## cr17200

## Source details

**Canonical URL:** [cr17200](https://www.imf.org/-/media/files/publications/cr/2017/cr17200.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2017/cr17200.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2017/cr17200.pdf.json)

---

### Overview
- FSAP stress tests assess Bulgarian bank resilience to solvency, concentration, liquidity, and balance sheet contagion risks.
- Coverage: 17 banks, making up 95 percent of banking system assets.
- Simulation horizon: three-year projections (June 2016—June 2019), based on data available through June 2016 and using end-June 2016 bank-by-bank balance sheets, income statements, and prudential reports.
- Both baseline and adverse scenarios incorporate AQR adjustments based on year-end 2015 data.
- Key methodological differences: Top-down (IMF) vs Bottom-up (BNB) approaches, including treatment of interest accrual on NPLs, provisioning rules, profitability of performing loan portfolios, market risk treatment, and funding cost shocks.

### Solvency Stress Tests — Key Findings
- Baseline scenario (modest growth, declining unemployment, stable/low interest rates):
  - Two banks—including a systemic one—exhibit weakness in capital buffers to cope with accumulated losses.
  - IMF ST approach excludes interest income from NPLs in both baseline and adverse scenarios; AQR-driven increase in NPLs reduces the performing loan base that generates cash-based interest income.
  - Result: Two banks do not generate enough recurring income to cover interest expense and credit costs in the baseline, producing negative profits and declining capital levels.
- Adverse scenario (large drop in output; increase in unemployment; decline in collateral prices; increase in risk premiums):
  - Five banks fall below the minimum regulatory CET1 ratio threshold of 4.5 percent.
  - Three banks fall below the threshold by the second year of the stress period.
  - Two additional banks fall below the regulatory minimum in the third year.
  - Aggregate capital shortfall of 1.8 percent of the annualized nominal GDP at the end of the simulation period.
  - Of the five banks below the regulatory minimum at the end, three become technically insolvent.
- Main driver of capital deterioration:
  - Loan loss provisions are the largest driver of the change in capital positions in the adverse scenario.
  - Provisioning increased by 20 percentage points from the assumed minimum provisioning level of 60 percent in the baseline scenario for NPLs, producing large provisioning charges that reduce capital.
  - NPLs increase significantly in the second year as economic growth turns negative; recovery starts in the third year but loss provisions continue to weigh on profitability.

### Solvency Stress Tests — Design and Sensitivities
- Two scenarios: baseline and adverse; time horizon three years (2016 Q2–2019 Q2).
- Baseline scenario linked to IMF WEO October 2016 projections; key baseline projections (In percent):
  - Annual GDP Growth: 3.6 3.0 2.8 2.5
  - Unemployment Rate: 8.9 8.2 7.1 6.9
  - SOFIBOR: 0.13 0.13 0.13 0.13
- Adverse scenario calibration and key path (In percent / basis points):
  - Annual real GDP growth rates: -3 percent (June 2017), -3.6 percent (June 2018), -0.9 percent (June 2019).
  - Unemployment Rate: 10.4 percent (June 2017), 11.5 percent (June 2018), 12.0 percent (June 2019).
  - SOFIBOR shocks relative to baseline: 33 basis points (June 2017), 23 basis points (June 2018), 6 basis points (June 2019).
- Provisioning assumptions:
  - Minimum provisioning ratio of 60 percent assumed in baseline.
  - Adverse scenario provisioning rates uniformly increased by 20 percentage points from baseline levels and capped at 100 percent.
  - In adverse scenario, an additional 20 percentage points of provisions added for loans re-classified as NPLs in the AQR results; these provisions equal 0.2*X million for loans reclassified totaling BGN X million, divided equally over the three years.
- Sensitivity tests (static, June 2016 after AQR adjustments; RWAs assumed constant):
  - Decline in sovereign security prices: aggregate CET1 ratio declines by 0.2 percentage points; no bank falls below regulatory minimum.
  - Interest rate shock (instantaneous upward shift of 500 basis points): aggregate CET1 ratio declines by 1.3 percentage points; one bank falls below regulatory minimum.
  - Increased provisions for existing stock of NPLs (uniform +20 percentage points): aggregate CET1 ratio declines by approximately 6 percentage points; two banks fall below regulatory minimum.

### Credit Risk Outcomes and Satellite Models
- Satellite models developed for three exposure classes: corporate loans; mortgages; other consumer loans; exposures to institutions excluded.
- NPL time series: starts in Q4 2009 and covers 2011–2016.
- Top-down ST projects significant increases in NPLs under the severe stress scenario versus declining NPLs in the baseline.
  - Under the severe stress scenario, the banking system’s average NPLs increase by approximately 8 percentage points between 2016 Q2 and 2019 Q2.
  - Corporate loans and mortgage loans experience the highest increases in NPLs.
- Credit losses over the three-year horizon in the loan book amount to BGN 4.4 billion in the adverse scenario (equivalent to approximately 5 percent of total banking system assets as of June 2016).
- Panel regression example coefficients (selected):
  - Corporate Loans: Dependent Variable(t-1): 0.7293*** (12.87); GDP(t-2): -2.1783*** (-3.92); Sofibor(t-2): 0.0353*** (-4.93); R-Squared: 0.97; Observations: 190.
  - Mortgage Loans: GDP(t-5): -0.7033*** (-2.79); Unemployment(t-2): 0.0841 (14.31); R-Squared: 0.99; Observations: 286.
  - Other Retail Loans: GDP(t-5): -1.7204*** (-3.71); Sofibor(t-3): 0.0956*** (14.64); R-Squared: 0.96; Observations: 198.

### Stress Test Results — System Metrics and Bank Outcomes
- Baseline scenario outcomes:
  - Average CET1 ratio increases modestly from slightly under 20 percent at June 2016 to 21 percent at June 2019 on a system-wide basis.
  - Two banks fall below the regulatory minimum CET1 threshold of 4.5 percent under the baseline scenario.
- Adverse scenario outcomes:
  - Five banks fall below the supervisory threshold of 4.5 percent CET1.
  - System-wide aggregate CET1 ratio falls to around 15 percent by June 2019.
  - System-wide leverage ratio falls from 11 percent to 8 percent, remaining above the Basel III reference level of 3 percent.
  - Three banks fall below the regulatory minimum CET1 ratio of 4.5 percent by June 2018.
  - Aggregate capital shortfall of 1.8 percent of the annualized nominal GDP by end of simulation; of the five banks below the regulatory minimum at the end, three become technically insolvent.
- Comparison with BNB ST:
  - BNB baseline: only one bank falls below regulatory minimum; average CET1 increases to 22 percent at year-end 2018.
  - BNB adverse: three banks fall below regulatory minimum; average CET1 falls to 14.4 percent at the end of the three-year simulation.
  - Methodological differences: IMF excludes accrued interest on NPLs in both scenarios and uses bank-specific funding shocks (zero for foreign-owned banks); BNB allows interest accrual on NPLs (if accounting permits) and applies uniform funding shocks.

