## _cr15213 - EXECUTIVE SUMMARY

## Source details

**Canonical URL:** [_cr15213 - EXECUTIVE SUMMARY](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr15213.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr15213.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr15213.pdf.json)

---

### Introduction
- Objective: assess the resilience of the banking system in Bosnia and Herzegovina (BiH) to major sources of risk as part of the 2015 Financial Sector Assessment Program (FSAP) Update.
- Stress test coordination and data:
  - Conducted in collaboration with the Central Bank of Bosnia and Herzegovina (CBBH) and the banking agencies of both entities.
  - Top-down solvency stress tests used bank-by-bank supervisory data as of March 31, 2014.
  - Bottom-up stress tests were conducted by all 27 banks using their internal models with FSAP-provided macroeconomic scenarios.
  - Liquidity stress tests, contagion risk analysis, and complementary sensitivity analysis covered all banks.

### Banking system and stress test coverage
- System structure and scale:
  - 27 licensed banks: 17 in the Federation of Bosnia and Herzegovina (FBiH) and 10 in Republika Srpska (RS).
  - Balance-sheet summary (In millions of KM, end-March 2014):
    - Number of banks: 27 17 10
    - Total assets: 22,619  15,238   7,381
    - Total loans: 15,955  10,956   4,999
    - Total deposits: 15,933  11,281   4,652
    - Total regulatory capital: 2,939    2,101    838
    - Total risk-weighted assets (RWAs): 16,943  12,003   4,940
    - Capital adequacy ratio (in percent of RWAs)17.317.517.0
  - Roughly 90 percent of assets are subsidiaries of foreign banks.
  - Gross loans account for over 70 percent of total assets in FBiH and over 65 percent in RS.
  - Trading book and holdings of securities remain limited (less than 5 percent of total assets).
  - Roughly two thirds of bank assets and liabilities are denominated in foreign currency, mainly euro.
- Stress test scope:
  - Top-down solvency tests for entire system (27 banks) and bottom-up tests by each bank.
  - Liquidity stress tests for all banks.
  - Contagion and spillover analysis included inter-linkages with the insurance sector, development banks, and other financial institutions.

### Macroeconomic scenarios
- Three macroeconomic scenarios over 2015-2019:
  - Baseline: projections generated by the IMF area department team.
  - Adverse Scenario 1: moderate external shock (euro area weakening and geopolitical deterioration in Ukraine).
  - Adverse Scenario 2: severe scenario adding severe reduction in external funding for banks and loss of confidence.
- The severe adverse scenario assigned a very small qualitative probability but included to assess tail-risk stability.
- Projections covered real GDP, CPI inflation, short and long-term interest rates, exchange rates, unemployment, real estate prices, equity index, and credit growth over 2015-2019.

### Solvency and key risk findings
- System-wide assessment:
  - System-wide solvency and liquidity indicators appear broadly appropriate, but significant pockets of vulnerability remain.
  - Aggregate stress losses, mainly related to increased provisions in the loan book, are non-negligible but broadly manageable at system level.
- Credit risk (principal vulnerability):
  - Loans represent more than two thirds of total banking sector assets.
  - System-wide NPLs stood at 15 percent of total loans at end-March 2014; ratio higher for private corporate sector loans.
  - Top-down credit risk losses on loan book:
    - KM 545mn in “Adverse Scenario 1” (equivalent to 2 percent of GDP)
    - KM 2,019mn in “Adverse Scenario 2” (equivalent to 7 percent of GDP)
  - Top-down and bottom-up results broadly align except for credit risk losses: aggregate increases in NPLs from top-down are noticeably larger than bottom-up in the most severe scenario.
  - About two thirds of loans are FX-indexed; information on borrowers’ currency of income/assets (“unhedged borrowers”) is not collected and inherent risks from unhedged borrowers could be substantial.
  - Estimation coefficients for NPL logit model:
    - real GDP growth: -0.15
    - short-term interest rate: 0.05
- Concentration risk:
  - For five banks, the combined amount of the two largest exposures exceeds the total regulatory capital of the bank.
  - Single largest exposure ≈ 33 percent of total regulatory capital on average; ≈ 29 percent (FBiH) and 41 percent (RS).
  - Sensitivity: default by largest exposure of each bank → implied capital shortfall ≈ KM 91mn (0.3 percent of GDP); simultaneous default of five largest exposures (zero recovery) → seven banks insolvent; total capital shortfall ≈ KM 1,094mn (3.8 percent of GDP).
- Market and exchange rate risks:
  - Direct exchange rate and other market risks appear contained:
    - Net open FX positions small and within regulatory caps; 93 percent of banks’ net open FX position relates to EUR.
    - Sensitivity: hypothetical 30 percent depreciation vs EUR → system-wide CAR would increase from 17.3 percent to 17.6 percent.
  - Indirect FX risks from unhedged borrowers could not be quantified due to lack of data.
- Operational risk:
  - Operational risk losses modeled as function of macro conditions; elasticity of logit(op. losses as percent of assets) to real GDP growth = -0.46.
  - In top-down stress test, operational risk losses represent close to 12 percent of total net losses in the adverse scenarios.
- Asset Quality Reviews (AQRs):
  - Several banks underwent AQRs; March 31, 2014 balance sheet data did not reflect AQRs.
  - Total capital deductions from AQRs for FBiH ≈ KM 22.2mn; effect on FBiH CAR: 17.5 percent → 17.3 percent; implied capital shortfall KM 14.3mn (0.05 percent of GDP).
  - AQR amounts small relative to stress-test estimated shortfalls (≈ KM 1,148mn in Adverse Scenario 2).

