div. payout rate in 2015 (EDPR) ### Single-factor tests and sensitivity checks - Four single-factor one-time shock tests (more severe than first year of stress scenario): - Interest rate shock in the banking book to isolate re-pricing impact (Table 8). - Market risk shock in the trading book, AFS securities and CVA (interest rates, spreads, exchange rate, equity, commodities) (Table 9). - Failure of 10 largest counterparties: instantaneous unexpected default of bank’s largest counterparties with two LGD assumptions (100 percent and banks’ best estimate). - Shutdown of FX swaps market to identify effects on capital requirements from change in net open positions and higher RWAs due to materialization of counterparty credit risk and default of largest counterparties in FX swap market. - Sensitivity analyses by FSAP team included (non-exhaustive list): - Using a credit loss model similar to authorities’ stress test. - Allowing banks to deleverage in stress scenario. - Changing projected corporate PDs to reflect debt-at-risk measures from corporate ICR simulation and authorities’ household stress test. - “Low for long” scenario to capture prolonged negative rates. - Using EBA interest rate shocks. - Using market-based measures of solvency. - Example interest rate shock (bps) by tenor and region (as reported): - Tenor / Scenario Shocks (bps) - Overnight: SWE +100; US +100; Rest of EU +125; Russia +300 - 3 month: SWE +100; US +100; Rest of EU +125; Russia +300 - 1 year: SWE +150; US +150; Rest of EU +180; Russia +350 - 2 years: SWE +200; US +200; Rest of EU +250; Russia +400 - 5 year: SWE +250; US +250; Rest of EU +300; Russia +500 - 10 year and longer: SWE +350; US +350; Rest of EU +350; Russia +550 - Example market risk factor shocks (selected): - Equities' price change (%): SWE -30%; US -30%; EU/UK -40%; JPY -40%; Nordic countries -50% - Commodities' price change (%): Energy -40%; Base Metals -40%; Precious Metals -25%; Grains and others -30% - FX change (%): USD/SEK +40% (USD appreciates); EUR/SEK +20%; GBP/SEK +30%; NOK/SEK +20% - Credit Spreads change (bps): Covered bonds +500 for SWE; Senior Financials +250 for SWE; Non-Financial HY +800 for SWE (table provides broader country-specific spread shocks) ### Results: baseline and stress scenarios - Overall solvency stress test conclusion: banks would be resilient to severe economic distress. One bank would fall below the supervisory hurdle rate and no bank would breach the regulatory threshold over the stress testing horizon. - Recapitalization: - Recapitalization needed to bring the bank to the supervisory hurdle rate peaked in 2018 at 100 percent of its 2015 net income—which corresponds to 2 percent of 2018 nominal GDP. - CET1 and leverage dynamics: - System-wide CET1 ratio fell by 2.5 percentage points in 2016 relative to the base year (or 2.5 percentage points relative to the baseline scenario in 2016) and by additional 1.5 percentage points in 2017-2018. - System-wide “all-in” leverage ratio fell by 0.3 percent point in the downturn period (from 4.1 percent in 2015 to 3.8 percent in 2018) and no banks fell below the internationally agreed minimum of 3 percent. - Key drivers of downturn (2016-2018): - Net income fell from SEK 76 billion to SEK 34 billion net loss in 2016. - Provisions for credit losses increased seven times by 2018 due to a severe contraction of the real economy. - Provisions subtracted 1.1 percentage points from CET1 ratio in 2015 and additional 3.4 percentage points in the 2017-2018 period. - Trading income subtracted 1 percentage points from the system-wide CET1 ratio and additional 0.3 percentage points in the 2017-2018 period. - Higher RWAs subtracted 1.3 percentage points from the system-wide CET1 ratio in 2016-2018 largely as a result of higher PDs and depreciation of SEK. - Recovery period: - Positive changes in CET1 ratios driven by positive profits from higher interest income and lower provisions; net interest income increased significantly as funding shock dissipated and provisions dropped with pickup in real activity. - Baseline projection: - CET1 ratios projected to be broadly stable in baseline; system-wide CET1 ratio increased marginally in 2016 compared to base period, but fell by 0.7 percentage points over 2017-2018 as higher RWAs from balance sheet expansion outweighed profits. - System-wide leverage ratio projected to increase in 2016 but decline steadily after 2016 due to balance sheet expansion outpacing profit growth. ### Impact of supervisory corporate risk weight initiative - An initiative to increase corporate risk weights could significantly reduce system-wide CET1 ratio: - Would push three additional banks below the supervisory threshold in the stress scenario (raising recapitalization need to 3 percent of GDP) and one bank in the baseline scenario. - Even with higher corporate risk weights no bank would breach the regulatory threshold. ### Single-factor shock findings (quantified) - Interest rate shock in the banking book: - Aggregate loss from materialization of interest rate risk corresponds to around 3.5 percent of 2015 aggregate CET1 capital. - Around half of losses due to shock in Sweden; 1/3 of losses from shocks in Nordic countries. - Market risk shock in trading book, AFS securities and CVA: - Aggregate loss rate (loss over mark-to-market value of portfolio) on trading book and AFS securities equal to 6.3 percent, mostly driven by credit spreads, and amounted to 6 percent of CET1 capital. - CVA on OTC derivatives would more than double, but total increase in price of counterparty credit risk would be less than 1 percent of total CET1 capital. - Failure of 10 largest counterparties: - Two banks significantly affected by simultaneous failure of their 10 largest counterparties. - For other banks losses (using banks’ estimated LGDs) as share of CET1 ranged from 8-13 percent. - EBA stress test of two largest counterparties in derivatives yielded losses around 1 to 3 percent of CET1. - Shutdown of FX swaps market: - Would increase net open position substantially, increasing RWAs and capital needs by almost 20 percent and lowering 2015 CET1 ratio by 2.3 percentage points. ### Sensitivity analyses (select outcomes) - Credit loss model similar to authorities’ model: - Credit losses would double compared to credit losses in the benchmark stress scenario. - System-wide CET1 capital ratio would fall by 8.7 percentage points over 2016-2018. - All banks would fall below the supervisory threshold and one bank would breach the regulatory threshold. - Recapitalization needed to bring banks back to supervisory thresholds would peak in 2018 at 7 percent (recapitalization needs expressed as percent of nominal GDP). - Less severe interest rate shock (EBA adverse scenario): - Cumulative drop in system-wide CET1 ratio was 3.3 percentage points over first three years (instead of 4 percentage points in IMF benchmark stress). BU exercise under similar stress scenario amounted to 2.3 percentage points. - Difference mainly due to more conservative projection of credit losses and depreciation of SEK in IMF top-down framework. - Loan dynamics / deleveraging: - Allowing balance sheet to shrink (deleverage) in stress scenario raises CET1 ratios by 0.1 percentage points relative to benchmark stress scenario—small effect because loans not projected to fall substantially. - Low for long in baseline: - If interest rates