### Concentration Risk and Single-name Exposures
- Average size of the largest exposure: 24 percent of total capital; considering only NFC exposures, average largest exposure is just under 16 percent of total capital.
- Base case (full recovery of collateral):
  - Default of largest NFC exposure of each bank implies average total CAR of more than 19 percent.
  - Two banks fall below the 8 percent total CAR hurdle; total regulatory capital shortfall equivalent to 0.2 percent of GDP.
  - Default of five largest NFC exposures increases banks below 8 percent total CAR to five; total capital shortfall increases to 1.5 percent of GDP.
- Severe case (50 percent haircut on collateral):
  - Default of five largest NFC exposures increases banks below 8 percent total CAR to seven; total capital shortfall equivalent to 1.7 percent of GDP.
- Exposure-to-parent analysis:
  - Default of largest exposure of each bank implies average total CAR of 17 percent; two banks fall below 8 percent total CAR; total shortfall equivalent to 0.2 percent of GDP.
  - Severe scenario leads to four banks falling below 8 percent total CAR; total shortfall slightly more than 0.6 percent of GDP.
- Caveat: concentration testing may underreport risks without highest-level consolidation of client relationships and related-party exposures.

### Liquidity Stress Tests — Framework and Key Results
- Four liquidity exercises:
  - LCR-type liquidity stress test.
  - NSFR-type liquidity stress test.
  - Cash flow-based liquidity stress test using maturity buckets.
  - Reverse liquidity stress test.
- Funding and liquidity context:
  - Nonbank deposits: 83 percent of liabilities.
  - Wholesale funding (including repo and long-term debt): 3.4 percent of total funding.
  - System loan-to-deposit ratios: LTD domestic currency 66 percent; LTD loans in Euro close to 100 percent; overall system LTD 76 percent.
- LCR/NSFR results (examples):
  - LCR baseline (BCBS-recommended minimum 10 percent run-off for less stable deposits): No bank underperformed; System LCR was 232 percent.
  - LCR extreme (50 percent deposit outflow): System LCR remained above 100 percent; only one bank underperformed; total liquidity shortfall less than BGN 740 million or 0.85 percent of 2015 GDP.
  - NSFR baseline (zero percent deposit outflows): Three banks underperform threshold; initial shortfall BGN 600 million, or 0.7 percent of GDP; System NSFR 146.2 percent.
  - NSFR extreme (50 percent less stable deposit outflow): Aggregate NSFR drops to 125.2; shortfall for the three underperforming banks increases to BGN 1 billion, or 1.2 percent of GDP.
- Cash-flow liquidity stress test (selected run-off and roll-over assumptions and short-term results):
  - Time horizons: 7 days, 30 days, 90 days, 1 year.
  - 7-day horizon: three banks vulnerable (including a large one); combined shortfall: 0.7 percent of GDP; simulated cash outflow BGN 16.7 billion.
  - 30-day horizon: two banks vulnerable; simulated cash outflow BGN 18.4 billion.
  - 90-day horizon: one bank vulnerable; total cash shortfall 0.5 percent of GDP; simulated cash outflow BGN 19.8 billion.
  - 1-year horizon: no banks vulnerable; all banks have adequate cash inflows.
  - System-wide shortage of liquidity under severe structural shocks is relatively small, less than 1 percent of GDP.
- Results sensitive to assumptions on availability/quality of securities pledged as collateral and timing of deposit withdrawals.

### Contagion, Interconnectedness, and Network Tests
- Domestic interbank contagion model (Espinosa-Vega and Solé (2010) style) using net bilateral exposures among 17 largest banks as of end-June 2016.
- Findings:
  - Interbank market is shallow; limited direct balance-sheet contagion.
  - Net domestic interbank positions are small relative to banks’ capitalization.
  - No single failure of a domestic bank would trigger failure of another bank through the net-exposure channel; no cascade effect detected.
  - Four banks present higher vulnerability indices; two banks have higher contagion indices.
- Caveats:
  - Psychological/ depositor-confidence channels (e.g., KTB episode in 2014) are not captured by balance-sheet contagion model and could transmit stress via deposit withdrawals.

### Policy Implications and Key Recommendations
- Analyze and stress test largest exposures separately, also taking into account potential cash flows from collateral in case of default.
  - Authority: BNB
  - Timeframe: I (Immediately, within one year)
- Introduce cash flow based liquidity reporting template; in the absence of common European Banking Authority (EBA) template use examples from other EU countries, like Austria or BIS recommended one.
  - Authority: BNB
  - Timeframe: I
- Promote high quality IFRS 9 implementation practices. See Technical Note on NPL Reduction Strategy.
  - Authority: BNB
  - Timeframe: I

### Key System Statistics and Structural Features (as reported)
- Financial system assets: 115 percent of GDP.
- Banking sector accounts for close to 87 percent of the financial system’s assets.
- Insurance companies: close to 4 percent of financial system assets.
- Pension funds: close to 10 percent of financial system assets.
- Banking system composition and assets (Assets in 000 LEV) — Total: 88,650,159 (selected banks listed in source).
- Asset quality (EBA definition of NPLs):
  - Ratio of gross NPLs to total loans and advances: 13.7 percent.
  - Loan loss reserve to NPL coverage: 56.8 percent.
- Asset composition:
  - Loans: close to 58 percent of total assets.
  - Balances with the central bank: 12 percent.
  - Investments in medium- and long-term government bonds: 11 percent.
- Profitability drivers:
  - Trading revenues: around 10.7 percent of total revenues.
  - Fees and commissions revenue: 20 percent of total revenue.
- Capital:
  - Tier 1 capital ratio as of June 2016: 21.1 percent.
- Funding and deposit insurance:
  - Eligible deposits (up to EUR 100,000): about 93.7 percent of total deposits.
  - Bulgarian Deposit Insurance Fund recapitalized via a one-off Ministry of Finance EUR 2 billion loan and World Bank and EBRD loans of EUR 300 million each.

*Source: cr17200 - EXECUTIVE SUMMARY and selected chapters (IMF staff, FSAP stress test results and analysis).*

### EXECUTIVE SUMMARY ___________________________________________________________________________ 5

### EXECUTIVE SUMMARY

### Overview
- FSAP stress tests assess Bulgarian bank resilience to solvency, concentration, liquidity, and balance sheet contagion risks.
- Coverage: 17 banks, making up 95 percent of banking system assets.
- Simulation horizon: three-year projections (June 2016—June 2019), based on data available through June 2016 and using end-June 2016 bank-by-bank balance sheets, income statements, and prudential reports.
- Both baseline and adverse scenarios incorporate AQR adjustments based on year-end 2015 data.
- Key methodological differences highlighted: Top-down (IMF) vs Bottom-up (BNB) approaches, including treatment of interest accrual on NPLs, provisioning rules, profitability of performing loan portfolios, market risk treatment, and funding cost shocks.

