## _cr11288

## Source details

**Canonical URL:** [_cr11288](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2011/_cr11288.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2011/_cr11288.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2011/_cr11288.pdf.json)

---

### Executive summary
- Sweden recovered from the financial crisis and the authorities have initiated exit measures from crisis response policies.
- Real GDP growth reached 5½ percent in 2010.
- Crisis impact on the banking sector arose from strains in funding markets and cross-border exposures in the crisis-hit Baltic region.
- Banks relied significantly on short-term wholesale markets—partly in foreign currency—to fund mortgage lending in Swedish krona.
- The FSAP Update undertook a financial stability analysis including comprehensive stress-testing of banks’ solvency and liquidity positions.

Key high-level findings:
- Solvency: Stress tests used a baseline and three adverse scenarios (double-dip recessions and slow growth scenarios) with external drivers (derailed global growth affecting Swedish exports) and internal drivers (house-price correction affecting household balance sheets and debt service capacity).
- Liquidity: Stress tests assessed dislocations in funding markets, including debt restructuring in fiscally strained countries, flight to high-quality sovereigns, deposit outflows refinanced by costlier funding, and structural shift to longer-term funding to meet Basel III liquidity requirements.
- Overall: Credit risk stress tests indicate resilience due to high profits and capital buffers and high-quality loan portfolios. Liquidity stress tests reveal weaknesses because of heavy reliance on wholesale funding, particularly short-term, to finance relatively illiquid mortgages in SEK.

### Structure and soundness of the banking sector
- Financial system assets account for 550 percent of GDP.
- Four large banking groups (Swedbank, Nordea, Skandinaviska Enskilda Banken (SEB), and Handelsbanken) account for 65 percent of financial system assets and are systemically important with extensive cross-border activities in the Nordic and Baltic regions.
- Insurance companies (in particular life insurance) account for 16 percent of total assets.
- Loans account for 71 percent of total bank balance sheets; government bond holdings are 5 percent of assets.
- Geographic distribution of lending: Sweden 50 percent; other Nordic countries 28 percent; Baltics 4.1 percent; Other 17.9 percent.
- Loan composition: mortgages account for approximately 40 percent of total loans; loans to property developers account for approximately 20 percent of total loans.
- Commitment: committed facilities for trade finance are the bulk of off-balance sheets; exposures to special purpose vehicles (SPVs) are limited.

Funding and liquidity profile:
- Sector loan-to-deposit ratio is 200 percent.
- Banks’ mortgage lending is mostly funded through the covered-bond market (mostly in Swedish krona (SEK)); a considerable part of issues were held as repo-collateral at the Riksbank (RB).
- Approximately 25 percent of foreign funding is swapped to SEK to lend to Swedish borrowers.
- Interest-rate and cross-currency swaps are used to hedge interest rate and exchange rate risks associated with FX borrowing.
- Wholesale market reliance is considered amongst the highest in Europe.

Capitalization and profitability:
- Aggregate capital adequacy ratio (CAR) of the four major banks is 12 percent as of September 2010.
- Capital is predominantly common equity Tier 1, strengthened by rights issues in late 2008 and 2009; retained earnings continued to support capitalization.
- Capital-to-assets ratio is 4.5 percent.
- QIS results suggest the four largest banks are in compliance with Basel III capital requirements; data on common equity Tier 1 ratios based on new Basel capital and RWA definitions suggest Swedish banks are well prepared for Basel III.
- Banks’ profits rebounded from 2009 lows; provisions booked in 2009 were largely reversed in 2010 for some banks.

Asset quality and market indicators:
- Nonperforming loans (NPL) ratio is 2 percent as of September 2010.
- Highest credit risk is in corporate lending, mostly commercial real estate (real estate developers); mortgage lending has historically exhibited low default rates.
- Market indicators (share prices, CDS spreads, joint probability of default, Banking Stability Index) point to decreasing risk since the height of the crisis, though banks with high Baltic exposure lag behind.

### Top-down banking stress tests: scope and methodology
- Coverage: credit, liquidity, contagion, and concentration risks for the four largest banks accounting for 90 percent of banking sector assets.
- Two complementary modelling approaches:
  - Mission-devised model: features from the July 2010 CEBS stress test methodology using bank-by-bank supervisory data.
  - RB’s credit-risk model: adapted for FSAP, using Moody’s KMV market-implied data.
- Differences in quantitative results stem from data sources, satellite models translating macro scenarios into bank losses, methodologies for loan losses and RWA, and assumptions on payout ratios and balance-sheet growth. The two approaches converged on overall solvency conclusions.

Macrostress components:
- Risks modeled: (i) credit risk; (ii) market risk from repricing of sovereign bonds due to upward shift in long-term yield curve; (iii) funding cost risk from run on deposits and dry-out in wholesale funding markets.
- Credit, sovereign, and funding risks were modeled as functions of macroeconomic scenarios devised by the FSAP mission and the RB.

### Macroeconomic scenarios (credit-risk stress tests span 2011–2015)
- Baseline: in line with October 2010 WEO projections for GDP growth of 2.56 percent in 2011 and 3.00 percent in 2012, with conservative projections for unemployment and interest rates.
- Adverse Scenario 1 (double-dip): deviations in GDP growth by 2.7 percent in 2011 and 3.2 percent in 2012 compared to the WEO baseline; includes a confidence shock to global demand and EU-specific yield-curve shock; linear return to baseline assumed for 2013–2015.
- Adverse Scenario 2 (more severe double-dip): assumes twice the deviations from baseline for all variables except long-term interest rates; for long-term interest rates only the country risk increase was doubled.
- Adverse Scenario 3 (prolonged low growth): calibrated to illustrate combined effect of (i) external demand shock; (ii) house-price crash inducing decline in domestic consumption; and (iii) oil-price shock; calibrated to be similar to the 2008/09 downturn but with a more persistent path and no return to baseline in 2015. RB calibrated this scenario using their DSGE RAMSES model.