stay at low levels in baseline, system-wide CET1 ratio would be higher by 0.7 percentage point over stress testing horizon versus baseline because lower interest rates contribute to lower credit losses, higher trading income, and lower funding costs (depends crucially on pass-through estimation method). ### Sectoral cross-checks and market-based analysis - Households’ stress test (FI) for residential mortgages: - FI’s stress scenarios (interest rate increase up to 5 percentage points; 10 percent unemployment; higher rates combined with 20 percent house price decline; higher unemployment combined with 20 percent house price decline) suggest Swedish mortgage borrowers’ payment ability is sufficient. FI encouraged to combine shocks and map results into PDs on mortgage exposures. - Non-financial corporate stress test (ICR-based): - Under combined scenario of interest rate increase (300 basis points), depreciation of SEK (15 percent) and profit drop (10 percent), share of firms with debt at risk would rise from 16 percent in baseline (2015) to 26 percent (i.e., by 70 percent). - In comparison, PDs on corporate exposure in IMF top-down test increase by 300 percent in 2016. - Market-price based analysis (contingent claims analysis, CCA): - Using equity prices and balance sheet information, historical time series of one-year expected default frequencies (EDFs), market capital to asset ratios (MCARs), and expected losses (ELs) were calculated using Moody’s CreditEdge EDFs and satellite econometric models with quarterly macro data 1999Q1–2015Q4 to project EDFs and MCARs for 2016-2020. - Results of market-based exercise broadly consistent with IMF top-down stress tests; one-year EDF projections under stress scenario increase 0.3 percent to an average of [text ends here in supplied content]. *Source: IMF Staff (excerpts from "1. Credit RWAs").* ### 0.7 percent (Panel 6). These stressed EDF levels are quite low—the range of peak EDF levels was ### cr17309 - 0.7 percent (Panel 6). These stressed EDF levels are quite low—the range of peak EDF levels was ### EDFs, Market Capital to Asset Ratios (MCAR) and Stress Scenario Projection - Stressed EDF levels: range of peak EDF levels was between 0.5 and 0.95 percent. - Interpretation: all banks have peak default probabilities of 0.95 percent or lower, which equates to investment grade or near investment grade ratings. - Linkage: bank EDFs are tightly linked to bank market cap to asset ratios. - Projected MCARs for 2016–2020 under the stress scenario: - MCARs decline on average about from eight to near four percentage points by 2017–18, then recover slowly afterwards. - These stress scenario MCARs reach levels seen in 2012 but do not reach levels seen during the 2009 crisis. - Econometric satellite model: - Data: quarterly EDF and macro data from 1999q1-2015q4. - Sample size: panel version with a total sample size of 272 bank-quarter observations. - Macro variables included: growth in GDP, inflation, exchange rate depreciation, stock market returns, real house price inflation, short term and long-term interest rates. ### Solvency Stress Tests: Approaches, Horizons, and Methodological Differences - Approaches compared: IMF top down (TD), Bottom up (EBA / banks), Top down by the authorities (FI/Riksbank). - Scenario horizons: - Bottom up exercise: three-year stress horizon. - IMF and authorities’ tests: five-year stress horizon. - Granularity and data: - Banks used much more granular data than IMF or authorities. - Authorities’ approach focused on modeling credit losses only. - Key methodological differences and consequences: - Authorities did not directly model effects of shocks on other income statement items (e.g., effect of interest rate shock on net interest income); these effects were indirectly taken into account when projecting credit losses—leading to higher projected credit losses in authorities’ test than would be if all income statement items modeled. - Losses on derivatives were not fully assessed in the IMF stress test because of reporting differences and lack of granular derivatives data. - IMF approach additionally assessed effects of higher corporate risk weights. ### Bottom-up (EBA) Stress Test Results — Highlights - Baseline projection: system-wide CET1 ratio projected to increase by 2.2 percent points from 2016 to 2018 (driven by positive net income and lower dividends). - Bottom-up stress outcome: - Over three years, system-wide CET1 would fall by 2.3 percentage points. - One bank fell below the supervisory threshold. - Drivers: higher RWAs (partly due to higher PDs and LGDs), higher provisions for credit losses, and lower net interest income from higher yields affecting trading income and interest expense. - Credit losses composition (bottom up): two thirds of total projected credit losses in the stress scenario came from corporate exposures; two thirds of losses came from non-Sweden exposures (mostly Finland, Norway and other countries (U.K., Russia)). ### IMF Top-down vs Bottom-up and Authorities’ Top-down: Quantitative Differences - Comparison of system-wide CET1 changes (first three years): - Bottom up cumulative drop: 2.3 percentage points. - IMF top down cumulative drop: 4 percent points. - Authorities’ top down cumulative drop: 4.9 percentage points. - Explanatory factors for IMF vs bottom-up difference: - (i) larger funding costs in IMF framework reducing net interest income; - (ii) depreciation of SEK affecting RWAs; - (iii) more conservative projection of credit losses under IMF framework. - Credit losses cumulative (2016-2018): - Authorities’ top down estimates were three times larger than in the IMF stress test. - Authorities’ top down estimates were five times larger than in the EBA exercise. ### Table 10: Key Methodological and Assumptions Differences (summary points) - Scenarios: - IMF top down: Baseline: IMF WEO; Stress: EBA + more conservative interest rate shocks + depreciation of SEK. - Bottom up/EBA: Baseline: EBA; Stress: EBA. - Authorities top down: Baseline: IMF WEO; Stress: EBA + more conservative interest rate shocks + depreciation of SEK. - Length of scenarios: IMF and authorities: 5 years; Bottom up/EBA: 3 years. - Capital hurdle rates: IMF applies Basel III minimum requirement+buffers; Bottom up/EBA and authorities: No hurdle rate (IMF hurdle rates imposed in comparisons). - Methodology highlights for IMF top down: - Balance sheet approach. - Net interest income: maturity gap and EBA guidance on lending and funding rates. - Trading income and AOCI: modeling and duration approach. - Provisions: credit losses model. - RWAs: - IMF: Credit risk RWAs: IRB formula; Market and Operational RWAs from BU test; Corporate risk weight floor. - Bottom up: Bank's own calculations consistent with EBA guidance. - Authorities: Assumption: RWAs increase by 7.5 percent each year in the first three years and 0 afterwards. - Projected loans and total assets: Held fixed in stress under bottom up and authorities; modeled in baseline for IMF. - Important items not considered by IMF: Net interest income on derivatives (not fully assessed). ### Bottom-up Stress Test (EBA) Visualization Findings - System wide CET1 under stress (2015–2018) fall and distributions illustrated in figures. - PDs and LGDs: PIT PDs, TTC PDs, LGDs shown over 2015–2018. ### Authorities’ Top-down Stress Test Results — Highlights - Baseline projection: authorities