### Solvency Stress Tests — Key Findings
- Baseline scenario (modest growth, declining unemployment, stable/low interest rates):
  - Two banks—including a systemic one—exhibit weakness in capital buffers to cope with accumulated losses.
  - IMF ST approach excludes interest income from NPLs in both baseline and adverse scenarios; AQR-driven increase in NPLs reduces the performing loan base that generates cash-based interest income.
  - Result: Two banks do not generate enough recurring income to cover interest expense and credit costs in the baseline, producing negative profits and declining capital levels.
- Adverse scenario (large drop in output; increase in unemployment; decline in collateral prices; increase in risk premiums):
  - Five banks fall below the minimum regulatory CET1 ratio threshold of 4.5 percent.
  - Three banks fall below the threshold by the second year of the stress period.
  - Two additional banks fall below the regulatory minimum in the third year.
  - Aggregate capital shortfall of 1.8 percent of the annualized nominal GDP at the end of the simulation period.
  - Of the five banks below the regulatory minimum at the end, three become technically insolvent.
- Main driver of capital deterioration:
  - Loan loss provisions are the largest driver of the change in capital positions in the adverse scenario.
  - As NPLs rise, loan loss provisions are increased by 20 percentage points from the assumed minimum provisioning level of 60 percent in the baseline scenario, producing large provisioning charges that reduce capital.
  - NPLs increase significantly in the second year as economic growth turns negative; recovery starts in the third year but loss provisions continue to weigh on profitability.

### Solvency Stress Tests — Design and Sensitivities
- Scenario-based solvency ST complemented by sensitivity tests:
  - Interest rate shocks: instantaneous upward and parallel shift in yield curve and increase in real interest rates by 500 bps.
  - Decline in sovereign security prices.
  - Significant increase in provisioning rates for NPLs.
  - Concentration credit risks (default of largest borrowers).
- Satellite models and methodologies by IMF staff used to forecast credit losses and profitability under macro scenarios.
- IMF-imposed assumptions differ from BNB’s, notably IMF assumes NPLs do not accrue income in both baseline and adverse scenarios; BNB applies that assumption only in the adverse scenario.

### Liquidity Stress Tests — Key Findings
- Four separate liquidity exercises:
  - LCR-type liquidity stress test.
  - NSFR-type liquidity stress test.
  - Cash flow-based liquidity stress test using maturity buckets.
  - Reverse liquidity stress test.
- All banks meet initial minimum total LCR ratios of 70 percent, with wide margins.
- Cash flow-based liquidity stress tests (time horizons of 7, 30, and 90 days, and one year) incorporate haircuts on sovereign securities and lower inflows from loan portfolios linked to AQRs and solvency ST scenarios; assumption: new nonperforming assets will not generate cash flows.
- Implied cash flow-based liquidity ST: banks have enough inflows from liquid and less liquid assets to withstand severe structural shocks up to a one-year horizon.
- Under simulated system-wide liquidity shocks and run-off rates:
  - Three banks fail to maintain positive cash flow over a seven-day horizon.
  - Two banks fail to maintain positive cash flow over one month.
  - One bank fails to maintain positive cash flow over three months.
  - Over the one-year horizon, all banks have positive flows.
  - System-wide shortage of liquidity is relatively small, less than 1 percent of GDP.
- Results are sensitive to assumptions on availability and quality of securities pledged as collateral and timing of deposit withdrawals.

### Contagion and Interconnectedness
- Contagion and spillover risks are minimal from balance-sheet exposures:
  - Local banks funded mostly by domestic deposits; little dependence on significant foreign funding.
  - Foreign-owned banks’ reliance on parents for funding is minimal; they are predominantly funded by domestic deposits.
  - Interbank market is shallow; limited room for direct (balance-sheet) contagion.
  - However, a loss of confidence could transmit via deposit withdrawals (psychological channel), as experienced in the 2014 Corporate Commercial Bank (KTB) episode.
- Balance-sheet contagion model simulations:
  - Failure of one or several banks tested to assess sources/destinations of spillovers.
- Concentration risks:
  - Relatively high, but concentrated in a subset of banks.
  - Several banks have large exposures that represent meaningful portions of their capital base.
  - Concentration stress test: two banks would fail in the case of a default of their largest nonfinancial corporate (NFC) exposures.

### Policy Implications and Key Recommendations
- Key recommendations (Authority responsible; Timeframe: I = Immediately, within one year):
  - Analyze and stress test largest exposures separately, also taking into account potential cash flows from collateral in case of default.
    - Authority: BNB
    - Timeframe: I
  - Introduce cash flow based liquidity reporting template; in the absence of common European Banking Authority (EBA) template use examples from other EU countries, like Austria or BIS recommended one.
    - Authority: BNB
    - Timeframe: I
  - Promote high quality IFRS 9 implementation practices. See Technical Note on NPL Reduction Strategy.
    - Authority: BNB
    - Timeframe: I

### Purpose and Use of FSAP Stress Tests
- Objective: assess capacity of the banking system to withstand extreme but plausible macroeconomic shocks and to explore weaknesses and transmission channels of adverse shocks.
- Tests aim to identify policy priorities: reducing specific exposures, building capital and liquidity buffers.
- FSAP stress testing also identifies informational and methodological gaps and assesses authorities’ preparedness for financial distress.

*Source: cr17200 - EXECUTIVE SUMMARY (IMF staff, FSAP stress test results and analysis).*

### 10.      Although stress tests are useful for exploring weaknesses in a financial system, results

### 10.      Although stress tests are useful for exploring weaknesses in a financial system, results must be interpreted with caution.

### Stress tests: limitations and implementation
- Stress tests are conceptually challenging and results must be interpreted with caution.
- Stress tests use macroeconomic and satellite models to calculate the impact of adverse scenarios or shocks on banks.
- These models are estimated using historical data and are subject to estimation uncertainty.
- Limitations can be mitigated, but not eliminated, by using state-of-the-art techniques.
- Design choices include the severity of shocks and imposing constraints, floors, and caps on certain variables.
- In adverse scenarios, the economy is typically affected by a combination of external and domestic shocks that (ex ante) have a very low probability of occurrence.
- By construction, adverse scenarios should not be interpreted as macroeconomic “forecasts.”
- Satellite models map the variables projected in the macroeconomic scenarios into credit factors that determine individual banks’ gains or losses.

### Structure of the Bulgarian financial and banking system
- Financial system assets: 115 percent of GDP.
- Banking sector accounts for close to 87 percent of the financial system’s assets.
- Insurance companies account for close to 4 percent of financial system assets.
- Pension funds account for close to 10 percent of financial system assets.
- Banking system composition:
  - 22 banks and 5 foreign branches.
  - Total banking assets: approximately 100 percent of GDP.
  - Nine local banks’ total asset share: 23.3 percent.
  - Foreign banks account for 74 percent of banks’ assets.
  - Foreign branches account for 2.6 percent of banks’ assets.
  - Top five banks hold 58 percent of total bank assets; only one of the top five is domestically owned.
- Currency composition of assets:
  - Approximately half of total system assets are denominated in foreign currency (mostly Euro).
  - Just under 50 percent of assets are denominated in Bulgarian lev.
- Asset composition:
  - Loans: close to 58 percent of total assets.
  - Balances with the central bank: 12 percent.
  - Investments in medium- and long-term government bonds: 11 percent.
- Funding composition:
  - Nonbank deposits: 83 percent of liabilities.
  - Very little reliance on wholesale funding markets.