### Key stress-test findings and vulnerabilities
- Solvency:
  - Credit risk stress tests indicate banks should be able to maintain adequate capital under severe macroeconomic shocks due to high profits, capital buffers, and high-quality loan portfolios.
  - Under adverse Scenarios 1 and 2, capital adequacy ratios remain above regulatory minimums: above the 4 percent Tier 1 regulatory minimum, the 6 percent CEBS Tier 1 threshold, and the 8 percent total CAR regulatory minimum.
  - Under the prolonged low-growth scenario no bank is projected to breach the regulatory minimum for either Tier 1 or total CAR in any given year under both models.
  - Basel III assessment: common Tier 1 capital ratios remain above the minimum required under Basel III using new common Tier 1 capital and RWA definitions and applying phased-in deductions from 2014–15.
- Liquidity vulnerabilities:
  - Heavy reliance on wholesale funding, particularly short-term, to finance relatively illiquid mortgages in SEK exposes banks to dislocations in wholesale and swap markets.
  - Short-term liquidity ratio (three-month horizon) is significantly less than 100 percent for banks, indicating insufficient liquidity reserves to cover modeled cash outflows.
  - Structural liquidity weakness reflects reliance on nonstable short-term market funding to finance illiquid assets; many mortgages remain on banks’ balance sheets rather than securitized.
- Systemic risks:
  - Performance of mortgage and property companies’ loans is key to banking-sector stability given high and rising household indebtedness, increasing loan-to-value ratios, and large share of housing-related loans.
  - Market-based indicators show reduced tail risk since the crisis peak, but elevated joint default probability and Banking Stability Index persist relative to pre-crisis levels.

Representative numeric results and mechanics:
- Aggregate CAR of four major banks: 12 percent (September 2010).
- Capital-to-assets ratio: 4.5 percent.
- NPL ratio: 2 percent (September 2010).
- Sector loan-to-deposit ratio: 200 percent.
- Approximately 25 percent of foreign funding is swapped to SEK.
- Short-term liquidity test assumptions:
  - Haircut of 50 percent applied to liquidity reserves.
  - Deposit runs: 10 percent on retail and SME deposits; 25 percent on corporate deposits.
  - 50 percent of issued securities maturing within three months assumed not to be refinanced.
  - 10 percent of committed credit facilities assumed to be drawn.

### Liquidity, contagion, and concentration stress-test results
- Liquidity:
  - Short-term liquidity ratio formula: Adjusted liquidity reserve / (stressed month cash flow) = SLR.
  - Banks lag in both short-term and structural liquidity metrics behind other European banks.
- Contagion:
  - Contagion stress tests evaluated interbank exposure defaults across unsecured lending, securities, and derivatives.
  - Banks’ largest interbank exposures were securities holdings.
  - Banks could withstand the default of another major bank: the Tier 1 ratio of the most affected bank would fall to about 8 percent.
  - Joint default of the three largest exposures would result in a bank reaching a low 4.4 Tier 1 ratio.
  - Market-based Banking Stability Index indicates failure of any one of the four banks is likely associated with failure of another bank.
- Concentration (assumed LGD of 60 percent) — Tier 1 Ratio (In percent):
  - Bank 1: 5.6
  - Bank 2: 4.4
  - Bank 3: 10.5
  - Bank 4: 8.3

### Methodology details and technical notes
- Two models translated scenario conditions into bank losses:
  - Mission model based on CEBS methodology and supervisory data.
  - RB model based on Moody’s KMV market-implied data and satellite models.
- Sovereign risk losses: assumed common upward shifts in long-term yield curves of Eurozone countries, the United Kingdom, and the United States; common shift compounded by country-specific additional upward shift; shocks phase out linearly in 2013–2015; bond haircuts calculated via discounted cash flow (DCF) methodology using representative five-year bonds.
- Profits and funding-cost mechanics:
  - All market debt maturing within three months assumed refinanced at higher cost using short-term interest rate shocks.
  - Additional funding cost not assumed to be entirely passed through to borrowers.
  - Additional funding cost added to interest expense and deducted from operating profits.
  - Profit projections based on market consensus; haircuts applied to net commissions and other income.
- Dividend payout assumptions:
  - Mission model assumed zero payout.
  - RB model used banks’ historical and disclosed target payout ratios ranging between 40 percent and 50 percent for the four largest banks.

### RB credit-risk model, satellite models, and adjustments
- RB uses market-implied Moody’s KMV EDFs for listed firms and supplements with supervisory proxies (NPL ratios) for Baltic and household exposures.
- PD and LGD treatment:
  - Stress PDs estimated via RB satellite models; LGDs held constant in RB approach.
  - Ad hoc RWA increases applied to capture risk migration: 3 percent for adverse Scenario 1, 4 percent for adverse Scenario 2, 5 percent for adverse Scenario 3.
- Satellite model structure (summarized):
  - Corporate PDs: reduced-form VAR (Equation (7)).
  - Household PDs: projection on unemployment changes (Equation (8)).
  - Baltic PDs: dynamic panel with fixed effects (Equation (9)).
  - Credit-growth satellite models for corporate and retail credit use dynamic panel specifications (Equations (10)–(13)).
  - EAD update equations (Equations (14)–(17)) update exposures after projected credit growth.
- Loan-loss calculation:
  - LLP_it = D_it × L G D_it × EaD_it (notation per source).
  - Total loan losses equal sum across portfolios and geographies after anticipated reversals and expert judgment adjustments.
- Limitations: EDFs are equity-price based and exclude nonlisted firms; supervisory data preferred as initial values despite utility of market-implied data for satellite estimation.

### Stress-testing framework characteristics and caveats
- Basel II floors:
  - AIRB RWA subject to an 80 percent floor of 1988 Basel Accord RWA until end-2011; FSAP maintained these floors throughout the stress period.
- Static approach:
  - Stress-testing is static; results are comparative static relative to base year 2010: Q3 and do not compound across years. The outcome is effectively driven by the single worst year.
- Multipliers on PDs and LGDs are independent of initial levels of reported PDs and LGDs; expert judgment used to fine-tune results to avoid penalizing more conservative banks.
- Geographic allocation:
  - Mission model lacked explicit geographic allocation; RB model addressed geographic allocation as advantage but has market-data limitations.
- Sovereign haircut methodology:
  - Bond price changes for representative five-year bonds computed and aggregated to sovereign haircuts; sovereign bond losses per bank estimated using exposures as of March 2010; sovereign-exposure gains ignored.