projected banks’ capital ratios to be stable under the baseline (negative effect from dividend pay-outs and higher RWAs matched by net income). - Authorities’ top down stress outcome: - Over three years, system-wide CET1 would fall by 4.9 percentage points. - All banks fell below the supervisory threshold but no bank fell below the regulatory threshold. - Key driver: higher provisions for credit losses which increased 18 times in 2016. - Differences vs IMF TD: - Cumulative credit losses (2016-2018) in authorities’ top down exercise were three times larger than IMF TD, five times larger than EBA. - Authorities modeled credit losses only and applied a haircut on pre-provision net income; interest shock effects on net interest income were not assessed consistently. ### Solvency Exercise Recommendations (Banks) - FI and Riksbank should improve scenario-based solvency stress test: - Reduce reliance on ad hoc assumptions (pre-provision income and RWAs). - Move to modeling approaches projecting all main elements of balance sheets and income statements consistent with scenarios. - Benefit: enhanced ability to validate EBA/bottom-up test results. - Use market-based solvency and shortfall measures for more frequent monitoring. - Data gaps to be closed as prerequisite: - FI should collect longer time series (possibly going back to the 1990s crisis) of balance sheet and income statement data adjusted for mergers and acquisitions and regulatory changes. - FI should collect more granular data on trading and net interest income, including derivatives income with breakdown between centrally-cleared and over-the-counter transactions to ensure comparability across banks. ### Liquidity Stress Tests — Framework and Assumptions - Top-down liquidity stress tests performed bank-by-bank. - Two primary tests: Swedish LCR and proxy NSFR by different currencies (stand-alone tests); thresholds set at 100 percent for both metrics. - Data: - LCR used supervisory consolidated data. - NSFR numbers mapped from NSFR templates banks report to COREP into Basel III NSFR categories. - LCR scenarios: - Baseline calibrated to reflect adjustment factors consistent with the Swedish LCR regulation. - Stress scenario applied larger run-off rates on wholesale funding and larger haircuts on liquid assets; included off-balance sheet items not reported in the LCR (e.g., guarantees to non-financial customers) in a separate baseline scenario. - NSFR scenarios: - Baseline mirrored Basel III NSFR parameters. - Stress scenario parameters adjusted by 10 percentage points compared to baseline. - Implied cash flow analysis: - Simulated outflows from 1 day to more than 10 years across maturity buckets. - More flexible definitions of liquid assets, inflows/outflows, haircuts, run-off and roll-off rates. - Objective calibration: scenario with outflows of at least 20 percent of funds within three months and additional 5–10 percent within next three months. - Run-off rate adjustments for foreign currency funding: wholesale funding in foreign currency higher by 10 percentage points; deposits in foreign currency higher by 5 percentage points. ### LCR and NSFR Calibration (selected exact parameters) - Table 11 (LCR Calibration) examples (Baseline / Adverse): - Level 2A Assets (Domestic sovereign or central bank debt... regarded as level 2A): 15% / 25% - Demand deposits (Stable deposits) outflow: 5% / 8% - Demand deposits (Less stable retail deposits) outflow: 10% / 15% - Operational deposits generated by clearing, custody, and cash mgt activities outflow: 25% / 50% - Cooperative banks in an institutional network outflow: 75% / 100% - Undrawn loan commitments (household) outflow: 5% / 10% - Valuation changes on Level 2a posted collateral: 15% / 30% - Net derivative cash outflows: 100% / 100% - Inflows: Level 2a assets: 5% / 5%; All other assets: 5% / 5%; Operational deposits held at other financial institutions: 25% / 25% - Table 12 (Proxy NSFR Calibration) examples (Baseline / Adverse): - ASF Factor 95%: Stable non-maturity (demand) deposits and term deposits ... 95% / 92% - ASF Factor 90%: Less stable non-maturity deposits ... 90% / 85% - ASF Factor 50% (a) Funding with residual maturity < 1 year provided by non-financial corporate clients: 50% / 40% - RSF Factor 5%: Unencumbered Level 1 assets - excluding coins, banknotes and central bank reserves: 5% / 10% - RSF Factor 65% (a) Unencumbered residential mortgages with residual maturity of one year or more and risk weight <= 35%: 65% / 75% - Off-Balance Sheet RSF Factor 5%: Irrevocable and conditionally revocable credit and liquidity facilities: 5% / 10% ### Liquidity Stress Test Results — Key Findings - LCR test: - Under baseline: aggregate LCR (in all currencies) is above 100 percent for all banks. - LCR in SEK for three banks is below 100 percent, but shortfall compensated by buffers in convertible currencies (USD, EUR). - When off-balance sheet items not included in the Swedish LCR are taken into account, the aggregate LCR of one bank falls below 100 percent. - Under adverse scenario (larger run-offs/haircuts), three banks would fall below 100 percent. - NSFR and medium-term findings: - Proxy NSFR suggests most banks would fall below 100 percent in the stress scenario. - Basel III NSFR numbers for all banks are almost 100 percent in baseline. - Authorities noted banks are prioritizing NSFR and have a strategy to be NSFR-compliant ahead of expected EU NSFR requirement of January 1, 2018. - In larger shock scenarios, all banks would not be able to maintain a stable funding profile relative to assets and off-balance sheet activities. - Implied cash flow analysis: - Two banks would have medium term funding problems if faced with severe and long-lasting liquidity and market shocks (1 month–12 months). - Main cause: large outflows relative to short-term paper issued in foreign currency and assumption that bond holders do not roll-over certain bonds at maturity. - Some banks show resilience due to large buffers of liquid assets providing large inflows—around 20-25 percent of total inflows mature within 3 months, mostly related to exposures to central banks. ### Liquidity Exercise Recommendations - NSFR: - Authorities encouraged to collect granular data on available and required stable funding based on Basel III NSFR for all significant currencies to calculate and monitor Basel III NSFR. - Maturity ladder exercise: - Authorities should improve maturity ladder exercise by including features of the 2016 FSAP cash flow stress test. ### Network Analysis (Contagion) — Setup and Results - Approach: - Based on Espinosa-Vega and Solé (2010) work; simulated credit and funding shocks within a network of four largest banks. - Default defined using regulatory hurdle rate (supervisory hurdle rate used in sensitivity analysis). - Credit shock: domino effects triggered if each of the four banks defaulted one at a time on credit commitments. - Funding shock: default leads to liquidity squeeze for banks funded by the defaulting bank; credit shock compounded by funding shock and fire-sale losses. - Dataset included unsecured lending (deposits, overnight loans), securities (mainly covered bonds on net basis), and derivatives reported end-2015; included cross-holdings of repo agreements and equities; repo instruments and equity excluded for confidentiality reasons. - Parameters: LGD, share of funding not replaceable, and