### Coverage of the stress test exercise
- ST exercise based on the 17 largest banks in Bulgaria covering 95 percent of the banking system assets.
- Bank asset list (Assets in 000 LEV) — Total: 88,650,159
  - Unicredit Bulbank: 17,351,145
  - Dsk Bank: 11,492,993
  - First Investment Bank: 8,533,291
  - United Bulgarian Bank: 7,020,804
  - Eurobank Bulgaria: 6,701,780
  - Raiffeisenbank (Bulgaria): 6,258,158
  - Societe Generale Expressbank: 5,449,107
  - Central Cooperative Bank: 4,695,231
  - Piraeus Bank Bulgaria: 2,886,726
  - Cibank: 2,781,561
  - Allianz Bank Bulgaria: 2,538,985
  - Investbank: 1,985,189
  - Bulgarian Development Bank: 1,595,301
  - Procredit Bank (Bulgaria): 1,539,578
  - Municipal Bank: 1,350,433
  - International Asset Bank: 1,173,876
  - Bulgarian-American Credit Bank: 1,026,024
  - Citibank Europe-Bulgaria Branch: 865,897
  - Bnp Paribas S.A.-Sofia Branch: 777,657
  - D Commerce Bank: 692,598
  - Ing Bank N.V.-Sofia Branch: 594,452
  - Tbi Bank: 506,547
  - Tokuda Bank: 398,964
  - Teximbank: 186,183
  - Commercial Bank Victoria: 137,256
  - T.C. Ziraat Bank-Sofia Branch: 102,490
  - Isbank Ag-Sofia Branch: 7,933

### Asset quality
- EBA definition of NPLs:
  - Ratio of gross NPLs to total loans and advances: 13.7 percent.
  - Loan loss reserve to NPL coverage: 56.8 percent.
- Credit risk concentration by sector (percent of total loans; NPL ratios indicated where provided):
  - Trade: 29.5 percent of total loans; NPL ratio 21.9 percent.
  - Industry: 27.7 percent of total loans; NPL ratio 16.6 percent.
  - Construction: NPL ratio 47.1 percent.
  - Real estate: NPL ratio 32.6 percent.
  - Construction and real estate combined: 25.1 percent of the portfolio (exposed to housing market vulnerabilities).
- Currency denomination:
  - About 50 percent of loans denominated in foreign currency (predominantly EUR).
  - Related currency risks limited by the country’s commitment to the currency board.
- Data notes:
  - NPL data until end-2014 are from the Bulgarian Credit Register and use the local definition of NPLs.
  - Starting in 2015, Bulgaria adopted the EBA definition of NPLs and source for 2015-onward is the Finrep Supervisory dataset.

### Profitability
- Reported profitability is high but sustainability is uncertain given sluggish loan portfolio growth.
- Drivers of reported profitability:
  - Substantially reduced provisioning expenses.
  - Lower interest and operational expenses.
  - Gains on portfolio of government securities in the declining interest rate environment after the crisis.
- Trading revenues: around 10.7 percent of total revenues.
- Fees and commissions revenue: 20 percent of total revenue.
- Regulatory development: new BNB ordinance regulating fees is being phased in and is likely to limit future profit growth.
- Given current low level of rates, future gains on government securities are unlikely.

### Funding and liquidity
- Funding
  - Deposit funding abundant in domestic currency.
  - System loan-to-deposit ratio (LTD): 66 percent (domestic currency).
  - LTD ratio for loans in Euro: close to 100 percent.
  - Overall system LTD ratio: 76 percent.
  - Wholesale funding (including repo and long-term debt): 3.4 percent of total funding.
- Deposits and deposit insurance
  - Eligible deposits (up to EUR 100,000): about 93.7 percent of total deposits (include foreign currency deposits).
  - Bulgarian Deposit Insurance Fund ran into operational difficulties in the 2014 event when all payments to customers of the affected bank were suspended for about six months.
  - Fund recapitalized through a one-off Ministry of Finance EUR 2 billion loan and further bolstered by World Bank (WB) and European Bank for Reconstruction and Development (EBRD) loans of EUR 300 million each.
- Liquid assets
  - Consist mostly of cash and balances with the BNB, with small amounts invested in liquid securities.
  - Little scope for other secured funding such as repo.

### Capital
- Capital improvements:
  - Quantity and quality of capital has improved to account for higher risk profile.
  - In 2009 the BNB issued recommendations to retain profits to support capital.
  - Banks are not allowed to pay dividends, or need express BNB approval to do so; most profits are retained.
  - Decline in risk weights; RWAs roughly constant since the GFC while assets increased slightly.
  - Tier 1 capital ratio as of June 2016: 21.1 percent.
- Regulatory capital composition and distributions presented in figures (data source: BNB supervisory data; IMF staff estimates).

### Key risks and vulnerabilities
- Main vulnerabilities: concentrated exposures of banks, high levels of NPLs, high corporate sector indebtedness.
- Potential consequences: significant credit losses for banks if key risk events materialize.
- Key risks identified in the Risk Assessment Matrix (RAM):
  - Protracted slowdown of European growth leading to slower growth and higher unemployment in Bulgaria, lower income for corporate sector and households, adverse effects on bank profitability and asset quality.
  - Change of investor sentiment toward emerging markets and resulting financial stress, leading to higher country risk premiums, wider sovereign spreads, and funding pressures for banks.
  - Increasing geopolitical tensions related to countries in the region, likely to lead to higher country risk premiums.
  - Asset price uncertainty over NPLs; declines in collateral values would require higher provisions, lowering banks’ profitability and capital levels.

### Solvency stress tests — Scenarios
- Time horizon: three-year horizon.
- Two scenarios: baseline and adverse.
- Baseline scenario:
  - Reflects IMF World Economic Outlook (WEO) projections as of October 2016.
  - Year-end WEO projections for 2016, 2017, and 2018 used as baseline for stress tests.
  - As latest balance sheet data were as of June 2016, the three-year WEO projections were used for June 2017, June 2018, and June 2019.
- Main macrofinancial variables used in baseline scenario: annual GDP growth rate, unemployment rate, short-term interest rate (Sofibor).
- Baseline projections (Actual June 2016; Baseline June 2017; June 2018; June 2019) — (In percent)
  - Annual GDP Growth: 3.6 3.0 2.8 2.5
  - Unemployment Rate: 8.9 8.2 7.1 6.9
  - SOFIBOR: 0.13 0.13 0.13 0.13
- Additional baseline assumptions:
  - Sofibor stays constant at its June 2016 level, 13 basis points, in the following three years.
  - These variables are combined with satellite models to project banks’ NPL ratios under the baseline scenario.
  - A minimum provisioning ratio of 60 percent is assumed for NPLs under this scenario.

*Source: IMF staff report (chapter text provided).*

### 28.      The adverse scenario envisions a significant contraction in GDP and increase in the

### The adverse scenario envisions a significant contraction in GDP and increase in the unemployment rate

### Calibration and scenario design
- Calibration based on the Flexible System of Global Models developed by the IMF Research Department, benchmarked with the BNB’s BU stress testing scenarios and expert judgment.
- Adverse scenario simulates annual real GDP growth rates of:
  - -3 percent in the first year (June 2017)
  - -3.6 percent in the second year (June 2018)
  - -0.9 percent in the third year (June 2019)
- These numbers represent a cumulative GDP growth deviation of close to 16 percentage points over the three-year period from the baseline scenario, corresponding to approximately 1.5 times the standard deviation of the three-year cumulative growth rates observed during the 1996–2015 period.
- Unemployment rate path under the adverse scenario:
  - 10.4 percent by June 2017
  - 11.5 percent as of June 2018
  - 12 percent as of June 2019
- Sofibor shocks relative to the baseline scenario:
  - 33 basis points for June 2017
  - 23 basis points for June 2018
  - 6 basis points for June 2019
- Adverse scenario uses average haircuts on sovereign bonds from the 2016 EBA stress test to simulate declining values of banks’ available-for-sale (AFS) sovereign securities holdings.