### Policy recommendations and supervisory actions
- Monitoring and data collection:
  - Regularly collect supervisory data on mortgage and property-company exposures (PDs, LGDs, default rates, NPLs) and conduct regular stress tests on these sectors.
  - Supervisor should verify reported credit risk measures and ensure consistent calculation and reporting across banks.
- Macroprudential policy and LTV cap:
  - The regulatory cap on the LTV ratio of 85 percent is a good start given house price increases, high household indebtedness, and rising LTVs.
  - Authorities should be prepared to tighten the cap if house prices and LTV ratios continue upward. If insufficient, explore other macroprudential measures.
- Liquidity-related measures:
  - Address heavy reliance on short-term wholesale funding for SEK mortgages to reduce exposure to money markets, bond, and swap market dislocations.
  - Extend liquidity stress-testing, incorporate feedback from solvency to funding costs (lower capital pushes up funding costs), and consider longer-term funding structures to meet Basel III liquidity requirements.
- Stress-testing enhancements for RB:
  - Use supervisory bank data for credit and liquidity stress tests once validated.
  - Regularly conduct mortgage loan and housing-price stress tests linking household debt-service stress to banks’ solvency.
  - Introduce market risk stress tests (interest rate and foreign exchange).
  - Expand coverage beyond the four largest banks.
  - Calibrate macro scenarios using macroeconomic models rather than reverse engineering shocks.
  - Model stressed profits and RWA and use them in stress tests.
  - Incorporate linkages between solvency and liquidity beyond the funding-cost channel.

### Selected numeric assumptions and illustrative entries
- Base year of stress test: 2010: Q3.
- Short-term liquidity example calculation (Adjusted reserve and stressed cash outflow):
  - Liquidity reserve decrease 50% -> 400 -> 200 -> 200/200 = 100%
  - Withdrawal private individuals and SMEs: 10% -> 800 -> 80
  - Withdrawal large companies: 25% -> 160 -> 40
  - Issued securities maturing within three months: 50% cannot be refinanced -> 150 -> 75
  - Credit facilities used: 10% -> 50 -> 5
  - Total stressed outflow: 200
- Funding cost and earnings assumptions (In percent) under adverse scenarios:
  - Adverse scenario 1: 5% 2% 95% 5% 20%
  - Adverse scenario 2: 10% 5% 85% 10% 40%
  - Adverse scenario 3: 15% 10% 80% 15% 60%
- Structural liquidity funding and asset factor examples (In percent):
  - Deposits and borrowing from private customers and nonfinancial small corporate customers with remaining maturity < 1 year: 90
  - Deposits and borrowing from nonfinancial large corporate customers remaining maturity < 1 year: 50
  - Loans to private individuals and companies irrespective of maturity: 85
  - Securities factor: 5–50
  - Repos: 5; Reverse repos: 5
  - All other assets, including derivatives: 100
  - Selected funding items listed with factor 0 include Tier 1 and Tier 2 capital; Liabilities with remaining maturity > 1 year; Derivatives; Cash; Loans to financial institutions remaining maturity < 1 year; Insurance assets; Pension assets.

Representative table excerpts for Bank 1 (Table 5 summary, In percent):
- Profits before Loan Losses (2011–2015, baseline grouping): -3.6, -3.7, -3.5, -3.4, -3.4
- Loan Losses (2011–2015): 21.2, 47.5, 30.3, 15.9, 0.0
- RWA (2011–2015): 7.5, 15.8, 10.5, 5.6, 0.0
- Tier 1 ratio (annual entries): 10% 9% 9% 10% 11%
- Total CAR (annual entries): 12% 11% 12% 13% 14%

### Basel II floors and Basel III conservation standards
- Basel II transitional floor: AIRB RWA cannot be less than 80 percent of 1988 Basel Accord RWA until end-2011; this 80 percent floor was included in 2010: Q3 regulatory reported RWA and maintained in stress tests.
- Basel III minimum capital conservation standards (Common Equity Tier 1 Ratio and Minimum Capital Conservation Ratios, In percent of earnings):
  - 4.5-5.125 → 100
  - >5.125-5.75 → 80
  - >5.75-6.375 → 60
  - >6.375-7.0 → 40
  - >7.0 → 0
- When a bank is subject to a 2.5 percent countercyclical requirement — Common Equity Tier 1 Ratio and Minimum Conservation Ratios:
  - 4.5-5.75 → 100
  - >5.75-7.0 → 80
  - >7.0-8.25 → 60
  - >8.25-9.5 → 40
  - >9.5 → 0

*Prepared by Nada Oulidi and Stephanie Stolz, with significant input from staff at the Riksbank (RB) and Finansinspektionen (FI).*

### 1.   The Core Set of Financial Soundness Indicators for Banks, 2003–10 ..............................17

### 1.   The Core Set of Financial Soundness Indicators for Banks, 2003–10 ..............................17

### Executive summary
- Sweden has recovered from the financial crisis and the authorities have already initiated exit measures from crisis response policies.
- Real GDP growth reached 5½ percent in 2010.
- The crisis impact on the banking sector stemmed from strains in the funding markets and cross-border exposures in the crisis-hit Baltic region.
- Banks relied significantly on short-term wholesale markets—partly in foreign currency—to fund mortgage lending in Swedish krona.
- The FSAP Update undertook a financial stability analysis including comprehensive stress-testing of banks’ solvency and liquidity positions.

Key high-level findings:
- Solvency: Stress tests assessed impact of sharp deterioration in the economic environment on loan losses and solvency using a baseline and three adverse scenarios (double-dip recessions and slow growth scenarios). Scenarios included external drivers (derailed global growth affecting Swedish exports) and internal drivers (house-price correction affecting household balance sheets and debt service capacity).
- Liquidity: Stress tests assessed impact of dislocations in funding markets, particularly wholesale, including scenarios of debt restructuring in fiscally strained countries, flight to high-quality sovereigns, deposit outflows refinanced by costlier funding, and structural shift to longer-term funding to meet Basel III liquidity requirements.
- Overall: Credit risk stress tests indicate resilience due to high profits and capital buffers and high-quality loan portfolios. Liquidity stress tests reveal weaknesses because of heavy reliance on wholesale funding, particularly short-term, to finance relatively illiquid mortgages in SEK.