### Liquidity findings
- System-level liquidity:
  - Aggregate LCR for the entire system: around 292 percent.
  - LCR by entity: FBiH 349 percent; RS 205 percent.
  - System-wide NSFR: around 123 percent (FBiH 127 percent; RS 114 percent).
  - Bank deposits at central bank represent around 76 percent of banks’ HQLA, of which roughly 40 percent are “required reserves.”
  - If required reserves excluded from HQLA, system-wide LCR would fall to around 200 percent.
- Heterogeneity and vulnerabilities:
  - A few domestically-owned banks present relatively low liquidity ratios.
  - Banks with LCR below 100 percent: five banks (two FBiH, three RS).
  - Banks with NSFR below 100 percent: seven banks (four FBiH, three RS).
  - System implied liquidity shortfalls:
    - LCR: around BAM 70mn (0.2 percent of GDP)
    - NSFR: close to BAM 323mn (1.1 percent of GDP)
- Interbank liquidity and overnight transactions:
  - Domestic interbank exposures less than 1 percent of total regulatory capital; hypothetical default of any bank would not create significant cascade effects.
  - Some overnight bilateral transactions may not be fully captured in monthly supervisory reporting.

### Contagion risks and spillovers
- Cross-border exposures:
  - Foreign claims as of March 31, 2014: KM 2,146mn (about 10 percent of banks’ total assets).
  - Direct exposures to Austria and Germany represent roughly half of all foreign claims.
  - Network analysis (Espinosa-Vega and Solé (2010)-based) simulated credit and funding shocks:
    - Key assumptions: credit shock LGD = 100 percent on interbank exposures; funding shock haircut = 50 percent on fire sales; roll-over ratio = 80 percent; loss on BiH banks’ exposures to foreign public sectors = 20 percent.
    - Default of the Austrian banking system would cause five BiH banks’ regulatory capital to fall below the 12 percent minimum requirement, equivalent to a loss of 20 percent of BiH’s total regulatory capital.
    - Joint default of Germany and Austria in the credit shock would impair nearly half of total regulatory capital.
    - Adding funding shocks increases number of undercapitalized banks, especially when shocks come from top five systematically important countries for BiH.
- Domestic contagion:
  - Banks with significant vulnerabilities (mainly small domestically-owned banks) are somewhat interconnected among themselves but have low direct linkages with rest of BiH banks; direct spillovers likely limited.
  - Indirect contagion risks (e.g., reputational) not quantified but could pose significant systemic risk.
- Spillovers to insurance sector:
  - Banks’ liabilities to insurance sector ≈ KM 550mn (less than 3 percent of total banking sector deposits).
  - In Adverse Scenario 2: five banks become insolvent; joint default implies capital impairment of around KM 43mn (14 percent of insurance sector’s capital); at company level, two insurance companies would lose over half of total capital; one would lose its entire capital.

### Main recommendations based on the stress tests
- Conduct additional Asset Quality Reviews (AQR) in banks with weak solvency and liquidity indicators.
  - Responsibility: Banking agencies
  - Timeline: Short-term
- Improve modeling of credit risk losses, particularly by computing provisions based on "expected losses" and risk parameters (PDs, LGDs, EAD).
  - Responsibility: CBBH
  - Timeline: Medium-term
- Collect information on "unhedged borrowers."
  - Responsibility: Banking agencies
  - Timeline: Medium-term
- Fine-tune parameters embedded in the LCR and NSFR to better reflect domestic system specificity; in particular regarding inclusion of "required reserves" in the HQLA.
  - Responsibility: Banking agencies
  - Timeline: Short-term
- Regular compilation of an LCR in euro.
  - Responsibility: Banking agencies
  - Timeline: Short-term
- Continue monitoring systemic risks through periodic (at least once a year) bottom-up stress tests.
  - Responsibility: Banking agencies and CBBH
  - Timeline: Ongoing
- Monitor overnight interbank transactions not captured in the banks' monthly supervisory report.
  - Responsibility: Banking agencies
  - Timeline: Short-term

Notes: “short-term” denotes within the next 12 months; “medium term” denotes 12 to 24 months.

### Solvency stress test quantitative results (system-wide, 27 banks; data as of 31-March-2014)
- Pre-shock (actual data):
  - Total regulatory capital - before shock: 2,938,843 (thousands of KM)
  - Total risk-weighted assets (RWAs): 16,942,976
  - Total assets: 21,783,153
  - CAR before shock: 17.3 percent
- Stress-test estimated capitalization after shock (2015):
  - Baseline (Top-Down): Total regulatory capital - after shock: 3,157,302; CAR after shock: 18.6 percent; Implied capital shortfall: -
  - Baseline (Bottom-Up): Total regulatory capital - after shock: 3,216,311; CAR after shock: 19.0 percent; Implied capital shortfall: 144 (thousands of KM)
  - Adverse Scenario 1 (Top-Down): Total regulatory capital - after shock: 2,753,549; CAR after shock: 16.3 percent; Implied capital shortfall: 26,147 (thousands of KM)
  - Adverse Scenario 1 (Bottom-Up): Total regulatory capital - after shock: 2,916,570; CAR after shock: 17.2 percent; Implied capital shortfall: 2,941 (thousands of KM)
  - Adverse Scenario 2 (Top-Down): Total regulatory capital - after shock: 951,673; CAR after shock: 5.6 percent; Implied capital shortfall: 1,147,991 (thousands of KM)
  - Adverse Scenario 2 (Bottom-Up): Total regulatory capital - after shock: 2,374,494; CAR after shock: 14.0 percent; Implied capital shortfall: 122,731 (thousands of KM)
- Number of banks with CAR below 12 percent:
  - Baseline: Top-Down -, Bottom-Up 1
  - Adverse Scenario 1: Top-Down 5, Bottom-Up 2
  - Adverse Scenario 2: Top-Down 22, Bottom-Up 13
- Number of banks with CAR below "zero":
  - Adverse Scenario 2 (Top-Down): 5