### Macrofinancial variable table (as presented)
- Actual / Adverse Scenario Projections (In percent)
  - June 2016 — June 2017 — June 2018 — June 2019
  - Annual GDP Growth: 3.6-3.0-3.6-0.9
  - Unemployment Rate: 8.9 10.4 11.5 12.0
  - SOFIBOR: 0.13 0.46 0.36 0.19

### External and funding shocks
- Adverse scenario driven by external shocks: protracted slower growth in key export markets, leading to lower external demand, lower growth, higher unemployment, and lower corporate income.
- Provisioning ratio uniformly increased by 20 percentage points under the adverse scenario to capture declining collateral values in an extreme adverse environment.
- Bank-specific funding shocks for domestic banks calibrated using each bank’s liquidity ratio (total deposits to liquid assets) and the largest historical shock to deposit interest rates during 2007–2016.
  - Average funding shock for domestic banks from this calibration: 80 basis points (first year).
  - Average funding shock trajectory for domestic banks:
    - 80 basis points in the first year
    - 160 basis points in the second year
    - 240 basis points in the third year
  - System-wide average funding shock including all banks:
    - 24 basis points in the first year
    - 48 basis points in the second year
    - 72 basis points in the third year
- BNB system-wide average funding shocks used in its stress tests:
  - 97 basis points in the first year
  - 106 basis points in the second year
  - 75 basis points in the third year
- Methodological difference: BNB shocks are lower but applied uniformly to all banks; IMF applies funding shocks only to domestic banks based on liquidity ratios and assumes zero-funding shock for foreign-owned banks.

### Comparison with BNB adverse scenario
- Similar structures in the first two years; differ in the third year.
  - IMF adverse scenario: third-year GDP contraction at a rate of 0.9 percent.
  - BNB adverse scenario: third-year GDP growth of 1.8 percent (economy starts recovering).
- Cumulative deviation from baseline by end of second year:
  - IMF adverse: close to 12 percentage points
  - BNB adverse: close to 10 percentage points
- Both use identical Sofibor shocks relative to baseline: 33, 23, and 6 basis points in years 1–3.

### Satellite models, AQR adjustments, and provisioning assumptions
- Credit risk satellite models developed for three broad exposure classes: corporate loans; mortgages; and other consumer loans; exposures to institutions excluded.
- NPL time series used for satellite models starts in Q4 2009 and covers 2011–2016.
- CRR standard risk weights used for migration of loans under the Basel II STA portfolio.
- Expected losses (ELs):
  - STA portfolios (except sovereigns): flows of NPLs multiplied by bank-level historical average provisioning rates.
  - IRB portfolios: ELs = hybrid PDs × LGD × EAD.
  - Sovereign exposures: RWAs not stressed; average haircuts from EBA 2016 ST applied; expected losses calculated for own sovereign exposures.
- AQR frontloading:
  - 2016 Q2 stock of performing loans, capital levels, RWAs, and provisions adjusted to reflect bank-level AQR findings provided by the BNB.
  - Interest income from loans adjusted proportionally to the percent adjustment applied to the stock of loans.
- Provisioning modeling:
  - Baseline provisioning rates based on 2011–2016 bank-level historical averages, floored at 60 percent by judgment.
  - Adverse scenario provisioning rates uniformly increased by 20 percentage points from baseline levels and capped at 100 percent.
  - In adverse scenario, an additional 20 percentage points of provisions added for loans re-classified as NPLs in the AQR results; these provisions equal 0.2*X million for loans reclassified totaling BGN X million, divided equally over the three years.

### Credit risk outcomes and losses
- Top-down ST projects significant increases in NPLs under the severe stress scenario versus declining NPLs in the baseline.
- Under the severe stress scenario, the combined effect of higher unemployment and slowdown leads to an increase of approximately 8 percentage points in the banking system’s average NPLs between 2016 Q2 and 2019 Q2.
- Corporate loans and mortgage loans experience the highest increases in NPLs compared to starting points.
- Credit losses over the three-year horizon in the loan book amount to BGN 4.4 billion in the adverse scenario (equivalent to approximately 5 percent of total banking system assets as of June 2016).

### Key simulation assumptions for balance sheets and profits
- Growth of banks’ balance sheets projected to grow in line with nominal GDP, with a zero-growth floor (assets remain constant relative to the economy).
- Non-interest profit items projected to grow in line with balance sheet size; extraordinary income assumed not to recur; accrued interest from nonperforming exposures excluded in both scenarios.
- Dividend distribution:
  - Baseline: banks satisfying capital requirements assumed to distribute 50 percent of after-tax profits.
  - Adverse scenario: banks not allowed to distribute dividends.

### Stress test results: baseline and adverse
- Baseline scenario:
  - Average CET1 ratio increases modestly from slightly under 20 percent at June 2016 to 21 percent at June 2019 on a system-wide basis.
  - Two banks fall below the regulatory minimum CET1 threshold of 4.5 percent under the baseline scenario due to substantial NPL increases from AQR adjustments; these banks generate insufficient recurring income and record negative profits and declining capital.
- Adverse scenario:
  - Five banks fall below the supervisory threshold, including the two that were below the threshold in the baseline scenario.
  - System-wide aggregate CET1 ratio falls to around 15 percent by June 2019.
  - System-wide leverage ratio falls from 11 percent to 8 percent, remaining above the Basel III reference level of 3 percent.
  - Three banks fall below the regulatory minimum CET1 ratio of 4.5 percent by June 2018.
  - By the end of the simulation period, two additional banks fall below the regulatory minimum, with an aggregate capital shortfall of 1.8 percent of the annualized nominal GDP.
  - Of the five banks falling below the regulatory minimum at the end of the simulation period, three become technically insolvent.

*Source: IMF staff estimates and BNB.*

### 45.      The largest driver of the change in capital positions in the adverse scenario is loan loss

### 45.      The largest driver of the change in capital positions in the adverse scenario is loan loss

### Drivers of change in capital positions
- The largest driver of the change in capital positions in the adverse scenario is loan loss provisions.
- NPLs increase significantly as the economic outlook deteriorates in the simulated adverse scenario.
- An assumed increase in the provisioning ratio of 20 percentage points combined with rising NPLs leads to large loan loss provisions, which reduce banks’ capital.
- Estimated satellite models indicate changes in macroeconomic variables (GDP growth, unemployment, and interest rates) lead to increases in NPLs with a lag of several quarters.
- Lagged positive growth up to June 2016 results in only modestly higher NPLs by June 2017; as growth turns negative from the first year of the scenario period, NPLs increase significantly in the second year.
- As the economy starts recovering in the third year, the flow of NPLs slows but loss provisions continue to exert a significant drag on profitability.
- For domestic banks, interest rate-related losses stemming from sizeable funding shocks are the second largest driver of capital decline.