### Structure and soundness of the banking sector
- Financial system assets account for 550 percent of GDP.
- Four large banking groups (Swedbank, Nordea, Skandinaviska Enskilda Banken (SEB), and Handelsbanken) account for 65 percent of financial system assets and are systemically important with extensive cross-border activities in the Nordic and Baltic regions.
- Insurance companies (in particular life insurance) account for 16 percent of total assets.
- Loans account for 71 percent of total bank balance sheets; government bond holdings are 5 percent of assets.
- Geographic distribution of lending: Sweden 50 percent; other Nordic countries 28 percent; Baltics 4.1 percent; Other 17.9 percent.
- Loan composition: mortgages account for approximately 40 percent of total loans; loans to property developers account for approximately 20 percent of total loans.
- Commitment: committed facilities for trade finance are the bulk of off-balance sheets; exposures to special purpose vehicles (SPVs) are limited.

Funding and liquidity profile:
- Sector loan-to-deposit ratio is 200 percent.
- Banks’ mortgage lending is mostly funded through the covered-bond market (mostly in Swedish krona (SEK)); a considerable part of issues were held as repo-collateral at the Riksbank (RB).
- Approximately 25 percent of foreign funding is swapped to SEK to lend to Swedish borrowers.
- Interest-rate and cross-currency swaps are used to hedge interest rate and exchange rate risks associated with FX borrowing.
- Wholesale market reliance is considered amongst the highest in Europe.

Capitalization and profitability:
- Aggregate capital adequacy ratio (CAR) of the four major banks is 12 percent as of September 2010.
- Capital is predominantly common equity Tier 1, strengthened by rights issues in late 2008 and 2009; retained earnings continued to support capitalization.
- Capital-to-assets ratio is 4.5 percent.
- QIS results suggest the four largest banks are in compliance with Basel III capital requirements; data available on common equity Tier 1 ratios based on new Basel capital and RWA definitions suggest Swedish banks are well prepared for Basel III.
- Banks’ profits rebounded from 2009 lows; provisions booked in 2009 were largely reversed in 2010 for some banks.

Asset quality and market indicators:
- Nonperforming loans (NPL) ratio is 2 percent as of September 2010.
- Highest credit risk is in corporate lending, mostly commercial real estate (real estate developers); mortgage lending has historically exhibited low default rates.
- Market indicators (share prices, CDS spreads, joint probability of default, Banking Stability Index) point to decreasing risk since the height of the crisis, though banks with high Baltic exposure lag behind.

### Top-down banking stress tests: scope and methodology
- Stress tests covered credit, liquidity, contagion, and concentration risks.
- Tests covered the four largest banks accounting for 90 percent of banking sector assets.
- For credit risk, both the FSAP mission team and the RB carried out stress tests and cross-checked results; liquidity, contagion, and concentration tests relied on RB’s stress tests due to data confidentiality.

Macro stress tests components:
- Risks modeled: (i) credit risk; (ii) market risk from repricing of sovereign bonds (banking and trading books) due to upward shift in long-term yield curve; (iii) funding cost risk from run on deposits and dry-out in wholesale funding markets.
- Credit, sovereign, and funding risks were modeled as functions of macroeconomic scenarios devised by the FSAP mission and the RB.

Macroeconomic scenarios (credit-risk stress tests span five years, 2011 to 2015):
- Baseline: in line with October 2010 WEO projections for GDP growth of 2.56 percent in 2011 and 3.00 percent in 2012, with conservative projections for unemployment and interest rates.
- Adverse Scenario 1 (double-dip): implies an increase in unemployment and interest rates, assuming deviations in GDP growth by 2.7 percent in 2011 and 3.2 percent in 2012 compared to the WEO baseline. Assumes a confidence shock affecting global demand and an EU-specific shock to the yield curve following a sovereign-debt crisis; translated into a common upward shift in long-term sovereign-bond yields plus country-specific shifts; linear return to baseline assumed for 2013–2015.
- Adverse Scenario 2 (more severe double-dip): assumes twice the deviations from baseline for all variables except long-term interest rates; for long-term interest rates only the country risk increase was doubled; yield calibrations for vulnerable European countries updated based on recent bond-market developments.
- Adverse Scenario 3 (prolonged low growth): developed with RB; calibrated to illustrate combined effect of (i) external demand shock; (ii) house-price crash inducing decline in domestic consumption; and (iii) oil-price shock. Calibrated to be similar to the 2008/09 downturn but with a more persistent path and no return to baseline in 2015. RB calibrated this scenario using their DSGE RAMSES model.

### Key stress-test findings and vulnerabilities
- Solvency: Credit risk stress tests indicate banks should be able to maintain adequate capital under severe macroeconomic shocks due to high profits, capital buffers, and high-quality loan portfolios.
- Liquidity vulnerabilities: Heavy reliance on wholesale funding, particularly short-term, to finance relatively illiquid mortgages in SEK exposes banks to dislocations in wholesale and swap markets. Disruptions could increase funding costs and reduce profits and solvency if costs cannot be passed through to borrowers.
- Systemic risks: Performance of mortgage and property companies’ loans is key to banking-sector stability given high and rising household indebtedness, increasing loan-to-value ratios, and large share of housing-related loans.
- Market-based indicators show reduced tail risk and improved stability since crisis peak, but elevated joint default probability and Banking Stability Index persist relative to pre-crisis levels.

### Appendix and methodological notes (selected)
- Appendices referenced include: I. The Credit Risk Models Used for the FSAP Update; II. Sovereign Haircut Calculations; III. Basel III New Capital Requirements.
- Appendix tables include: geographic and loan portfolio allocation used in Riksbank’s credit-risk models; overview on Basel III minimum capital requirements; individual bank minimum capital conservation standards and variants when subject to a 2.5 percent countercyclical requirement.
- Unemployment benefits cover up to three years (relevant to mortgage default dynamics).