fire-sale discount set to 25 percent in baseline; doubled to 50 percent in sensitivity analysis. - Baseline results (parameters = 25 percent; regulatory hurdle): - No contagion chains among the four banks. - Capital losses borne by each bank ranged between 1 percent and 9.5 percent of bank’s CET1 capital. - Sensitivity (parameters = 50 percent): - Second-round defaults not triggered; losses increased to 3.5–22 percent of CET1 capital. - Second-round defaults triggered if supervisory hurdle rate used to define default with same doubled parameters; losses amounted up to half of CET1 capital in that scenario. - Caveats and limitations (likely lead to underestimation of contagion risks): - Not all interbank transactions included (repo instruments and equity not part of exercise). - Unable to assess contagion feedbacks from other financial sector segments in Sweden or abroad (e.g., default of an insurance company, common funding sources). - Liquidity pressures from replacing covered bonds of a defaulted bank by mortgages not captured; liquidity profile changes when taking on illiquid mortgages for covered bonds not included. - Recommendations for contagion analysis improvements: - Collect all interbank exposures and exposures between different financial institutions, including collateralized and uncollateralized and on- and off-balance sheet items, including all types of derivatives. - Consider richer market dynamics in simulations (e.g., counterparty risk in FX swaps, changing liquidity profiles when taking mortgages for covered bonds). - Use market-price based network analysis to extract interconnectedness risks from market prices. ### Solvency Stress Tests of Insurance Companies — Scope and Scenarios - Regulatory basis: Solvency II implemented in 2016; reference date for ST exercises: January 1, 2016. - Coverage: - Seven insurance companies participated. - Life sector coverage: 78 percent by including four large undertakings. - Non-life sector coverage: 53 percent by including three undertakings (below IMF target of 70 percent). - Scope: stress test run at highest consolidation level; 99 percent of technical provisions in life sector are domestic exposures; non-life sector: 36 percent of technical provisions related to Sweden. - Scenarios: - Scenario 1: EIOPA “double-hit” scenario (solo basis expanded to consolidated level). - Scenario 2: IMF Adverse scenario aligned with banking sector macrofinancial scenario; for insurance test all shocks assumed instantaneous at start of first year. - Market risk shocks included: risk-free interest rate, equity and property prices, corporate and sovereign credit spreads, external value of Swedish krona. - Mass lapse event: assume 20 percent of insurance policies for which discontinuance would increase technical provisions are discontinued. - Additional sensitivity analyses requested (stand-alone; not additive to scenario results): - Credit spreads of Swedish covered bonds increase by 500 basis points (life and non-life). - Life insurers: longevity (permanent 20 percent decrease in mortality rates), mortality (permanent 15 percent increase), pandemic (temporary 35 percent increase in disability-morbidity for 12 months). - P&C insurers: repeat of winter storm Gudrun (Erwin) and hurricane Andrew events, with reinsurance recoveries from participant’s five largest reinsurers. ### Insurance Stress Test Market and Credit Shock Specifications (selected exact entries from Table 13) - Risk-free (swap) rate shocks (Scenario 1 / Scenario 2): - SEK, 1 year: -60bp / -33bp - SEK, 10 years: -63bp / +32bp - EUR, 1 year: -60bp / -33bp - EUR, 10 years: -61bp / +32bp - USD, 1 year: -61bp / -33bp - USD, 10 years: -63bp / +32bp - Stock prices: - Sweden: -28.4% / -25.4% - Other countries: -33.4% / -25.4% - Property prices: - Sweden, residential: -4.6% / -31.5% - Sweden, commercial: -4.2% / -31.5% - Other countries, residential: -6.7% / -25.0% - Other countries, commercial: -6.0% / -25.0% - Covered bond spreads: - Sweden: +20bp / +78bp - Corporate bond spreads (Scenario 2 examples): - AAA: +50bp - AA: +80bp - A: +120bp - BBB: +180bp - BB: +300bp - B or lower: +300bp - Unrated: +200bp - Sovereign bond spreads (Scenario 2 examples): - Sweden: +136bp - AAA e.g. +152bp for DE: +50bp - AA e.g. +172bp for FR: +80bp - A e.g. +196bp for LV: +120bp - BBB e.g. +226bp for IT: +180bp - BB: +300bp - B or lower: +300bp - Equity, subordinated bonds, uncollateralized derivative exposure (Scenario 2): - Equity: -100.0% - Subordinated bonds: -100.0% - Uncollateralized derivative exposure: -100.0% - Unsecured bonds: -50.0%; Secured bonds: -15.0%; Deposits: -15.0%; Collateralized derivative exposure: -15.0% - Exchange rates: External value of SEK in Scenario 2: -13.9% ### Insurance Stress Test Capital Standards, Modeling and Top-down Benchmark - Main output: effect on available own funds eligible for coverage of SCR; recalculation of SCR after stress encouraged (optional). - Companies used Solvency II Standard Formula (no approved internal models among participating firms). - Risk-mitigating mechanisms (profit sharing, deferred taxes, LTG measures) reported separately; management actions only included if non-discretionary rules existed at reference date. - Undertakings requested to provide three-year projection under Scenario 2 for gross written premiums, gross claims, lapse rates, investment returns, net earnings, gross technical provisions, own funds, and SCR. - Top-down benchmark: - Used granular breakdown of investment assets (geographical breakdown of sovereign holdings, rating distribution of corporate bond portfolio, durations and coupon rates). - Cash-flow projections for upcoming 60 years in line with Solvency II contract boundaries. - TD stress applied scenarios to cash flows to recalculate technical provisions after interest rate shock and to investment assets; liquidity drain from mass lapse (life) and catastrophic event (P&C). - Asset allocation kept stable (rebalancing annually); dividends paid out of net income per historic payout ratios as long as SCR coverage given; no capital increases modeled in TD stress. ### Insurance Stress Test Results — Key Quantitative Findings - Life insurance sector (aggregate): - Pre-stress SCR coverage: 196 percent. - Post-stress Scenario 2 SCR coverage: 132 percent (BU) and 135 percent (TD). - Majority of companies still able to meet SCR. - Non-life sector (aggregate): - Pre-stress SCR coverage: 152 percent. - Post-stress Scenario 2 SCR coverage: 140 percent (BU) and 154 percent (TD). - Capital needs for companies below SCR: - Two companies fall slightly below 100 percent SCR in both BU scenarios. - Aggregate capital needed: SEK 17.7 bn in scenario 1 and SEK 4.0 bn in scenario 2. - These amounts amount to less than 1 percent of Swedish GDP. - Asset valuation effects under Scenario 2: - Aggregated market value of assets of life insurers declines by about 15 percent; liabilities remain broadly unchanged. - Non-life sector asset-side decline: between 5 and 8 percent. - Composition of effect on own funds in Scenario 2 (TD): - For life sector: available own funds decline by 35 percent on aggregate. - Biggest single impact: equity shock—assumed 25.4 percent drop in Swedish stock prices reduces life insurers’ available own funds by 22 percent on aggregate. - Interest rate change effect on assets and liabilities: 8 percent decline in own funds. - Other