### Selected stress losses components (system-wide, 2015, Top-Down / Bottom-Up where shown)
- Credit risk: Increase in provisions due to loan migration:
  - Baseline TD: (4,067)
  - Baseline BU: (55,062)
  - Adverse 1 TD: (545,431)
  - Adverse 1 BU: (326,040)
  - Adverse 2 TD: (2,019,179)
  - Adverse 2 BU: (829,554)
- Expected net losses on BIH government bond holding (HTM):
  - Baseline TD: (1,267)
  - Baseline BU: (230)
  - Adverse 1 TD: (2,025)
  - Adverse 1 BU: (537)
  - Adverse 2 TD: (3,127)
  - Adverse 2 BU: (1,071)
- Expected operational risk losses (examples):
  - Baseline TD: (1,810)
  - Baseline BU: (9,221)
  - Adverse 1 TD: (22,343)
  - Adverse 1 BU: (10,454)
  - Adverse 2 TD: (238,144)
  - Adverse 2 BU: (11,651)
- Total net expected "stress losses":
  - Baseline TD: 730,724
  - Baseline BU: 668,328
  - Adverse 1 TD: 122,551
  - Adverse 1 BU: 365,411
  - Adverse 2 TD: (1,684,089)
  - Adverse 2 BU: (179,600)

### Non-performing loan (NPL) outcomes under Adverse Scenario 2 (Implied NPL ratios, percent of total loans) — Top-Down medians (selected)
- FBiH medians:
  - 2014: 12.2
  - 2015: 31.1
  - 2016–2019: median ranges 31.1 to 33.7 (detailed table in source)
- RS medians:
  - 2014: 13.7
  - 2015: 34.8
  - 2016–2019: median ranges 34.8 to 37.6 (detailed table in source)
- Bottom-Up series show lower medians in later years for bottom-up compared with top-down in some years.

### Liquidity metrics (system-level)
- LCR (including "required reserves"):
  - System-wide LCR (in percent): 291.7 (BiH, 27 banks); 348.6 (FBiH, 17 banks); 205.2 (RS, 10 banks)
  - Implied liquidity shortfall (in millions of BAM): 70.2 (BiH); 49.7 (FBiH); 20.4 (RS)
  - Number of banks with LCR below 100 percent: 5 (BiH); 2 (FBiH); 3 (RS)
- LCR (excluding "required reserves"):
  - System-wide LCR (in percent): 200.5 (BiH); 241.5 (FBiH); 138.1 (RS)
  - Implied liquidity shortfall (in millions of BAM): 324.6 (BiH); 142.7 (FBiH); 181.9 (RS)
  - Number of banks with LCR below 100 percent: 11 (BiH); 6 (FBiH); 5 (RS)
- NSFR:
  - System-wide NSFR (in percent): 123.0 (BiH); 127.4 (FBiH); 114.0 (RS)
  - Implied liquidity shortfall (in millions of BAM): 322.8 (BiH); 209.9 (FBiH); 112.9 (RS)
  - Number of banks with NSFR below 100 percent: 7 (BiH); 4 (FBiH); 3 (RS)

### Risk assessment highlights (Appendix I)
- Main risks (1–3 year horizon) and staff assessments:
  - Risk 1 — Bond market stress from euro-area sovereign risk: Staff assessment Low; Likelihood Medium; Impact Medium.
  - Risk 2 — Geopolitical tensions surrounding Russia/Ukraine: Staff assessment Medium; Likelihood Low; Impact Low.
  - Risk 3 — External funding: parent banks remove support: Staff assessment Medium; Likelihood Medium; Impact Medium.
  - Risk 4 — Further deterioration in commercial banks and depositor confidence: Staff assessment Medium; Likelihood High; Impact High.
  - Risk 5 — Unavailability of official budget financing and sovereign debt restructuring: Staff assessment Low; Likelihood Low; Impact Low.

*International Monetary Fund staff assessment based on supervisory data and IMF staff calculations as presented in the source content.*

### EXECUTIVE SUMMARY ___________________________________________________________________________ 5

### _cr15213 - EXECUTIVE SUMMARY

### Introduction
- Objective: assess the resilience of the banking system in Bosnia and Herzegovina (BiH) to major sources of risk as part of the 2015 Financial Sector Assessment Program (FSAP) Update.
- Stress test coordination and data:
  - Conducted in collaboration with the Central Bank of Bosnia and Herzegovina (CBBH) and the banking agencies of both entities.
  - Top-down solvency stress tests used bank-by-bank supervisory data as of March 31, 2014.
  - Bottom-up stress tests were conducted by all 27 banks using their internal models with FSAP-provided macroeconomic scenarios.
  - Liquidity stress tests and contagion risk analysis, plus complementary sensitivity analysis, covered all banks.

### Banking system and stress test coverage
- System structure and scale:
  - 27 licensed banks operating in BiH: 17 in the Federation of Bosnia and Herzegovina (FBiH) and 10 in Republika Srpska (RS).
  - As of end-March 2014, total banking sector assets amounted to KM 21.8bn (prose summary).
  - Balance-sheet summary (In millions of KM, end-March 2014) as reported:
    - Number of banks: 27 17 10
    - Total assets: 22,619  15,238   7,381
    - Total loans: 15,955  10,956   4,999
    - Total deposits: 15,933  11,281   4,652
    - Total regulatory capital: 2,939    2,101    838
    - Total risk-weighted assets (RWAs): 16,943  12,003   4,940
    - Capital adequacy ratio (in percent of RWAs)17.317.517.0
  - Banking system composition:
    - Roughly 90 percent of assets are subsidiaries of foreign banks.
    - Gross loans account for over 70 percent of total assets in FBiH and over 65 percent in RS.
    - Corporate loans represent about half of total loans in FBiH and close to 63 percent in RS.
    - Trading book and holdings of securities remain limited (less than 5 percent of total assets).
    - Domestic interbank exposures are fairly small; some banks have noticeable foreign funding.
    - Roughly two thirds of bank assets and liabilities are denominated in foreign currency, mainly euro.
- Stress test scope:
  - Top-down solvency stress tests for entire system (27 banks) and bottom-up tests by each bank.
  - Liquidity stress tests for all banks.
  - Contagion and spillover analysis included inter-linkages with the insurance sector, development banks, and other financial institutions.

### Macroeconomic scenarios
- Three macroeconomic scenarios were used:
  - Baseline scenario based on the latest Area Department staff projections.
  - Two alternative scenarios: one moderate and one severe.
  - Full-fledged five-year macroeconomic projections were quantified for each scenario.