*Prepared by Nada Oulidi and Stephanie Stolz, with significant input from staff at the Riksbank (RB) and Finansinspektionen (FI).*

### 14.      To ensure the robustness of the overall results of the top-down stress tests, the

### To ensure the robustness of the overall results of the top-down stress tests

### Methodology and models
- Two models translated scenario conditions into bank losses:
  - Mission-devised model: includes features from the July 2010 Committee of European Banking Supervisors (CEBS) stress test methodology and uses bank-by-bank supervisory data.
  - RB’s credit-risk model: adapted to the needs of the FSAP, using Moody’s KMV data.
- The two methodologies are complementary (not necessarily comparable) and were used for consistency checks of overall solvency conclusions.
- The FSAP and RB stress tests produced different quantitative results due to differences in:
  - data sources (supervisory data used by FSAP; market implied data used by RB),
  - satellite models translating macro scenarios into bank losses,
  - methodologies used in calculation of loan losses and RWA,
  - assumptions on payout ratios and balance sheet growth.
- The two approaches converged in overall conclusions on banking-sector solvency.

### Sovereign risk losses
- Assumed upward shifts in long-term yield curves of Eurozone countries, the United Kingdom, and the United States (U.S.) to simulate an aggravated sovereign-debt crisis.
- A common upward shift was applied to each country, compounded by a country-specific additional upward shift reflecting fiscal situations and/or market perceptions.
- Shocks were assumed to phase out linearly in the three outer years (2013–15).
- The mission broadly followed a discounted cash flow (DCF) methodology in calculating bond haircuts under the four scenarios.

### Profits and funding costs
- Stress tests assumed a funding cost increase associated with a run on deposits and a dry-out in wholesale funding markets.
- Assumptions and mechanics:
  - All market debt maturing within three months was assumed to be refinanced at higher cost using short-term interest rate shocks from the macro scenarios.
  - The additional funding cost was not assumed to be entirely passed through to borrowers (conservative assumption relative to historical Swedish bank behavior).
  - The additional funding cost was added to interest expense and deducted from operating profits.
- Profit assumptions:
  - Based on market consensus projections because no viable model to forecast earnings was available.
  - Haircuts applied to net commissions and other income.
  - Earnings assumptions summarized in Table 2 (source document).

### Dividend payout assumptions
- Mission model assumed zero payout.
- RB model assumptions:
  - More conservative than Basel III (Appendix I).
  - Based on banks’ historical and currently disclosed target payout ratios, which range between 40 percent and 50 percent for the four largest banks.

### Macro credit stress-test results — key findings
- Credit risk is the primary vulnerability; nonetheless, under both RB and FSAP mission models, banks are found resilient.
- Under adverse Scenarios 1 and 2:
  - Banks’ capital adequacy ratios remain above regulatory minimums despite weak growth, high unemployment, and increased bank losses.
  - Capital ratios remain well above:
    - the 4 percent Tier 1 regulatory minimum,
    - the 6 percent CEBS Tier 1 threshold,
    - the 8 percent total CAR regulatory minimum.
  - Increases in loan losses and RWAs from credit risk are the main negative drivers.
  - Profits and reversals in provisions provide significant buffers.
  - Shocks to sovereign-bond portfolios and funding costs have limited impact due to banks’ limited exposures to vulnerable European countries and pass-through of increased funding cost.
  - Post-shock credit losses are not significant owing to high-quality loan portfolios and low initial values of credit-risk measures.
  - Assuming full pretax profit retention (and even with a 40 percent dividend payout assumed under the RB model), Tier 1 and total capital ratios under both IMF and RB methodologies remain above regulatory minima.
- Under a prolonged low-growth scenario:
  - Low growth and high unemployment persist; bank losses and RWA increases are significant with no recovery in outer years.
  - No bank is projected to breach the regulatory minimum for either Tier 1 or total CAR in any given year under both models.
- Basel III assessment:
  - Similar tests suggest banks’ common Tier 1 capital ratios remain above the minimum required under Basel III using new common Tier 1 capital and RWA definitions and applying phased-in deductions from 2014–15.

### Liquidity risk stress tests — framework and assumptions
- Due to data confidentiality, the mission used RB’s framework; RB conducted the liquidity stress tests and presented results and methodology.
- Two measures analogous to Basel III ratios were calculated:
  1. Short-term liquidity ratio (three-month horizon):
     - Assesses liquidity buffers under stressed scenarios to cover unexpected outflows from wholesale dry-out, deposit withdrawals, and draw-downs of unused credit lines.
     - Calibration assumptions summarized in Table 3 (source document); key assumptions explicitly stated:
       - Haircut of 50 percent applied to liquidity reserves.
       - Deposit runs: 10 percent on retail and SME deposits; 25 percent on corporate deposits.
       - 50 percent of issued securities maturing within three months assumed not to be refinanced.
       - 10 percent of committed credit facilities assumed to be drawn.
     - RB’s short-term liquidity ratio formula: Adjusted liquityreserve / (stressed month cash flow) = SLR.
  2. Structural long-term liquidity ratio:
     - Relates weighted average of liabilities to weighted average of assets (weights based on liquidity of assets and stability of funds for liabilities).
     - Example weights:
       - Assets: cash allocated a zero weight; loans allocated an 85 percent weight.
       - Liabilities: equity capital and liabilities maturing in less than a year allocated a weight of 100 percent; short-term market funding allocated a zero weight.
     - RB’s structural liquidity ratio estimated via weighted sums (formula provided in source).
- Interpretation:
  - Banks with high shares of liquid assets and longer-maturity funding fare better in respective tests.

### Liquidity stress-test results — findings
- Banks lag in both short-term and structural liquidity metrics behind other European banks due to heavy reliance on short-term market funding.
- The short-term liquidity measure is significantly less than 100 percent, indicating insufficient liquidity reserves to cover modeled cash outflows over three months.
- Structural weakness reflects heavy reliance on nonstable, short-term market funding to finance relatively illiquid assets (large share of mortgages kept on banks’ balance sheets rather than securitized).

### Large exposures and contagion risk — framework and results
- RB carried out contagion and concentration risk stress tests due to data confidentiality.
- Contagion stress tests:
  - Evaluated impact of interbank exposure defaults across unsecured lending, securities (gross positions and positions net of risk mitigants), and derivatives.
  - Interbank exposure data not directly available to mission; RB ran the contagion-risk stress tests.
  - Banks’ largest interbank exposures were securities holdings.
- Concentration risk stress tests:
  - Based on banks’ 15 largest counterparty exposures compiled quarterly by RB.
  - Assumed an LGD of 60 percent.
- Results:
  - Banks could withstand the default of another major bank: the Tier 1 ratio of the most affected bank would fall to about 8 percent.
  - The joint default of the three largest exposures would result in a bank reaching a low 4.4 Tier 1 ratio.
  - Broader interconnections suggest more complex spillovers; market-based Banking Stability Index indicates that failure of any one of the four banks is likely associated with failure of another bank.