### Stress test results and system capital metrics
- Under the BNB ST baseline scenario:
  - Only one bank falls below the regulatory minimum threshold.
  - Average CET1 ratio for the 17 banks included in the IMF ST exercise increases to 22 percent at year-end 2018.
- Under the adverse scenario of the BNB ST:
  - Three banks fall below the regulatory minimum threshold, including the bank that fell below in the baseline.
  - Average CET1 ratio for the 17 banks falls to 14.4 percent at the end of the three-year simulation period.
  - Two of the three banks not meeting minimum capital requirements become technically insolvent at the end of the simulation period.
- Differences between IMF ST and BNB ST:
  - IMF ST does not allow for interest income accrual on NPLs; BNB ST allows it if in line with international accounting standards.
  - IMF ST uses bank-specific funding shocks and assumes zero shock for foreign-owned banks; BNB ST applies uniform funding shocks to all banks.
  - Under the adverse scenario, IMF ST shows larger variation across banks: some banks receive sizeable funding shocks and large losses, others do not, widening the distribution of final capital levels.

### Sensitivity analysis — stand-alone shocks (static, as of June 2016 after AQR adjustments; RWAs assumed constant)
- Shocks assessed: decline in prices of sovereign securities; increase in interest rates affecting net interest income; increase in provisions for NPLs due to decline in collateral values.
- Decline in prices of sovereign securities:
  - Valuation haircuts based on EBA (2016) ST adverse scenario applied to HFT and AFS holdings.
  - Banks would suffer minimal losses given modest sovereign holdings.
  - No bank falls below the regulatory minimum threshold.
  - Aggregate CET1 ratio declines by 0.2 percentage points.
- Interest rate shock (maturity gap analysis; instantaneous upward shift of 500 basis points in the yield curve):
  - Impact is highly heterogeneous across banks due to different maturity gap profiles.
  - Only one bank falls below the regulatory minimum threshold.
  - Aggregate CET1 ratio declines by 1.3 percentage points.
  - Aggregate CET1 remains relatively high compared to close to 20 percent in June 2016.
- Increased provisions for existing stock of NPLs (uniform increase of provisions by 20 percentage points):
  - Two banks fall below the regulatory minimum threshold for capital.
  - Aggregate CET1 ratio declines by approximately 6 percentage points.
  - Due to large stock of NPLs, this shock leads to much larger declines in capital levels compared to sovereign-price and interest-rate shocks.
- Note: sensitivity tests are static and capture instantaneous impacts at June 2016 balance sheet positions.

### Concentration risk (single name exposures)
- Method:
  - Assessed default of up to five largest NFC borrowers per bank using supervisory large exposure data (exposures > 10 percent of regulatory capital considered; sovereign exposures excluded).
  - Two recovery scenarios: full recovery of collateral (base) and 50 percent haircut on collateral plus zero recovery on unsecured part (severe).
  - Also ran sensitivity including exposures to financial institutions (parents).
- Key metrics:
  - Average size of the largest exposure is 24 percent of total capital.
  - Considering only NFC exposures, average largest exposure declines to just under 16 percent of total capital.
- Base case (zero recovery on unsecured part, full recovery of collateral):
  - Default of largest NFC exposure of each bank implies average total CAR of more than 19 percent (well above 8 percent total CAR hurdle).
  - Two banks fall below the 8 percent total CAR hurdle rate.
  - Total regulatory capital shortfall equivalent to 0.2 percent of GDP.
  - Default of five largest NFC exposures increases banks below 8 percent total CAR to five; total capital shortfall increases to 1.5 percent of GDP.
- Severe case (zero recovery on unsecured part, 50 percent haircut on collateral):
  - Default of largest NFC exposure leads to two banks falling below 8 percent total CAR; total shortfall slightly over 0.2 percent of GDP.
  - Default of five largest NFC exposures increases banks below 8 percent total CAR to seven; total capital shortfall equivalent to 1.7 percent of GDP.
  - Haircut of 50 percent on collateral in stress is conservative; for financial guarantees it implies losses on both obligor and guarantor.
- Exposure-to-parent analysis (including financial institution exposures):
  - Default of largest exposure of each bank implies average total CAR of 17 percent.
  - Two banks fall below 8 percent total CAR; total shortfall equivalent to 0.2 percent of GDP.
  - Severe scenario (zero recovery unsecured, 50 percent haircut on collateral) leads to four banks falling below 8 percent total CAR; total shortfall slightly more than 0.6 percent of GDP.
  - Potential losses from parent exposures correspond to modest amounts at the system level.
- Caveats:
  - Concentration risk testing ideally requires highest level of consolidation of exposures and clients; absence of such information can underreport risks for connected clients/related parties.
  - Formal large exposure definition floored at 10 percent of capital (CRR) may miss de facto connected exposures.

### Liquidity stress tests — scenarios and models
- Framework:
  - Short-term and long-term liquidity stress tests approximating Basel III LCR and NSFR using bank-by-bank supervisory data for 17 domestically incorporated solo banks and foreign subsidiaries.
  - Tests assessed vulnerability to increasing outflows of less stable deposits.
  - About 80 percent of deposits have maturities of less than 90 days.
  - 43 percent of retail deposits are classified as less stable (EBA reporting templates).
- Run-off rate assumptions:
  - LCR benchmark: 5 percent for stable deposits and simulation of 10 percent to 50 percent for less stable deposits.
  - NSFR simulation: 10 percent to 50 percent for less stable deposits.
  - Simulations covered range from 0 percent to 50 percent run-off of less stable deposits.
- A cash flow-based liquidity stress test using central bank liquidity reporting templates was also performed.

### Liquidity stress tests — results
- LCR baseline scenario (BCBS-recommended minimum 10 percent run-off for less stable deposits):
  - No bank underperformed the standard.
  - System LCR was 232 percent.
- LCR extreme scenario (50 percent deposit outflow):
  - System LCR remained above 100 percent.
  - Only one bank underperformed the standard.
  - Total liquidity shortfall was less than BGN 740 million or 0.85 percent of 2015 GDP.
- NSFR baseline (zero percent deposit outflows):
  - Three banks underperform the threshold in baseline.
  - Initial shortfall for these three banks amounts to BGN 600 million, or 0.7 percent of GDP.
  - System-wide NSFR is 146.2 percent.
- NSFR extreme scenario (50 percent less stable deposit outflow):
  - No new banks underperform the standard (the same three remain below threshold).
  - Aggregate NSFR level drops to 125.2.
  - Shortfall for the three underperforming banks increases to BGN 1 billion, or 1.2 percent of GDP.
- Additional LCR/NSFR summary figures:
  - System LCR remained above 100 percent even in extreme 50 percent outflow scenario.
  - System NSFR is 146.2 and remains above 100 percent throughout the scenario simulation range.
  - Liquidity shortfall grows from BGN 600 million to BGN 1 billion (from 0.74 to 1.24 percent of GDP) across scenarios.