important factors: default of largest banking counterparty; decline in property prices. - SEK depreciation tends to benefit Swedish insurers (hold more assets than liabilities in foreign currency). - Non-life sector: - Overall decline in own funds: 33 percent. - Relative contributions differ: lower relative investments in equity, property and sovereign funds reduce vulnerability; larger and riskier corporate bond portfolios add vulnerabilities. - Default of the largest banking counterparty has a more prominent effect in non-life sample. - Bond portfolio composition (median/aggregate): - Life corporate bond portfolio (excl. covered bonds): around 58 percent AAA to A; 3 percent below investment grade. - Sovereign holdings (life): Swedish government bonds 53 percent; France, Germany, the United Kingdom and the United States each close to 10 percent. - Durations: sovereign holdings mostly between four and ten years. - Non-life corporate bond portfolio: investment grade assets account for more than three quarters; 4 percent BB or lower. - Sovereign holdings (non-life): 77 percent Swedish; 15 percent other Nordic countries. - Recovery and medium-term projections: - After instantaneous stress, insurers expect investment returns between 2 and 3 percent in 2017 and 2018. - Net income projected around 6 percent of own funds in 2017 and 2018. - Under Scenario 2, stress translates into median investment return of -14 percent and median ratio of net income to own funds of -33 percent immediately post-stress. - Underwriting sensitivities: - Life insurers: longevity shock (permanent 20 percent decrease in mortality) declines own funds by 3 percent; mortality increase (permanent 15 percent) and pandemic (temporary 35 percent morbidity increase) have even smaller effects. - Non-life insurers: repetition of Gudrun windstorm would cause median company a 2 percent loss in own funds; catastrophic events outside Nordic region have marginal sector-wide effects due to limited exposures and diversified reinsurance. - Concentration risk: - Severe deterioration in Swedish covered bonds market (credit spread +500 bp) could cause median life company own funds to decline by 5 percent and median non-life firm own funds to drop by 16 percent. ### Insurance Stress Test Observations on Low-Interest Environment (Low-for-long) - Challenge: prolonged low interest rate environment severe for life insurers due to long-duration liabilities under market-consistent valuation. - Asset rollover risk: - For life insurers included in exercise, 73 percent of all investments with a fixed coupon have remaining maturity of five years or less—these assets pay a coupon of between 3.4 percent on average. - Next 5-year cohort (five to ten years) average coupon: 2.7 percent. - Implication: coupon rates for upcoming maturities are substantially above average guaranteed interest rates (text truncated at end of supplied content). *Source: IMF staff and authorities material as provided in the supplied content unit.* ### 2.1 percent, the current yield environment will make it difficult to find assets with sufficiently high ### cr17309 - 2.1 percent, the current yield environment will make it difficult to find assets with sufficiently high ### Life insurers: product shifts, asset-side risks, and reinvestment pressures - Aggregate product changes led to a decline in guaranteed interest rates through reducing guarantees in new business and increasing sales of unit-linked policies. - Financially stronger companies, particularly mutuals, have shifted from sovereign bonds into corporate bonds and equity. - Going forward, insurers may increase investment yields by: - extending durations of the bond portfolio; - accepting more credit risk; - investing in less traditional asset classes, e.g., infrastructure. - Increasing equity allocations would raise sensitivity to equity prices further. ### Supervisory and stress-testing recommendations for FI (Finansinspektionen) - FI should strengthen its stress testing framework and ensure reporting data completeness and accuracy: - Scenarios used for regular macroprudential stress tests should complement both the Solvency II Standard Formula and companies’ Own Risk and Solvency Assessment and incorporate a multi-period perspective. - If FI considers that the Standard Formula does not adequately capture a company’s risk profile (e.g., high concentration to the domestic banking sector), the need for a Pillar 2 capital add-on should be assessed. - FI should ensure reporting data quality to perform stress tests and analytical work. Not having the full set of Solvency II reporting data for large life insurance companies complicates analysis and should be mitigated by expanding, in a risk-based way, the current use of voluntary reporting. ### Investment funds: growth, asset composition, and liquidity transmission risks - Investment funds’ total net assets increased by 70 percent since the end of 2011. - Funds’ asset allocation (shares of total assets): domestic equity 36 percent; foreign equity markets 36 percent; domestic mortgage bonds 8 percent. - Premium Pension Authority (PPM) assets represent 25 percent of total assets and partially alleviate liquidity risks, but redemption risk remains relevant because redeemed pension money must be reinvested in other investment funds. - Open-ended funds are exposed to redemption risk; large asset liquidations can transmit shocks to other sectors if they lead to fire-sale discounts. ### IMF top-down liquidity stress test: design and scope - Purpose: assess investment funds’ capacity to meet severe redemptions in an orderly manner and markets’ capacity to absorb assets sold. - Shock definition: a one-time, tail event redemption shock using net flow rates on a quarterly basis. - Shock magnitude: the first percentile of net flow rate distribution of all funds of the same style over 2005Q1–2015Q4. - Data cut-off: fourth quarter 2015. - Coverage: funds accounting for about 95% of the asset management industry; Swedish UCITS and Swedish special funds were included. - Two approaches for assumed asset sales to meet redemptions: - Approach 1 (“pro rata”): sell assets pro-rata to keep portfolio weights intact (natural for index funds). - Approach 2 (“waterfall”): sell assets in descending order of liquidity based on LCR haircut hierarchy (asset ranking based on economic reasoning and expert insight). ### Stress test methodology and market absorption metric - Assets sold by funds under stress were aggregated across funds and asset markets. - Market turnover (quarterly market turnover in specific asset markets) was used as a standard metric of available trading liquidity and compared to assets sold. - If assets sold exceeded market turnover, potential liquidity pressure and fire-sale risks were indicated. ### Key quantitative results and caveats - Corporate bond markets may face stress when confronted with tail event redemptions: - Under the pro-rata assumption, the volume of corporate bonds sold by funds under severe stress could be 50 percent higher than the average turnover of corporate bonds. - For other bond types, volumes sold under the tail event shock are smaller than average market turnover under both approaches, reflecting funds’ asset structure dominated by equity. - Domestic stock market turnover is much larger than the equity that would be sold by funds facing severe redemptions. - A caveat: results do not account for liquidity risk materialization in other sectors holding the same assets, which would further reduce market capacity to absorb assets sold (i.e., exercise likely overestimates market capacity). - Example: foreign investors, including funds domiciled abroad, hold 28 percent of Swedish covered bonds—twice as much as investment funds domiciled in Sweden—and could also sell covered bonds in stress. ### Policy implications and recommendations - Authorities are encouraged to develop monitoring tools to continuously assess markets’ susceptibility to extreme mutual fund redemptions. - Liquidity shocks to investment funds can materially affect funding of non-financial corporations and the cost of funding. - As investment funds grow and potentially change investment structures, authorities should incorporate liquidity risk analyses into mutual fund industry oversight. *Source: IMF staff calculations and Finansinspektionen data as presented in the Sweden country chapter (cut-off data fourth quarter 2015).