### Solvency and key risk findings
- System-wide view:
  - System-wide solvency and liquidity indicators appear broadly appropriate, but significant pockets of vulnerability remain.
  - Aggregate stress losses, mainly related to increased provisions in the loan book, are non-negligible but broadly manageable at system level.
- Credit risk (principal vulnerability):
  - Potential credit risk losses on the loan book represent the most important risk factor.
  - Top-down tests show changes in non-performing loans (and loan category migrations) are sensitive to macroeconomic conditions.
  - Top-down and bottom-up results broadly align except for credit risk losses: aggregate increases in non-performing loans from top-down are noticeably larger than bottom-up in the most severe scenario.
  - Some banks’ bottom-up models show little reaction of non-performing loans to macroeconomic stress, biasing aggregate bottom-up results downward.
- Concentration risk:
  - Concentration risks are high in specific segments despite moderate average large exposure share at system level.
  - For five banks, the combined amount of the two largest exposures exceeds the total regulatory capital of the bank.
  - Sensitivity analysis: system-level potential losses remain broadly manageable, but a few banks could become insolvent if a small number of their largest exposures default.
- Market and exchange rate risks:
  - Direct exchange rate risk and other market risks appear contained:
    - All banks comply with regulatory limits and have fairly small net open FX positions.
    - Net open FX positions do not necessarily move in the same direction across currencies, providing a natural hedge.
    - Large share of loans are de facto floating rate; repricing gap between assets and liabilities is quite small.
    - Limited securities holdings reduce other market risk channels.
  - Indirect exchange rate risks and “unhedged borrowers”:
    - Most banks do not collect information on currency denomination of borrowers’ income or assets.
    - Due to lack of data and no historical exchange rate shocks against the euro, inherent risks from unhedged borrowers could not be quantified but are likely substantial given widespread FX-linked loans.
- Operational risk:
  - Operational risk losses are part of the solvency assessment and found to be sensitive to macroeconomic developments (figure referenced).
- Asset Quality Reviews (AQRs):
  - Several banks underwent comprehensive AQRs, but March 31, 2014 balance sheet data did not reflect AQR results.
  - Box 2 presents analysis showing the impact from AQRs on stress test results is small at system level.

### Liquidity findings
- System-level liquidity:
  - Liquidity stress tests based on Basel III LCR and NSFR-type proxies show the system has ample liquidity: system-wide LCR exceeds 250 percent.
  - Most banks hold sizeable deposits at the central bank.
- Heterogeneity and vulnerabilities:
  - A few banks, mainly domestically-owned, present relatively low liquidity ratios.
  - Liquidity stress tests and distribution of liquidity ratios show most banks are sound but pockets of low liquidity exist.
- Interbank liquidity and overnight transactions:
  - Domestic interbank exposures reported are limited (less than 1 percent of total regulatory capital), implying hypothetical default of any bank would not create significant cascade effects.
  - Some domestic-owned banks engage in overnight transactions of deposits, FX, and cash based on bilateral agreements; these may not be fully captured in monthly supervisory reporting.

### Contagion risks and spillovers
- Cross-border exposures:
  - Foreign claims represent about 10 percent of total banking sector assets.
  - Direct exposures to Austrian and German banks represent roughly half of all foreign claims.
  - Three types of cross-border exposures identified:
    - Credit lines from parent companies to foreign-owned banks.
    - Large amounts of deposits abroad for FX transactions and settlement.
    - “Round-trip” cross-border exposures exhibited by a few (mainly domestically-owned) banks.
  - Network analysis suggests direct and indirect effects from potential credit and funding shocks abroad could be sizeable for the domestic banking system.
- Domestic contagion:
  - Banks with significant vulnerabilities (mainly small domestically-owned banks) are somewhat interconnected among themselves but have low direct linkages with other BiH banks; direct spillovers to rest of system likely limited.
  - Indirect contagion risks (e.g., reputational risk) were not quantified in the stress tests but could pose significant systemic risk.
- Spillovers to insurance sector:
  - Contagion risks from banks to the insurance sector appear contained and would operate mainly through insurance companies’ deposits in banks.

### Main recommendations based on the stress tests
- Conduct additional Asset Quality Reviews (AQR) in banks with weak solvency and liquidity indicators.
  - Responsibility: Banking agencies
  - Timeline: Short-term
- Improve modeling of credit risk losses, particularly through the computation of provisions based on "expected losses" and risk parameters (PDs, LGDs, EAD).
  - Responsibility: CBBH
  - Timeline: Medium-term
- Collect information on "unhedged borrowers."
  - Responsibility: Banking agencies
  - Timeline: Medium-term
- Fine-tune the parameters embedded in the LCR and NSFR to better reflect the specificity of the domestic banking system; in particular regarding the inclusion of "required reserves" in the HQLA.
  - Responsibility: Banking agencies
  - Timeline: Short-term
- Regular compilation of an LCR in euro.
  - Responsibility: Banking agencies
  - Timeline: Short-term
- Continue monitoring of systemic risks through periodic (at least once a year) bottom-up stress tests.
  - Responsibility: Banking agencies and CBBH
  - Timeline: Ongoing
- Monitor overnight interbank transactions not captured in the banks' monthly supervisory report.
  - Responsibility: Banking agencies
  - Timeline: Short-term

Notes: “short-term” denotes within the next 12 months; “medium term” denotes 12 to 24 months.

*EXECUTIVE SUMMARY — Bosnia and Herzegovina, INTERNATIONAL MONETARY FUND*

### 6.      Stress tests were based on the use of full-fledged macroeconomic scenarios. To assess

### 6.      Stress tests were based on the use of full-fledged macroeconomic scenarios. To assess

### Macroeconomic scenarios used in the stress tests
- Three macroeconomic scenarios over the period 2015-2019:
  - Baseline: projections generated by the IMF area department team (continuous monitoring of the economy in BiH).
  - Adverse Scenario 1: a moderate adverse scenario illustrating an external shock driven mainly by a further weakening in the economic outlook of euro area countries, combined with a further deterioration of the current geopolitical crisis in Ukraine.
  - Adverse Scenario 2: the severe adverse scenario in which the external risks in the moderate scenario are accompanied by a severe reduction in external funding for banks (mainly through the removal of support of parent banks to their local subsidiaries), compounded with a further deterioration in the health of commercial banks and loss of confidence.
- The two adverse scenarios relate to risks identified in the Risk Assessment Matrix (RAM):
  - Moderate adverse scenario: Risks #1 and 2 in the RAM.
  - Severe adverse scenario: Risks #1, 2, 3, and 4 in the RAM.
- Probability assessment:
  - The severe adverse scenario is assigned a very small qualitative probability but is included to assess tail-risk stability.