### Conclusions and policy recommendations
- Overall solvency:
  - The banking sector appears resilient to credit risk stress tests because of high profits, capital buffers, and high-quality loan portfolios.
  - Structural factors supporting mortgage credit quality include nonspeculative mortgage borrowing, full recourse from lenders, generous social and unemployment benefits, and so far stable housing prices.
- Monitoring and data collection:
  - Performance of mortgage and property companies’ loans should be monitored closely by regularly collecting supervisory data on these sectors (e.g., PDs, LGDs, default rates, and NPLs) and conducting regular stress tests on them.
- Macroprudential policy and LTV cap:
  - Given continuing house price increases, high and rising household indebtedness, increasing loan-to-value (LTV) ratios, and the significant share of housing-related loans in banks’ portfolios, the recently introduced regulatory cap on the LTV ratio of 85 percent is a good start.
  - Authorities should continue to closely monitor the situation and be prepared to tighten this cap if house prices and LTV ratios continue upward trends.
  - If the LTV cap proves insufficient, authorities could explore other macroprudential measures.
- Liquidity-related vulnerabilities:
  - Liquidity stress test concerns stem from heavy reliance on wholesale funding—particularly short-term—to finance relatively illiquid mortgages in SEK.
  - Current funding structure exposes banks to dislocations across wholesale market segments (money markets, bonds, swaps).
  - Disruptions could increase funding costs, impacting profits and solvency if costs cannot be passed through to borrowers, particularly if borrowers’ balance sheets deteriorate under high unemployment combined with a sharp and sustained decline in housing prices, potentially forcing banks to deleverage.

*Source: IMF staff summary and RB analyses presented in the source document.*

### 30.      While the RB has made significant strides in its stress-testing capacities,

### _cr11288 - 30.      While the RB has made significant strides in its stress-testing capacities,

### Stress-testing enhancements and recommendations
- In-house methodologies for top-down stress tests are "fairly advanced and appropriate" but "enhancements are warranted."
- Use supervisory bank data for credit and liquidity stress tests once data are validated by the supervisor; if results cannot be published in the financial stability report, use them for internal financial risk assessments.
- Regularly conduct stress tests of mortgage loans and housing prices. Suggested sequence:
  - Test households’ debt service capacity assuming lower housing prices and higher LTV ratios and unemployment rates (or other relevant factors).
  - Link household stress results to banks’ solvency.
- Collect and monitor default data and NPL ratios by sector (in particular, mortgage and property developer loans) and geographical allocation on a regular basis to verify reported credit risk measures (e.g., PDs and LGDs).
- Supervisor should verify reported credit risk measures (for example cross-checked against actual defaults) and ensure consistent methods of calculation and reporting to enhance comparability across banks.
- Introduce market risk stress tests (including interest rate and foreign exchange).
- Expand stress-test coverage beyond the four major banks to include smaller banks.
- Calibrate macroeconomic scenarios based on macroeconomic models as conducted by the FSAP (instead of reverse engineering macro shocks).
- Model stressed profits and RWA and use them in the stress tests.
- Extend the framework linking solvency and liquidity stress tests beyond the current funding cost channel; specifically, incorporate the feedback that lower capital pushes up funding costs.

### Data sources and verification
- UC AB, a private credit registry with granular loan information, "might be used for this purpose."
- Supervisory validation of bank data is emphasized before use in credit and liquidity stress tests.
- Reported PD and LGD data are understood to exclude already-defaulted exposures.

### Scope, scenarios, and selected numeric assumptions
- Base year of the stress test is 2010: Q3.
- Stress-scenario calibration: PDs and LGDs are linked to changes in three macroeconomic variables (real GDP growth, unemployment rate, and long-term interest rate) using CEBS elasticities.
- Short-term liquidity measure example (Adjusted reserve and stressed cash outflow):
  - Liquidity reserve: Decrease 50% -> 400 -> 200 -> 200/200 = 100%
  - Withdrawal of deposits by private individuals and small and medium-sized companies: 10% -> 800 -> 80
  - Withdrawal of deposits by large companies: 25% -> 160 -> 40
  - Issued securities that will mature within three months: 50% cannot be refinanced -> 150 -> 75
  - Credit facilities used: 10% -> 50 -> 5
  - Total stressed outflow: 200
- Assumptions for funding cost and earnings under adverse scenarios (In percent):
  - Adverse scenario 1: 5% 2% 95% 5% 20%
  - Adverse scenario 2: 10% 5% 85% 10% 40%
  - Adverse scenario 3: 15% 10% 80% 15% 60%
- Structural liquidity measure funding and asset factors (selected entries, In percent):
  - Deposits and borrowing from private customers and nonfinancial small corporate customers with remaining maturity < 1 year: 90
  - Deposits and borrowing from nonfinancial large corporate customers remaining maturity < 1 year: 50
  - Loans to private individuals and companies irrespective of maturity: 85
  - Securities factor: 5–50
  - Repos: 5; Reverse repos: 5
  - All other asset, including derivatives: 100
  - Tier 1 and Tier 2 capital; Liabilities with remaining maturity > 1 year; Derivatives; Cash; Loans to financial institutions remaining maturity < 1 year; Insurance assets; Pension assets: 0 (as listed in funding/assets matrix)
- Concentration risk stress-test results (Tier 1 Ratio, In percent):
  - Bank 1: 5.6
  - Bank 2: 4.4
  - Bank 3: 10.5
  - Bank 4: 8.3