* ### Appendix I. Risk Assessment Matrix ### Appendix I. Risk Assessment Matrix ### Overview - The matrix shows events that could materially alter the baseline (the scenario most likely to materialize in the IMF staff’s view). It reflects staff's views on the source of risks and overall level of concern at the time of discussions with the authorities. - The relative likelihood of risks (in case the baseline does not materialize) is the staff’s subjective assessment of the risks around the baseline (“low” indicates a probability below 10 percent, “medium” between 10 and 30 percent, and “high” between 30 and 50 percent). ### 1. Sharp rise in risk premia with flight to safety — global financial volatility - Likelihood: Medium - Impact: High - Description: - Sharp asset price adjustment and decompression of credit spreads as investors reassess underlying risk and respond to unanticipated changes in growth prospects, Fed policy rate path, and increases in U.S. term premia, with poor market liquidity amplifying the effect on volatility. - For Sweden: a reassessment of Swedish-specific risk, i.e. a reassessment of household risk which would translate into a reassessment of covered bonds risks. This scenario could also be triggered by a fall in housing prices. The scenario could further affect stock prices (with stocks accounting for a large share of insurance investments). - Key possible effects for Sweden: - Renewed stress in global wholesale funding markets would led to liquidity strains for Swedish banks that rely on foreign exchange wholesale funding. - Concerns about covered bonds would impose higher refinancing risks for banks. - In the face of higher volatility, banks would be constrained in their ability to post additional collateral to maintain the bonds’ cover ratios. - High loss rates due to real estate collateral devaluation would put pressure on loan generation and banks’ and MC’s profitability. - Mark-downs of covered bonds would hurt the solvency of banks, life insurance companies and pension funds. - A drop in stock prices would affect insurers given that stocks are a big share of their investments. - Adverse impact could be partially mitigated by safe-haven flows. - Regional extension to Nordic countries: - Likelihood: Medium - Description: - Similar effects as above are extended to Swedish banks’ exposures in the other Nordic countries. - 70 percent of Nordea’s lending and 22 percent of Handelsbanken’s lending goes to Nordic economies. - Higher bank funding costs translate into higher lending rates and curtailed lending, leading to a decline in house prices in the region. Nordea is the second bank in Denmark (after Danske Bank). ### 2. Structurally weak growth in key advanced and emerging economies, including the Euro Area and China - Likelihood: Medium - Impact: /High - Description: - Euro Area/Japan: Weak demand and persistently low inflation from a failure to fully address crisis legacies and undertake structural reforms, leading to low medium-term growth and accumulation of financial imbalances. - Sweden, Nordic, and Baltic countries: Lower growth in AE and in particular Euro Area would affect exports of Sweden and other Nordic and Baltic countries. This scenario is likely to be combined with low interest rates. - Key possible effects: - Weaker GDP growth and higher unemployment would increase NPL and lead to higher loan loss impairment, weighing on banks’ profitability. - Life insurance companies and pension funds would face difficulties in attracting long term savings in an environment of low interest rates; servicing contracts with guaranteed interest rates would weigh on profitability. ### 3. Economic fallout from political fragmentation in Europe (post-Brexit, migration, populism) - Likelihood: Medium - Impact: (described qualitatively) - Description: - The UK is an important trading partner (Sweden exports over 2 percent of its GDP to the UK). - Sweden is a small open economy highly dependent on unrestricted movement of labor, goods and services. - Key possible effects: - Uncertainty during post-Brexit negotiations could weigh on confidence and investment. - Renewed large scale refugee inflows would increase spending and support activity but would further strain capacity to receive and integrate migrants, raising unemployment and undermining social cohesion. - Higher barriers to trade would dampen exports and investment and weaken the growth outlook. - Weaker GDP growth and higher unemployment in Europe would increase NPL and lead to higher loan losses in banks with cross-border exposures. ### 4. Significant house price decline in Sweden - Likelihood: Medium - Impact: Large (descriptive) - Description: - High house prices largely reflect demographic, balance sheet, and interest rate factors driving up demand faster than supply. - Price levels remain high despite the recent moderation, but the slow reduction in supply shortfalls mitigates downside risks. - Key possible effects: - Large impact on consumption and employment lowers growth. - Loan quality impacted, primarily of firms serving domestic market. - Lending could be curtailed if doubts about the quality of covered bonds rise, elevating bank funding costs. - Mortgage loans are funded with covered bonds. ### Definitions and measured exposures cited in the matrix - Likelihood categories: “low” indicates a probability below 10 percent, “medium” between 10 and 30 percent, and “high” between 30 and 50 percent. - Trade exposure note: Sweden exports over 2 percent of its GDP to the UK. - Nordic bank exposure statistics: 70 percent of Nordea’s lending and 22 percent of Handelsbanken’s lending goes to Nordic economies. *Source: Appendix I. Risk Assessment Matrix (IMF staff).