### Construction and coverage of the scenario projections
- Projections covered a large number of macroeconomic and financial variables in a consistent manner over 2015-2019, including:
  - real GDP, CPI inflation, short and long-term interest rates, exchange rates (e.g., against the USD), unemployment rate, real estate prices, equity price index, and credit growth.
- The full numerical representation of the macroeconomic scenarios appears in Table 5 (in the source).

### Behavioral assumptions for the stress tests
- Constant balance sheet assumption: balance sheets reflect initial positions and effects of shocks only; limited managerial actions allowed to ensure comparability across banks.
- Specific behavioral assumptions included:
  - RWAs will remain constant at their pre-shock level.
  - Banks are not allowed to raise fresh capital from existing shareholders or the market.
  - Apart from changes in the ratio of non-performing loans, the composition of the loan book will remain unchanged.
  - Loan write-offs are not allowed.
- No managerial optimization allowed:
  - No optimization of banking or trading book portfolios; balance sheet positions each year must reflect initial portfolios and stress effects.
- Taxes and dividends:
  - Taxes assumed to be paid at a fixed rate of 11.5 percent on positive net profits.
  - Dividend payouts represent a fixed share of after-tax-profits equal to 50 percent, provided the bank's regulatory capital ratio (CAR) exceeds a predetermined threshold.
  - Banks should be allowed to pay dividends only when their CAR exceeds 14.5 percent.
- Other income items:
  - Net fees and commissions, operating expenses (wages and salaries, rental income, sale of fixed assets) modeled as functions of macroeconomic variables when relevant, based on expert judgment and historical trends.
  - Other sources of income or losses (e.g., legal-case losses not related to collateral execution) were excluded.
  - Net interest income and operational risk losses treated separately in dedicated sections.

### Solvency stress test scope
- Risks covered (extensive STeM coverage, Appendix II referenced):
  - Credit risk (loan book and debt instrument holdings / “issuer default risk”)
  - Market risk (interest and exchange rate risks)
  - Risks related to equity instrument holdings
  - Concentration risk
  - Operational risk
  - Contagion risk through interbank exposures and real sector exposures, domestic and foreign

### A. Credit Risk — loan book
- Importance and exposure:
  - Loans represent more than two thirds of total banking sector assets.
  - System-wide NPLs stood at 15 percent of total loans at end-March 2014; ratio higher for private corporate sector loans.
- Modeling approach:
  - Due to limited PD and LGD data, the NPL ratio (percent of total loans) was used to assess credit risk.
  - Losses computed based on increase in provisions implied by loan migration associated with increased NPLs.
  - NPL-ratio modeled via a logit transformation to constrain predictions between 0 and 1:
    - x = ln(NPL/(1 - NPL))  [equation as in the source]
  - The logit transform assumed linear in exogenous macroeconomic factors with individual bank fixed effects; estimation equation provided in the source.
- Estimation details:
  - Equation [2] estimated with annual data over 1999-2013.
  - NPL data disaggregated by 13 economic sectors and by currency (KM, EUR, and “other currencies”).
  - Real GDP growth and the (short-term) interest rate were the main macro drivers of NPLs.
  - Resulting coefficients on the logit transform of the NPL-ratio:
    - real GDP growth: -0.15
    - short-term interest rate: 0.05
- Key quantitative credit-risk findings:
  - Top-down credit risk losses on the loan book:
    - KM 545mn in “Adverse Scenario 1” (equivalent to 2 percent of GDP)
    - KM 2,019mn in “Adverse Scenario 2” (equivalent to 7 percent of GDP)
  - Losses appear evenly distributed across the banking systems in both entities proportional to loan book size.
  - Loss estimates depend heavily on supervisory data quality; actual economic losses could be higher than those implied by supervisory data.
- Unhedged borrowers risk:
  - Approximately two thirds of loans are FX-indexed.
  - Large portion of FX-linked loan borrowers likely do not receive income or possess assets in euros (especially retail borrowers with domestic-currency income).
  - In the event of domestic currency depreciation vs. the euro, exposure of unhedged borrowers could produce sharp deterioration in the loan portfolio.
  - Authorities and most banks do not collect information on “unhedged borrowers.”
- Top-down vs bottom-up:
  - Top-down implied increases in aggregate NPLs are noticeably larger than bottom-up losses estimated by banks.
  - Differences evenly affect loans denominated in domestic and foreign currencies.
  - Some banks’ bottom-up NPLs show little reaction to macro scenarios, which appears less plausible.
- Additional methodological notes:
  - The stress framework models NPLs as loans classified in categories C, D and E together, and assumes the proportion among these categories remains the same before and after shocks.

### A. Credit Risk — debt instrument holdings
- Coverage and methodology:
  - Top-down and bottom-up tests included expected losses on debt instrument holdings in the banking book (HTM): domestic government bonds, corporate bonds, other domestic debt instruments.
  - Methodology for top-down estimation described in Box 1 (in the source).
  - Expected losses due to valuation changes in HFT and AFS portfolios treated under Market Risk.
- Exposure and results:
  - Banking sector exposure to fixed income instruments is limited; government bonds represent nearly all HTM holdings.
  - Expected losses from increased credit spreads are relatively small (Tables 6 and 7 in the source).
  - In FBiH, government bond holdings represent 100 percent of domestic fixed income (HTM) holdings; in RS government bonds represent 97 percent and corporate bonds 3 percent.
  - Losses from debt instrument holdings are dwarfed by loan-book stress losses.