### FSAP mission model: methodology and formulas (summary of technical approach)
- The FSAP mission model largely followed the CEBS stress test methodology used in the 2010 CEBS exercise and other recent EU FSAPs.
- Supervisory data by geographical exposure were not available; the RB devised a model accounting for geographical dimension using market-implied Moody’s KMV data.
- Comparative static shocks relative to the base year were calculated for each year using CEBS elasticities.
- Credit and sovereign bond mark-to-market (MTM) losses were estimated each year to reflect cumulative stressed changes since the base year. The methodology did not model the buildup of retained earnings or bank exposures (static approach).
- Notation and inputs used (as described in the source):
  - Base-year bank-specific AIRB PD and LGD values represented by i,0 LGD,PD.
  - RR, M, B, C, S represent total bank exposures to sovereign, corporate, banking and financial institutions, mortgage, and revolving retail loans, respectively.
  - 0 RWA represents a bank’s actual reported initial regulatory total RWA.
  - tt UN,GDP and t LR represent the level of real GDP growth, unemployment rate and long-term interest rate in stress scenario year t.
- PDs and LGDs in stress period t (PDit and LGDit) are linked to base-year PD0 and LGD0 via CEBS elasticities and changes in macro variables (∆UN,GDP, ∆LR, etc.).
- Expected stress losses for each loan portfolio (corporate, banking, mortgage, and revolving retail) were estimated for the base year and each stress year; total stress losses in a year equal the sum of portfolio losses.
- Change in expected losses relative to base year: ΔEL = Et − E0.
- Stress test loan-loss provisions in stress year t: LLPt = LLP0 + ΔELt.
- RWA estimation:
  - RWA for portfolios computed using IRB(PD,LGD,m) functions where m (maturities) are assumed fixed over the stress test.
  - Base year and stress scenario total RWA for corporate, banking, mortgage and retail credit risks summed as: At = IRB(PD,LGD,m) for each exposure type.
  - Cumulative stressed change in RWA relative to base year: ΔRWA = Rt − R0.
  - Estimate of bank RWA used in stress test to estimate regulatory capital ratios: St = R0 + ΔRWA.

### Selected macro-stress test results (high-level outcomes shown in tables/figures)
- Base year and scenario outcome presentation conventions:
  - Profits before loan losses, loan losses, and RWA presented in percent deviations from baseline.
  - Tier 1 ratios and total CARs presented in percent.
- Representative entries from "Table 5. Sweden: Summary of Macro Stress Testing Results, 2011–15 (In percent)" for Bank 1 (rows by item across years):
  - Profits before Loan Losses: -3.6, -3.7, -3.5, -3.4, -3.4 (for 2011–2015 under baseline column grouping)
  - Loan Losses: 21.2, 47.5, 30.3, 15.9, 0.0
  - RWA: 7.5, 15.8, 10.5, 5.6, 0.0
  - Tier 1 ratio: 10% 9% 9% 10% 11% (yearly entries shown)
  - Total CAR: 12% 11% 12% 13% 14%
- Comparable entries are reported for Bank 2, Bank 3, and Bank 4 across Adverse 1, Adverse 2, and Adverse 3 scenarios in the same table format (profitability deviations, loan losses, RWA, Tier 1 ratio, Total CAR).
- Aggregate contributions to changes in the aggregate CAR are decomposed into:
  - Change owing to change in RWA
  - Net profits
  - Revaluation of sovereign exposures
  - Reversals in provisions
  - Change in expected credit losses
- Structural liquidity stress-test results (June 2010) are presented for Bank 1–4 and an "Average European banks" benchmark in percentage terms (figure shown with range roughly between 72% and 94% on the chart).

*Source: Extracted from the provided content of the IMF staff report PDF chapter/section.*

### 35.      The stress tests maintained the Basel II floors throughout the entire stress-

### _cr11288 - 35.      The stress tests maintained the Basel II floors throughout the entire stress-

### Basel II floors and baseline RWA/provisions
- Under current Basel II rules, a bank’s RWA calculated under the AIRB approach are subject to a floor which requires that banks may not report an RWA value that is less than 80 percent of their RWA value calculated under the 1988 Basel Accord rules.
- This transitional floor will remain in place until end-2011.
- The initial stress-test scenario RWA used banks 2010: Q3 regulatory reported RWA, which include the 80 percent floors.
- These floors were kept by the FSAP mission throughout the stress period.
- Equations (3) and (6) estimate stress-scenario loan-loss provisions and RWA as the sum of a bank’s initially reported values plus the changes in expected loss and RWA (respectively), where these changes are calculated relative to the base-year Basel II EL and RWA prescriptions.
- The approach is necessary because bank balance sheet provisions and RWA are not equal to the Basel formula in the initial period, i.e., 00 LLPEL, and 00 A RWARWA.

### Static approach and multiplier characteristics
- The stress-testing approach is static: stress-scenario results do not compound across subsequent years and outcomes are calculated as a series of comparative static results (relative to the base year).
- The stress-test capital adequacy outcome is essentially determined by the single worst year of the test.
- Model multipliers are applied to a bank’s initial PDs and LGDs to produce stress PDs and LGDs.
- These multipliers are determined by the magnitudes of the macroeconomic shocks and the model parameters and are independent of the level of a bank’s reported PDs and LGDs.
- This raises a potential issue: more conservative banks (higher PDs and LGDs) could face relatively larger stress impacts compared with underreporting banks; expert judgment from banking sector experts at the RB and FI was used to fine-tune results.

### Geographic allocation caveat and complementary modeling
- The mission model uses bank-by-bank supervisory data but does not account for geographic allocation of loan exposures.
- The caveat is somewhat mitigated by the predominance of loan exposures in Sweden and other Nordic countries with similar macroeconomic conditions.
- The caveat motivated the use of a complementary model devised by the RB in collaboration with the FSAP team to corroborate solvency results.

### The Riksbank (RB) credit risk model: data sources and portfolio breakdown
- The RB designed a credit risk stress-testing methodology in close collaboration with the mission team.
- The RB bases their stress tests on market-implied Moody’s KMV default data.
- RB arguments for market-implied data:
  - (i) results of stress tests based on supervisory data could not be published for confidentiality reasons;
  - (ii) results based on market-implied data are more comparable across banks.
- RB derives PDs and LGDs from available market-implied Moody’s KMV data for listed companies in the Nordic countries, Germany, and the United Kingdom.
- Loan breakdown used in RB models includes: property companies, other nonfinancial companies, financial institutions, and households.
- Geographic breakdown includes: Sweden, other Nordic countries, Baltic countries, Germany, the United Kingdom, and the rest of the world.
- NPL ratios are used as proxies for corporate loan PDs in the Baltic countries, and household loan PDs in all countries.