* ### 3. Tail shocksScenario ### 3. Tail shocksScenario ### Scenario design and macro assumptions - Baseline: EBA baseline scenario; WEO baseline scenario as of April 2016 and EBA baseline (where noted). - Stress: EBA stress scenario reflecting Sweden-specific risks and spillovers from a recession in the Nordic and Baltic region; Sweden-specific stress scenario driven by a confluence of shocks producing a deep recession with a slowed recovery (overall GDP profile somewhat “L-shaped”). - Stress scenario macro outcomes (Sweden-specific): - unemployment rate rose by a 6 percentage point rise over a four period. - cumulative growth rate of real GDP was equal to 7 percent over the first three years (GDP growth rates were negative for three years). - equity prices fell by 25 percent in the first year. - house prices declined by 35 percent over the first three years. - Supplementary scenarios: Sweden-specific macro scenarios supplemented with a set of scenarios for the Nordic and Baltic region. - Alternative stress specification: EBA stress scenario with more severe interest rate and exchange rate shocks. ### Market and one-time add-on shocks (Top-down by FSAP Team) - Interest rate risk in the banking book: - steepening of the yield curve depending on currency (e.g. 100 bps widening in the short end of the curve; 350 bps widening in the long end of the curve for Sweden). - Market risk shocks on Trading, AFS securities and CVA: - Equity index shocks: OMX by 40 percent, S&P 500 by 30 percent, Stoxx 50 by 50 percent, MSCI Asia (ex Japan) by 40 percent, Nikkei 225 by 30 percent. - Currency valuation shocks: 40 percent depreciation of the SEK against the U.S. dollar, 20 percent depreciation against the euro, 30 percent depreciation against GBP. - Commodity price decline: energy, base metals, precious metals and grains by 40, 40, 25 and 30 percent respectively. - interest rates (depending on the currency and maturity) and credit spreads (depending on exposure) increase. - Counterparty default shock: top 10 exposures. - Shutdown of foreign exchange swaps market: impact on net open position and capital, CCR RWAs and credit risk RWAs. - Low interest rates for a long time. - Constant loan supply versus dynamic forecast of credit. - Less severe interest rate shock in the baseline. - Corporate sector stress test. ### Risks assessed and parameter calibration - Credit risk: - Banks’ internal models based on stressed point‐in‐time PD and LGD parameters and grade migration; prescribed loss parameters for sovereign exposures. - Credit risk assessed for households, corporates, sovereign, financial institutions exposures estimated according to Basel III framework, under IRB approach. - For corporate exposures (and retail SME) PD modelled using EDFs for all country exposures, except for the Baltic countries. PDs for household exposures based on the level of indebtedness and the level of unemployment. Model estimates subject to expert judgment. LGDs mostly based on expert judgment. - Projected point in time PDs for each asset class (and country) calculated as projected loan losses for each asset class/(LGD x projected exposures by asset class). - Downturn LGDs provided by banks—stayed constant at 2015Q4 level. - Point in time PDs and downturn LGDs used for both credit losses and stressed RWA calculations. - Market risk, CCR and CVA: - Full revaluation of the trading and AfS/FVO portfolio; default of the two most vulnerable of the 10 largest stressed CCR exposures. - Market RWAs and Operational RWAs were not considered in some approaches; market risk RWAs based on a common set of stressed market parameters calibrated from the macroeconomic scenario and historical experience; VaR constant in the baseline and replaced by SVaR in the adverse; Stressed IRC and CVA capital requirements. - Operational risk: - Banks’ own projections for AMA, basic approach and standard approach. - Taxes: - Bottom-up: common simplified tax rate of 30%. - Authorities: taxes set at the pooled average tax rate over the last 20 years. - Calibration of risk parameters: - Banks used their models rather than benchmarks to determine stressed PD and LGD parameters. - Banks employed a rating transition matrix‐based approach, considering effects of PD/LGD grade migration on defaults and impairments. - Projected losses distributed across asset classes and countries; RWAs floored by 2015 value in some top-down approaches. - Top-down adjustments translated downturn LGDs, changes in through-the-cycle PD (adjusted for projections of point-in-time PDs), changes in asset correlation, maturity adjustment parameter and exposures (also adjusted for depreciation of SEK) into stressed RWAs. ### Behavioral adjustments and balance sheet assumptions - Bottom-up banks: - Dividend, fees and commission remained constant in the baseline; in stress minimum of the ratio to total assets of 2015 and the average of the 2 years with the smallest value that occurred 2011‐2015. - For dividends paid: pay‐out ratio based on publicly declared dividend policies; if no policy available, baseline pay‐out ratio = maximum of 30% and median of pay‐out ratios in profitable years 2011‐2015; in the adverse, same amount of dividends assumed (0 for loss‐making banks). - Static balance sheet assumption; asset disposals and acquisitions over time not considered; portfolio composition remained unchanged over time, with maturing exposures replaced with similar ones. - Authorities/top-down: - If after tax net profit was positive, a dividend ratio of 75 per cent was assumed. - Static and dynamic balance sheet assumptions analyzed; dynamic case modeled balance sheet growth and funding growth using a panel regression with fixed effects and macro variables as exogenous variables. - Dividend payout schedule followed capital conservation rule; banks could distribute maximum dividend amount equal to dividend payout ratio in the base-year if not capital constrained; dividends paid only if bank records profits. ### Regulatory and market-based standards, capital and RWAs - Capital standards and metrics: - Basel III capital used; metrics include Tier 1 common capital ratio, common equity tier 1 ratio, Tier 1 capital ratio, total capital ratio and the leverage ratio; all ratios reported on a transitional and a fully loaded basis. - Top-down: CET1 capital, leverage ratio; hurdle rate: regulatory hurdle rate (Basel III regulatory minimum, other Pillar 2 own-fund requirements associated with pension risk, concentration risk and interest rate risk in the banking book, microprudential mortgage floors) and local supervisory requirements accounting for buffers (systemic risk surcharge, the countercyclical capital buffer, and macroprudential mortgage floors). - 100 percent of AOCI phase out from CET1 capital from 2016 onwards; 100 percent phase in factor on deductions from CET1. - Risk-weighted assets: - CRR requirements based on stressed PD and LGD parameters; prescribed increase in REA for securitization exposures; RWAs floored by 2015 value. - Authorities/top-down: risk weights went up by 7.5 percent per year for the first three years and 0 afterwards. - Top-down: RWAs for market and operational risk taken from the banks, reported for the BU test. - For top-down: risk-weighted assets—credit risk using Basel II, IRB formula translating downturn LGDs, changes in through the cycle PD (adjusted for projections of point-in-time PDs), changes in asset correlation, maturity adjustment and exposures (also adjusted for depreciation of SEK) into stressed RWAs. - Market RWAs and operational RWAs treatment varied across approaches; prescribed haircuts for AFS/FVO sovereign positions; RWAs for IRC and CVA floored by the increase for IRB REA; prescribed simplified approach based on historical NTI volatility for HFT. ### Reporting format and output presentation - Distribution of CET1 ratios/leverage ratios by bank in the baseline and stress scenario. - Contribution to the change in system wide CET1 ratio in the baseline and stress