### B. Market Risk
- General approach:
  - Market risks assessed included interest rate risk, exchange rate risk, and valuation changes in securities.
  - Given limited trading-book securities, interest and exchange rate risks were the main market risks considered.
  - Equity investment losses addressed separately.
- Interest rate risk:
  - Assessed via time-to-repricing buckets; assets and liabilities grouped by repricing horizon (less than 3 months; 3–6 months; 6–12 months; 12–24 months; 24–36 months; more than 36 months).
  - Time-to-repricing gap = repricing-sensitive assets minus repricing-sensitive liabilities within a bucket.
  - Expected interest income loss/gain = gap × change in interest rate (direct effects only).
  - Indirect effects via credit risk (asset-quality impact) were handled in the credit risk section.
  - Expected interest income losses from increased NPLs (defaults on interest payments) were added to net interest income on top of direct repricing effects.
- Key quantitative interest-rate findings:
  - Potential losses on interest income are small due to widespread use of floating-rate loans; time-to-repricing gaps are fairly small.
  - System shares with time-to-repricing less than three months:
    - 53 percent of interest rate sensitive assets and 56 percent of interest sensitive liabilities (system-wide figure used in one context).
    - For banks in FBiH: 61 percent of interest rate sensitive assets have time-to-repricing less than three months.
    - For banks in RS: 36 percent of interest rate sensitive assets have time-to-repricing less than three months.
  - Net interest margins are relatively large; banks generate sizeable net interest income profits across stress scenarios (Tables 6 and 7).
  - Sensitivity example:
    - An interest rate increase of 5 percentage points would lead to an increase in the system-wide CAR from 17.3 percent to 17.4 percent.
    - An interest rate increase of 10 percent would lead to an increase in the system-wide CAR from 17.3 percent to 17.6 percent.
  - Timing assumptions for instantaneous losses:
    - Largest net losses on any gap with time-to-repricing less than 12 months considered as instantaneous loss in year 1 (2015).
    - Amount related to “more than 36 months” gap split evenly among years 2018 and 2019.

*Source: IMF staff calculations and the FSAP stress test chapter (as provided in the source content).*

### Box 1. Computation of Potential Losses due to Debt Instruments Risk

### Box 1. Computation of Potential Losses due to Debt Instruments Risk

### Sovereign and corporate debt expected loss computation
- For sovereign bond holdings in the banking book (HTM) the implied expected loss should, in principle, be covered by provisions.
- Risk parameters required: PD, LGD, and EAD.
- If a sovereign rating is available from a main agency (Moody’s, S&P, or Fitch), use that rating to determine the corresponding default probability via mapping tables; this implied default probability is the starting point in the FSAP stress test (call it, ܦܲ
଴
).
- Using panel regression analysis (sample: 117 countries; 2,120 observations; panel fixed effects), an elasticity ߛ was estimated to be -0.088792.
- The estimated relationship (elasticity) used: logit(ܦܲ
௧
) = ߛ · Δ(real GDP growth)݌݀݃ݎ݀
௧
.
- Rearranged formula to compute scenario PDs:
  - ܦܲ
௧
 = [ (ܦܲ
௧ିଵ / (1 - ܦܲ
௧ିଵ)) · exp(݌݀݃ݎ݀∆ · ߛ
௧
) ] / [ 1 + (ܦܲ
௧ିଵ / (1 - ܦܲ
௧ିଵ)) · exp(݌݀݃ݎ݀∆ · ߛ
௧
) ]
- If no historical technical default evidence exists, use recovery rate = 36 percent (World Bank Doing Business Report for Bosnia and Herzegovina) as proxy for LGD in stress tests.
- For corporate bond holdings, apply the same methodology. In absence of corporate ratings, assume corporate sector ratings equal the sovereign’s rating and use the same elasticity ߛ to obtain corporate PDs. LGD again based on the World Bank Doing Business Report.

### Interest rate risk — valuation effects on debt instrument holdings
- Valuation changes on government and corporate bond holdings in AFS and HFT portfolios, and foreign bond holdings (all marked-to-market), were assessed through duration-based valuation.
- Steps:
  - Compute duration of each holding (Macaulay duration depends on maturity date, annual yield, periodic coupon payment and frequency).
  - Calculate portfolio average duration as weighted average of individual durations weighted by amount (in KM) of each bond holding.
  - Expected gains or losses = size of bond portfolio × average duration × change in the relevant interest rate (bond yield).
- Directional effect: an increase in interest rates → valuation loss; a decrease → valuation gain.
- Empirical finding: total holdings of government bonds (HTM, AFS, HFT) represent close to 2 percent of total assets. Under an extreme hypothetical 50 percent loss on these holdings, system-wide CAR would fall from 17.3 percent to 16 percent, with recapitalization needs of around BAM 20mn (less than 0.1 percent of GDP).
- Overall conclusion: potential valuation losses on fixed income instruments remain limited and are relatively small compared to credit risk losses.

### Exchange rate risk (direct effects)
- Direct effects assessed using banks’ net open FX positions by currency categories: USD, EUR, CHF, HRK, RSD, and “other currencies.”
- Implied FX gains or losses = net open position × expected depreciation of corresponding currency under each scenario.
- Note: this section addresses only direct FX effects; indirect effects through credit risk (important due to FX-linked loans) could not be adequately quantified.
- Most banks keep net open FX positions (excluding EUR) close to zero; regulatory caps: net open FX positions capped at 30 percent of capital for EUR, and 20 percent of capital for aggregate of all other currencies.
- “Other currencies” position ≈ 2 percent of banks’ total FX positions.
- Explicit exchange rate paths were provided for USD, EUR, HRK, and RSD; for “other currencies” the NEER path was used.
- Empirical finding: 93 percent of banks’ net open FX position relates to EUR; net open EUR position represents 87 percent and 96 percent of total net open FX position in FBiH and RS, respectively.
- Sensitivity test: hypothetical 30 percent depreciation of local currency against the euro → effect on banking sector capitalization is fairly small; most banks are “long EUR” and would make slight gains.
- Quantitative result: under 30 percent depreciation vs EUR, system-wide capital ratio would increase from 17.3 percent to 17.6 percent.