### RB satellite models for PDs and LGDs
- Stress scenario PDs are estimated using elasticity estimates from the RB’s in-house satellite models; LGDs are held constant.
- Notation:
  - ,C it PD = corporate loan PDs in the Nordic countries, U.K., and Germany
  - ,H it PD = household loan PDs in the Nordic countries, United Kingdom, and Germany
  - ,B it PD = household and corporate loan PDs in the Baltic countries
- RB satellite models summarized as:
  - Equation (7): ,,1,1,2,,3, CC it i t i t i t i t PD c PD GDP SR        
    - (Equation (7) is a reduced form VAR model.)
  - Equation (8): ,,1, H it i t PDb UN  
    - (Equation (8) is a simple projection where the percentage change in PD is explained by the percentage change in the unemployment rate.)
  - Equation (9): ,,1,1,2, BB it i t i t i t PD d PD GDP C        
    - (Equation (9) is a dynamic panel with fixed effects.)

### RB credit-growth satellite models and EAD projection
- Baseline scenario: credit growth forecasted using RB satellite models.
- Explanatory variables:
  - Corporate lending: investments considered main driving factor.
  - Retail credit growth: disposable income and short-term interest rates considered main explanatory variables.
- Notation for exposures:
  - CI = corporate loan exposures in Nordic countries, United Kingdom, and Germany
  - CB = corporate loan exposures in the Baltic countries
  - HI = household loan exposures in Nordic countries, United Kingdom, and Germany
  - HB = household loan exposures in the Baltic countries
  - ,it I = investments in country i and year t
  - ,it DI = disposable income
  - ,it SR = short-term interest rates
  - i C = crisis dummy
- RB credit growth satellite models summarized as:
  - Equation (10): ,,1,1,2,it ii i ti t CI ca CI a I       
  - Equation (11): ,,2,1,2,,3,it ii i titi t HI cb HI b DI b SR        
  - Equation (12): ,,1,1,2,,3,,4 it ii iti iti iti CB cd CB d I d SR d C         
  - Equation (13): ,,1,1,2,,3 iti iti iti ii HB ce HB e SR e C       
  - (Equations (10)–(13) are dynamic panel models with fixed effects.)
- EAD update equations:
  - Equation (14): ,1, (1) it t CI CI CI     
  - Equation (15): ,1, (1) it t HI HI HI     
  - Equation (16): ,,1, (1) it it CB CB CB     
  - Equation (17): ,1, (1) it t HB HB HB     

### Loan-loss calculation, RWA adjustments, and scenario-specific ad hoc RWA increases
- Using equations (7)–(17), and fixed LGDs, loan losses are calculated for each loan portfolio and geographic allocation as:
  - ,,,it it it LLP D L G D EaD   
- Total loan loss in each bank is estimated as the sum of loan losses in each loan portfolio and geographic allocation after accounting for anticipated reversals and using expert judgment from banking and macro prudential staff at the RB to fine-tune results.
- Loan losses are deducted from RWA at the end of the year.
- Additional risk migration assumed as ad hoc increases in RWA of:
  - 3 percent for adverse Scenario 1
  - 4 percent for adverse Scenario 2
  - 5 percent for adverse Scenario 3

### Model limitations and data considerations of RB approach
- Advantages: model accounts for geographic allocation of loan exposures.
- Limitations:
  - Model is based on market-implied data (Moody’s KMV EDFs).
  - EDFs are based on a Merton-type model assuming efficient markets; EDFs infer PDs from equity prices, which are typically volatile in crisis periods.
  - Nonlisted firms are excluded from EDF-based model, potentially omitting an important share of banks’ loan portfolios.
  - Actual/supervisory bank data should be used as initial values of credit-risk measures in stress tests despite the utility of market-implied data for satellite model estimation.

### Sovereign haircut and bond loss methodology (Appendix II)
- Method: discounted cash flow (DCF) methodology for bond haircuts under four scenarios.
- Representative maturity: five-year bonds assumed representative of banks’ bond holdings.
- Bonds included: those with market quotes from Bloomberg with maturities between 4.5 and 6.5 years.
- Bond price calculation (per year stress year t for bond b):
  - ,bt P = (1/(1+r_t)) * sum over k=1..n of cf_k/(1+r_t)^{k-1} ??? (DCF formula described using coupon payments, coupon frequency, maturity dates, redemption value, and yield-to-maturity under the four scenarios)
- Corresponding haircuts for each bond from changes in bond prices relative to the base year under the three scenarios:
  - 1100 0, , , b tb tb P P P  (where ,0b P is the bond price in the base year)
- Sovereign haircut at each stress date c t haircut calculated as weighted average change in price of all bonds of that sovereign, weights are amounts outstanding:
  - , c t haircut = (sum_b b Amt * , b t P) / (sum_b b Amt)
- Sovereign bond losses per bank estimated as:
  - 0, 1 exp m c t t c c Losses haircut s osures    
  - Where exposures 0,c represent banks’ total sovereign-bond exposures to country c in the banking and trading books as of March 2010.
- Sovereign-exposure gains are ignored for stress-test purposes.
- Note: More recent data on banks’ sovereign bond exposures were not available to the mission.

### Basel III new capital requirements and conservation standards (Appendix III)
- Table 9: Individual Bank Minimum Capital Conservation Standards — Common Equity Tier 1 Ratio (In percent) and Minimum Capital Conservation Ratios (In percent of earnings):
  - 4.5-5.125 → 100
  - >5.125-5.75 → 80
  - >5.75-6.375 → 60
  - >6.375-7.0 → 40
  - >7.0 → 0
- Table 10: Individual Bank Minimum Capital Conservation Standards, When a Bank is Subject to a 2.5 Percent Countercyclical Requirement — Common Equity Tier 1 Ratio (Including other fully loss- absorbing capital) and Minimum Capital Conservation Ratios (In percent of earnings):
  - 4.5-5.75 → 100
  - >5.75-7.0 → 80
  - >7.0-8.25 → 60
  - >8.25-9.5 → 40
  - >9.5 → 0

*Source: _cr11288 - 35.      The stress tests maintained the Basel II floors throughout the entire stress-*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2011/_cr11288.pdf_