scenario. - Contribution of each component of the income statement to change in profits in the baseline and stress scenario. - Evolution of sectoral losses (also by countries). - Evolution of PDs. - Number of banks and share of total assets below hurdle rates. - Capital shortfall under each scenario in nominal terms and relative to GDP. - Results of sensitivity analysis. ### Banking sector: liquidity risk (Top-down by FSAP Team) - Institutional perimeter: - 4 largest banks included. - Market share: 75 percent. - Data and baseline date: Supervisory data (COREP, FINREP); consolidated banking group as of 2015Q4. - Methodology: - Swedish LCR (“old” Basel III version) by currency. - Proxy NSFR by currency. - Cash flow analysis using maturity ladder by currency. - Shocks and calibration: - Shocks reflected in adjustment factors (haircuts and run-off rates) applied to high-quality liquid assets/counterbalancing capacity and outflows; factors informed by the Basel III liquidity metrics (baseline scenario) and more severe episodes of market and funding risks (stress scenario). - Baseline scenario: haircuts and run-off rates calibrated based on the LCR/NSFR. - Stress scenario: stressed LCR and NSFR. - Regulatory thresholds: - For LCR and NSFR threshold set to 100. - For maturity ladder based on survival horizon. - Outputs: - Distribution of LCRs and NSFRs by currency, by bank. - Survival period of each bank in the baseline and stress scenario. - Drivers of banks’ liquidity position and high quality liquid assets/counterbalancing capacity, for each scenario. ### Banking sector: spillover risks (Top-down by FSAP Team) - Institutional perimeter: - 4 largest banks included. - Market share: 75 percent of banks’ assets. - Data baseline date: 2015Q4. - Methodology: - Espinosa and Sole (2013) network analysis. - Tail shocks: - Credit (default of each bank) and funding shocks. - Risks: - Contagion risk. ### Insurance sector: solvency risks - Institutional perimeter: - 4 life insurers and 3 non-life insurers included. - Market share: Life: 78 percent (gross written premiums); Non-life: 53 percent (gross written premiums). - Consolidation level: Groups’ worldwide consolidated business (if applicable). - Data: Companies’ own data, mainly based on regular supervisory reporting. - Baseline date: 01/01/2016. - Methodology and horizon: - Balance sheet-based approach. - Companies’ asset and liability cash flow projections (60 years). - Stress test horizon: participants provided projections for a three-year horizon (2016-2018) assuming unchanged interest rates, risk premia and asset prices after the shock has occurred; asset shocks assumed to occur instantaneously. - Top-down: long-term projection for asset-liability mismatches; asset shocks assumed to occur instantaneously. - Scenarios and sensitivity: - EIOPA “double-hit” scenario: low risk-free interest rates and increase in risk premia. - IMF stress scenario: broadly in line with macrofinancial scenario used in the banking ST. - Top-down low-for-long interest rate scenario. - Sensitivity analyses (bottom-up examples): 500 bps increase in Swedish covered bond spreads; Longevity shock: Permanent 20 percent decrease in mortality rates (life insurers only); Mortality shock: Permanent 15 percent increase in mortality rates (life insurers only); Pandemic event: Temporary 35 percent increase in disability/morbidity rates for the next 12 months; Catastrophic event (non-life insurers only): Windstorm (repetition of windstorm Gudrun, 2005) or US hurricane (repetition of hurricane Andrew, 1992). - Risk factors and shocks (Bottom-up / Authorities / Top-down comparators): - EIOPA “double-hit” shocks: Interest rate shock: 1-year SEK risk-free rate: -60 bps; 10-year: -63 bps. Equity shock: -28.4 percent for Swedish stocks. Property shock: -4.6 percent for Swedish residential real estate; -4.2 percent for Swedish commercial property. Sovereign credit spread: +141 bps for 10-year Swedish government bonds. Corporate credit spread: Increase between 16 bps (for AAA financials) and 516 bps (for unrated financials). - IMF stress scenario shocks: Interest rate shock: 1-year SEK risk-free rate: -33 bps; 10-year: +32 bps. Equity shock: -25.4 percent (domestic and global). Property shock: -32.5 percent (domestic), -20 percent (global). Sovereign credit spread: Increase in credit spreads of Swedish government bonds by 136 bps. Corporate credit spread: Increase between 50 bps (for AAA) and 300 bps (for BB and lower). Default of largest banking counterparty: 100 percent write-off for equity and subordinated bonds, 50 percent write-off for unsecured bonds, 15 percent write-off for secured bonds. Currency shock: 13.9 percent depreciation of SEK. Mass lapse event: discontinuance of 20 percent of insurance policies for which discontinuance would result in an increase of technical provisions without the risk margin are discontinued. - Aggregation and buffers: - All shocks of the scenarios assumed to occur together and simultaneously; summation of shock effects within each scenario, no diversification effect. - Bottom-up buffers: absorption effect of technical provisions (profit sharing); absorption effect of deferred taxes; absorption effect of “long-term guarantee measures” as part of the Solvency II framework. - Top-down: None. - Behavioral adjustments and management actions: - Bottom-up: management actions limited to non-discretionary rules in place at the reference date. - Top-down: None. - Regulatory and valuation parameters: - Regulatory regime: Solvency II; Solvency I for occupational pension business pursued under Art. 308 of the Solvency II Directive. - Valuation: National GAAP (market-consistent). - Reporting outputs: - Impact on solvency ratios. - Impact on net income. - Dispersion measures of solvency ratios and net income. - Capital shortfall under each scenario in nominal terms and relative to GDP. - Top-down outputs: impact on valuation of assets and liabilities; impact on available own funds and solvency ratios; capital shortfall under each scenario in nominal terms and relative to GDP. ### Asset managers: liquidity risks (Top-down by FSAP Team) - Institutional perimeter: - Largest investment funds divided into different styles included. - Baseline date: 2015Q4. - Methodology: - Comparing redemptions with capacity of market for a particular asset class to be sold in an orderly manner. - Assuming a ranking of assets to be sold to meet redemptions; pro rata approach. - Redemptions applied to individual investment funds of the same style; redemptions calculated by style of investment fund as an average redemption rate across all investment funds of the same style; shock: 1st percentile of redemption rates’ distribution. - Stress test horizon: - One quarter (shock characterized by a run on a fund represented by an assumed redemption rate). - Risks and buffers: - Risk: Liquidity risk (a run on investment fund). - Buffers: Liquid assets, assets sold; capacity of a particular market to absorb sell-off of corresponding asset to meet redemptions (by comparing investment fund’s portfolio of a particular security with market turnover of the same security). - Sensitivity analysis: N.A. in top-down design. _Italic: Source — cr17309 - 3. Tail shocksScenario (PDF chapter) from IMF materials provided in the content unit._ --- _Source: https://www.imf.org/-/media/files/publications/cr/2017/cr17309.pdf_