*Source: IMF staff estimates.*

### 67.      A cash flow-based liquidity stress test was performed using bank-level data on the

### 67. A cash flow-based liquidity stress test was performed using bank-level data on the temporal structure of cash flows generated by different liabilities and assets

### Cash-flow liquidity stress test: setup and scenarios
- The cash-flow stress test mapped temporal structures of cash flows from liabilities and assets to scenarios of run-off rates (liability withdrawals) and roll-over rates (asset renewals) for time horizons of 7 days, 30 days, 90 days, and 1 year (see Table 3 for scenario assumptions).
- Model parameters stated for the contagion/fire-sale component used elsewhere in the analysis: 50 percent haircut in the fire sale of assets and a 65 percent roll-over ratio of interbank debt.
- Selected run-off rate assumptions (Run-off Rates — 7 Days / 30 Days / 90 Days / 1 Year):
  - Withdrawal of deposits by non-residents (financial institutions) — Group entities: 0 / 0 / 0 / 0
  - Withdrawal of deposits by non-residents (financial institutions) — Non-Group entities: 25 / 25 / 50 / 100
  - Withdrawal of deposits by non-residents (non-financial institutions) — Group entities: 0 / 0 / 0 / 0
  - Withdrawal of deposits by non-residents (non-financial institutions) — Non-Group entities: 25 / 25 / 50 / 100
  - Inability to issue bonds which mature within one year: 50 / 50 / 25 / 0
  - Withdrawal of deposits from residents — Credit institutions: 100 / 100 / 100 / 100
  - Withdrawal of deposits from residents — Non-financial companies and households: 40 / 30 / 20 / 10
  - Flow of new loans: 3.5 / 3.5 / 3.5 / 3.5
- Selected roll-over and inflow rate assumptions (Roll-over Rates — 7 Days / 30 Days / 90 Days / 1 Year):
  - Cash and CB reserves: 100 / 100 / 100 / 100
  - Balances with other banks: 50 / 50 / 70 / 100
  - Tradeable debt securities: 80 / 85 / 90 / 95
  - Gold and other commodities: 90 / 75 / 70 / 65
  - Loans and receivables — Central governments: 100 / 100 / 100 / 100
  - Loans and receivables — Corporate: 90 / 90 / 90 / 90
  - Loans and receivables — Retail: 90 / 90 / 90 / 90

### Cash-flow stress test: key results and statistics
- Overall result: Results were similar to those from the LCR- and NSF-based analyses. Most banks had adequate buffers to meet simulated cash outflows; a few banks experienced cash shortfalls.
- Very short term (7-day horizon):
  - Number of banks vulnerable: three banks (including a large one).
  - Combined shortfall: 0.7 percent of GDP.
  - Simulated cash outflow for the banking system: BGN 16.7 billion.
- 30-day horizon:
  - Number of banks vulnerable: two banks.
  - Simulated cash outflow for the banking system: BGN 18.4 billion.
- 90-day horizon:
  - Number of banks vulnerable: one bank.
  - Total cash shortfall: 0.5 percent of GDP.
  - Simulated cash outflow for the banking system: BGN 19.8 billion.
- 1-year horizon:
  - Number of banks vulnerable: none; all banks have adequate cash inflows to meet cash outflows in the simulated scenario.
- Aggregate cash shortfalls across horizons correspond to a small portion of GDP.

### Contagion and interconnectedness tests — domestic interbank contagion
- Sample and data:
  - Network model based on Espinosa-Vega and Solé (2010).
  - Analysis uses matrix of bilateral domestic interbank net credit exposures among the seventeen largest banks in the sample, with information as of end-June 2016.
  - Interbank exposures in the dataset are on the asset side of banks’ balance sheets (net exposures).
- Model mechanics:
  - Hypothetical default of each bank is assumed one at a time; default occurs on all interbank obligations of the bank.
  - The test assesses direct credit and capital losses in other banks and subsequent rounds (cascade effects) if triggered.
  - Funding shocks cause banks to replace lost funding by selling assets at discounts (fire sales), causing further capital losses.
- Findings on interbank activity and interconnectedness:
  - Net exposures show limited interbank market activity and a limited number of connections among banks.
  - Only a few banks are systemically important, and their net exposure remains limited.
  - Higher interconnectedness exists among foreign-owned banks compared to domestic ones.
- Contagion test outcomes:
  - Contagion risks from domestic net interbank exposures are very limited.
  - Domestic net interbank positions are small, especially relative to banks’ capitalization.
  - For banks in the system, the sum of their net domestic exposures to other banks is smaller than their regulatory capital.
  - No single failure of a domestic bank would trigger the failure of another bank through the net-exposure channel; no cascade effect would take place in this four-bank market.
  - As of end-June 2016, none of the banks is found to be undercapitalized with regard to the Tier 1 capital regulatory minimum after a shock on one or several of its domestic interbank exposures.
  - Four banks present a significantly higher level of vulnerability of spillovers to others (vulnerability index — percentage of loss at a single institution due to the default of all other institutions).
  - Two banks have a higher index of contagion (average percentage of loss of other banks due to the failure of a given bank).
- Limitations and caveats:
  - Banks that fail stress tests (IMF) have very limited impact on the rest of the system via net exposures; banks failing both baseline and adverse scenarios would not trigger cascading effects for the rest of the banks in the net-exposure model.
  - The assessment does not include potential “psychological” contagion effects that markets might impose across banks perceived to be similar to failed institutions.

### Selected liquidity framework assumptions (appendices)
- LCR outflow and inflow assumptions include (examples shown exactly as in source):
  - Retail Deposits — Stable deposits: 5
  - Less stable retail deposits: 10
  - Term deposits, residual maturity > 30d: 0
  - Operational deposits generated by clearing, custody, and cash management activities: 25
  - Portion covered by deposit insurance: 5
  - Nonfinancial corporates, sovereigns, central banks, multilat development banks, PSEs — Fully covered by deposit insurance: 20; Not fully covered by deposit insurance: 40
  - Secured funding with a central bank, or backed by Level 1 assets: 0
  - Level 1 Assets inflow (1-haircut): 100
  - Level 2a assets inflow (1-haircut): 85
  - Level 2b — Qualifying Mortgage Backed Securities: 75; Qualifying corporate debt securities rated between A+ and BBB-: 50; Qualifying common equity shares: 50
- NSFR available stable funding factors and required stable funding factors are provided in tabular form covering maturity buckets "1-3 Months", "3-6 Months", "6-9 Months", "9-12 Months", "12 Months and Over" with specific factors per item (examples shown exactly as in source):
  - Own funds: 100 / 100 / 100 / 100 / 100
  - Retail deposits (Article 411(2) qualifying for Article 421(1)): 95 / 95 / 95 / 95 / 95
  - Liabilities from customers who are not financial customers — unsecured lending transactions: 50 / 50 / 50 / 50 / 100
  - Required stable funding for Transferable assets representing claims on or guaranteed by sovereigns: 50 / 50 / 50 / 50 / 50

_Source: IMF Staff analysis as presented in the provided content._

### 2. Financial stress in

### cr17200 - 2. Financial stress in

### Major identified risks and transmission (emerging markets)
- Medium: Adverse effects on banks and corporates due to increased investor risk aversion which leads to higher risk premiums.
- Medium: Increased country risk premium, leading to wider spreads for the sovereign, funding pressures for banks, higher borrowing costs for nonfinancial corporates, and lower FDI.  
  - In the stress tests, this risk will be modeled through wider spreads for the sovereign (losses on domestic mark-to-market securities) and higher funding costs for banks.
- High/Medium: Intensification of geo-political tensions related to Russia, Ukraine, and Turkey.
  - Negative spillovers from trade and investment channels. Increase in country risk, sovereign rating downgrade.
  - Adverse effect on economic activity and social tensions leading to the higher country risk premiums and potential sovereign rating downgrade by one to three notches.
  - This risk will be modeled through a standalone sensitivity shock (sovereign downgrade).
- High: Asset price uncertainty over NPLs.
  - Delayed recognition of NPLs, and loan foreclosures. Impediments to releasing collateral associated with NPLs, including judicial bottlenecks and administrative costs.
  - Medium: Collateral prices decline. The need for higher provisions lowers bank profits and capital; investment and growth prospects are in turn weakened.
  - In the stress tests, this risk will be modeled through higher provisioning rates for banks.