### Market risk — top-down vs bottom-up
- Market risk losses (valuation, interest rate, FX) are broadly similar in top-down and bottom-up stress tests.
- Net interest income estimates are very similar across methodologies and represent most net gains across scenarios.
- Small differences arise mainly from how individual banks translate macro scenarios into effective risk parameters.

### Equity instrument holdings
- Potential equity losses derive from credit risk (bankruptcies of invested companies) and market risk (price and volatility changes).
- Credit-losses on equity investments computed using same PDs and LGDs as corporate bond HTM holdings (methodology in this Box).
- Market-losses on equities computed as amount held × percent change in equity prices.
- Empirical finding: total equity holdings < 1 percent of total assets. Some banks have no equity investments. Result: potential equity losses are relatively small.

### Operational risk
- Operational risk losses computed as a direct function of macroeconomic scenarios using historical annual data for 27 banks, 2006–2013.
- Operational losses include internal/external fraud; business disruptions and system failures; damage to physical assets; execution, delivery and process management; etc.
- Losses correlate well with macro conditions, in particular real GDP growth.
- Regression estimate: elasticity of logit transformation of operational risk losses (in percent of total assets) to real GDP growth = -0.46.
- In top-down stress test, operational risk losses represent close to 12 percent of total net losses in the adverse scenarios.
- Recommendation: banks should continuously improve internal operating and monitoring systems to mitigate large operational risk losses.

### Concentration risk
- Name concentration tested via hypothetical default of largest N borrowers using supervisory data on largest exposures and collateral.
- Central case recovery rate used = 36 percent; alternative assumptions tested.
- On average, single largest exposure ≈ 33 percent of total regulatory capital for banks in BiH; by entity: ≈ 29 percent (FBiH) and 41 percent (RS).
- Regulatory limit for largest single exposure = 25 percent of bank’s core capital.
- Distribution of largest exposures ranges from 10 percent to 156 percent of capital.
- System-level results:
  - Default by largest exposure of each bank → capital shortfall ≈ KM 91mn (0.3 percent of GDP).
  - Simultaneous default by five largest exposures of each bank with zero recovery → seven banks insolvent; total capital shortfall ≈ KM 1,094mn (3.8 percent of GDP); roughly half of losses from each entity.
- Sector concentration: loan portfolio broadly spread among economic activity sectors.

### Impact of Asset Quality Reviews (AQRs)
- Several banks underwent comprehensive AQRs in first half of 2014 (based on balance sheet data as of December 31, 2013). At FSAP (November 2014) AQR results known for FBiH banks but not for RS banks.
- Top-down and bottom-up stress tests used supervisory data as of March 31, 2014, which did not incorporate AQR results.
- Total capital deductions from these AQRs for banks in FBiH ≈ KM 22.2mn.
- Effect on FBiH: overall capital ratio would fall from 17.5 percent to 17.3 percent; implied capital shortfall of KM 14.3mn (0.05 percent of GDP) at end-March 2014.
- Subsequent recapitalizations reduced undercapitalization: only one bank reported CAR below 12 percent at end-September 2014 with capital shortfall KM 7.5mn (0.03 percent of GDP).
- Comparative scale: AQR amounts dwarfed by stress test estimated shortfalls (≈ KM 1,148mn in the most severe scenario); thus AQRs do not materially affect main stress test conclusions.
- Caveat: future AQRs or larger capital deductions in banks (particularly medium-to-large) could have noticeable consequences for system-wide capitalization.

### Liquidity stress tests (overview)
- Liquidity stress tests based mainly on Basel III LCR and NSFR-type proxies; assumptions for run-off rates and haircuts based on BCBS guidance.
- LCR measures potential net outflows over next 30 days vs counterbalancing capacity (HQLA). Banks should maintain LCR > 100 percent; stressed run-off rates and asset haircuts applied.
- Graphical results reported for LCR and NSFR by BiH, FBiH, and RS (including and excluding required reserves), indicating distribution across banks.

*Source: _cr15213 - Box 1. Computation of Potential Losses due to Debt Instruments Risk*

### 38.      The NSFR measures the aptitude of banks to fund their activities with sufficiently

### _cr15213 - 38.      The NSFR measures the aptitude of banks to fund their activities with sufficiently

### NSFR definition and regulatory benchmark
- The NSFR is defined as the ratio of available stable funding over the amount of required stable funding.
- The amount of required stable funding is computed as the weighted average of a bank’s assets, using predetermined weights defined in BCBS documentation.
- The NSFR ratio should be equal to at least 100 percent on an ongoing basis.

### Data snapshot (reported and AQR-adjusted)
- Data as of Sep 30, 2014 (reported data / "AQR-adjusted" / reported data):
  - Total regulatory capital: 2,101     2,079     2,100
  - Total risk-weighted assets: 12,003     12,003     12,214
  - CAR (in percent of RWAs): 17.5 17.3 17.2
  - No. of banks with a CAR below 12 percent: 1     3     1
  - Implied capital shorfall  1/0.4: 14.3     7.5
- Data as of Mar 31, 2014 (used elsewhere in the analysis).

### System-wide liquidity and coverage metrics
- Aggregate LCR for the entire system in BiH: around 292 percent.
- LCR by entity:
  - FBiH: 349 percent.
  - RS: 205 percent.
- Bank deposits at the central bank represent around 76 percent of the banks’ HQLA, of which roughly 40 percent are “required reserves.”
- If required reserves were excluded from HQLA, system-wide LCR would fall to around 200 percent for the banking sector as a whole.
- System-wide NSFR:
  - Banking sector as a whole: around 123 percent.
  - FBiH: around 127 percent.
  - RS: around 114 percent.
- Overall potential “liquidity shortfall” implied by:
  - LCR: around BAM 70mn (0.2 percent of GDP).
  - NSFR: close to BAM 323mn (1.1 percent of GDP).