### Appendix V — Stress Testing Matrix (STeM) for the Banking Sector: Top-down Stress Test by FSAP Team — Solvency Risk (assumptions)
- Institutional perimeter
  - Institutions included: 17 banks.
  - Market share: 95 percent of the banking system assets.
  - Data and baseline date: Source: Supervisory and publicly-available data. Baseline date: End-June 2016.
  - Scope of Consolidation: Consolidated level data for banks which have their headquarters in Bulgaria and subsidiary level data for the subsidiaries of foreign banks.
- Channels of risk propagation / Methodology
  - Satellite models developed by the FSAP team.
  - Balance sheet-based approach.
  - Satellite models for macro-financial linkages.
  - Models for credit losses, pre-impairment income, credit growth, expert judgment on coverage ratios.
  - Models to integrate solvency-funding interactions.
  - Methodology to calculate sovereign risk.
  - Methodology to calculate losses from bonds and money market instruments (sovereign and other issuers).
  - Net fee income and commission income projected based on nominal GDP growth and expert judgment.
  - No accrued income on NPL loans.
  - Stress test horizon: 3 years (2016 Q2–2019 Q2).
- Tail shocks / Scenario analysis
  - Scenario-based tests assess impacts on the entire portfolio, including loans and, if applicable, the trading book.
  - Variables include domestic macro-financial variables (e.g., GDP and inflation), and GDP for key trading partners (EU, Russia, neighboring non-EU countries).
  - Bulgaria-specific severe stress scenario: the GDP growth rate declines to -6 percent, -6.4 percent, and -3.4 percent, in 2017, 2018, and 2019, respectively.
  - External sector shocks, including large declines in exports and FDI, calibrated to magnitudes similar to those observed in countries with currency board system during 2008–2009.
- Sensitivity analysis
  - Evaluate external shocks: sovereign rating downgrade and a decline in the prices of sovereign bonds.
  - Default of large corporate borrowers.
  - Decline in real estate prices.
  - Increase in interest rates (risk premiums).
- Risks and buffers
  - Risks/factors assessed: Credit losses; Losses from bonds and money market instruments (sovereign and other issuers) in the banking and trading books; Funding costs.
  - Behavioral adjustments: Balance sheet grows in line with nominal GDP. Dividends are paid out by banks that remain adequately capitalized throughout the stress period. Dividend payout ratio is determined by using historical data.
- Regulatory and market-based standards and parameters
  - Calibration of risk parameters: Through-the-cycle and point-in-time for credit risk parameters or proxies.
  - Regulatory/accounting and market-based standards: European and national regulation. Basel II/III STA approach.
- Reporting format for results
  - Output presentation: System-wide capital shortfall. Number of banks and percentage of banking assets in the system that fall below microprudential hurdle rates (Basel minimum: 4.5 percent and 8 percent for common equity tier 1 and total capital ratios, and 3 percent leverage ratio).

### Appendix V — Stress Testing Matrix: Liquidity Risk (assumptions)
- Institutional perimeter
  - Institutions included: 17 largest banks in the system.
  - Market share: 95 percent of the banking system assets.
  - Data and baseline date: Source: Supervisory data. Baseline date: End-June 2016.
  - Scope of Consolidation: Consolidated level data for banks that have their headquarters in Bulgaria, and subsidiary level data for the subsidiaries of foreign banks.
- Channels of risk propagation / Methodology
  - Basel III-LCR and NSFR type proxies.
  - Cash-flow based liquidity stress test using maturity buckets by banks.
- Risks and buffers
  - Risks: Funding liquidity (liquidity outflows); Market liquidity (price shocks and haircuts).
  - Buffers: Counterbalancing capacity (HQLA); Central bank facilities.
- Tail shocks / Size of the shock
  - Runoff rates calculated based on historical events and LCR/NSFR rates.
  - Bank run and dry up of wholesale funding markets, taking into account haircuts to liquid assets.
- Regulatory and market-based standards and parameters
  - Regulatory standards: European Commission Delegated Regulation (EU) 2015/61.
- Reporting format for results
  - Output presentation: Bank-level and aggregate banking-level liquidity gaps.
  - Format for Results: Survival period in days by bank, number of banks that can still meet their obligations.

### Appendix V — Stress Testing Matrix: Interconnectedness
- Institutional perimeter
  - Institutions included: 17 largest banks in the system.
- Modeling approach / Methodology
  - Simple balance sheet contagion model.

### Appendix VI — Specification of Satellite Models of Credit Risk
- Modeling approach and purpose
  - NPL ratios of individual banks under the stress test scenarios (baseline and adverse) were estimated using satellite models of credit risks.
  - Losses related to credit risk were computed based on the changes in provisions resulting from the loan migration as predicted by the changing levels of NPLs under each scenario.
  - Panel data models were developed to predict bank-specific NPL ratios for three exposure classes in bank’s portfolios: corporate loans, mortgage loans, and other retail loans.
  - Models used macrofinancial variables included in the scenarios (GDP growth, Sofibor, unemployment rate), and added the lagged NPL ratios when necessary.
  - Exact specification for each exposure class was chosen based on Bayesian Model Averaging approach.
- Dependent variable transformation
  - The dependent variable in the panel regression for each asset class was the logit transformation of the NPL ratio:
    - )1()1/ln( ititit NPLNPLY
  - General specification estimated:
    - )2(,,1,,1 ,1,, SsandTtforXYY tisttiiti   
- Estimation results (selected coefficients and statistics)
  - Corporate Loans
    - Dependent Variable(t-1): 0.7293*** (12.87)
    - GDP(t-2): -2.1783*** (-3.92)
    - Sofibor(t-2): 0.0353*** (-4.93)
    - R- Squared: 0.97
    - Observations: 190
  - Mortgage Loans
    - GDP(t-5): -0.7033*** (-2.79)
    - Unemployment(t-2): 0.0841 (14.31)
    - Unemployment(t-2): 0.1250*** (10.39) [note: unemployment appears in specification table]
    - R- Squared: 0.99
    - Observations: 286
  - Other Retail Loans
    - GDP(t-5): -1.7204*** (-3.71)
    - Sofibor(t-3): 0.0956*** (14.64)
    - R- Squared: 0.96
    - Observations: 198
  - Source note: t-Statistics in Parentheses. * Denotes significance at 10 percent, ** at 5 percent, and *** at 1 percent.

*International Monetary Fund — Appendix extracts from the Bulgaria FSAP stress testing chapter*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2017/cr17200.pdf_