### Heterogeneity and vulnerable banks
- System-wide liquidity ratios mask differences across individual banks; smaller mainly domestically-owned banks exhibit much lower liquidity ratios.
- Banks with LCR below 100 percent: five banks (two from FBiH and three from RS).
- Banks with NSFR below 100 percent: seven banks (four from FBiH and three from RS).
- Shallow interbank market exacerbates liquidity risks because banks with excess liquidity are usually not interested in lending to other banks on the interbank market.

### Cross-border contagion risk (network analysis)
- Foreign claims of BiH banks as of March 31, 2014: KM 2,146mn (about 10 percent of banks’ total assets).
- Over 75 percent of these foreign claims are on foreign banking systems, mainly deposits in foreign banks.
- Direct exposures to Austria and Germany represent roughly half of all foreign claims.
- Identified cross-border exposure types:
  - Credit lines from parent companies to foreign-owned banks (decreasing over time).
  - Large deposits in corresponding accounts abroad for FX transaction and settlement purposes.
  - “Round-trip” cross-border exposures where claims and liabilities to a particular counterpart are roughly of the same amount (mainly involving domestically-owned banks; overall amounts very small).
- Network simulation methodology (Espinosa-Vega and Solé (2010)-based) considered:
  - Credit shock: banking system defaulting on liabilities to foreign banks.
  - Funding shock: banking system deleveraging and withdrawing funding, triggering asset fire sales.
  - Key shock assumptions:
    - Credit shock: loss given default = 100 percent on interbank exposures.
    - Funding shock: haircut = 50 percent on asset fire sales; roll-over ratio of interbank debt = 80 percent.
    - Loss on BiH banks’ exposures to foreign public sectors = 20 percent.
- Key results:
  - Direct spillovers from Germany and Austria are the key contagion risks for BiH banks.
  - Indirect contagion from the U.K. could be substantial via cascade effects through other foreign banking systems.
  - Default of the Austrian banking system would cause five BiH banks’ regulatory capital to fall below the 12 percent minimum requirement (four in FBiH and one in RS), equivalent to a loss of 20 percent of BiH’s total regulatory capital.
  - In the credit shock scenario, joint default of Germany and Austria would impair nearly half of total regulatory capital of BiH’s banking system.
  - Adding funding shocks increases the number of undercapitalized banks, especially when shocks come from the top five systematically important countries for BiH.
  - Knock-on effects from exposures to foreign public sectors are very limited except for Romania and Slovenia, where government bonds are held in relatively large amounts by a few foreign banks in BiH.

### Domestic interconnectedness and contagion risks
- Major domestic interconnections:
  - Banks are mainly connected to development banks and insurance companies.
  - Largest three linkages reflect liabilities of the largest three RS banks to the development bank in RS (IRBRS).
  - Insurance deposits in banks amount to over KM 550mn (equivalent to about 40 percent of the insurance sector’s total assets but only 3 percent of total banking sector assets).
  - Bank claims and liabilities on all MCOs account for BAM 48mn and BAM 37mn (roughly 0.2 percent of banks’ assets), respectively.
- Dependence on public sector support:
  - In RS, IRBRS and six development funds exhibit large credit lines with all commercial banks in RS and hold sizeable shares and subordinated debt in some domestically-owned banks.
  - In FBiH, the development bank plays a less important role; some cantons and state-owned enterprises hold subordinated debt or capital in a few small domestically-owned banks.
  - Widespread cross-ownership of shares among small domestically-owned banks in FBiH, though amounts are fairly small.
- Interbank market spillovers:
  - Domestic interbank exposures are less than 1 percent of total regulatory capital.
  - Hypothetical default of any bank would not have significant cascade effects, except in one particular case where default of a large bank on interbank obligations would trigger undercapitalization of a small bank.
  - Some domestically-owned and medium-sized foreign-owned banks engage in “overnight” transactions (deposits, FX, cash) on the interbank market based on bilateral agreements; monthly reported exposures appear relatively small due to short maturities.
- Insurance sector spillovers:
  - Banks’ total claims on insurance sector: roughly KM 5mn.
  - Banks’ total liabilities to insurance sector: roughly KM 550mn (less than 3 percent of total banking sector deposits).
  - 80 percent of these deposits are concentrated in large foreign-owned banks; deposits at domestically-owned banks account for 3 percent of their total balance sheet.
  - In Adverse Scenario 2:
    - Five banks become insolvent.
    - Total claims of insurance companies on these five banks account for 3 percent of total insurance sector assets.
    - Joint default of these five banks implies capital impairment of around KM 43mn (14 percent of insurance sector’s capital).
    - At company level, two insurance companies would lose over half of total capital; one would lose its entire capital.

### Conclusions (stress-testing and assessment scope)
- Stress tests assessed financial stability of the banking sector in Bosnia and Herzegovina using:
  - Top-down stress tests performed jointly by the FSAP team and CBBH staff to assess solvency and liquidity of the entire banking system.
  - Bottom-up stress tests coordinated by both banking agencies and carried out by all 27 banks using their internal models and FSAP-provided macroeconomic scenarios.

*Italic: International Monetary Fund staff assessment based on supervisory data and IMF staff calculations as presented in the source content.*

### 53.      The quantitative analysis included macroeconomic scenario-based stress tests,

### _cr15213 - 53.      The quantitative analysis included macroeconomic scenario-based stress tests,

### Methodology
- Scenario-based stress tests used three full-fledged macroeconomic scenarios (one baseline and two adverse scenarios of varying severity) to assess the solvency of the banking system.
- Risk coverage included:
  - credit risk on the loan book,
  - issuer default risk on the debt instrument holding and equity investment portfolios,
  - market risk effects on interest income and valuation effects on debt instrument holdings,
  - exchange rate risks related to net open FX positions,
  - operational risk, among others.
- Sensitivity analysis was used to assess potential concentration risks.
- Network analysis evaluated contagion and spillover risks through both the interbank market and cross-border exposures.
- Liquidity stress tests were carried out based on proxies of the Basel III liquidity ratios (LCR and NSFR) to assess the liquidity positions of the banks.

### Key findings from the stress tests
- Credit risk:
  - Credit risk in the loan book is by far the most significant risk factor on the balance sheet of the banks.
  - Asset quality seems to be highly sensitive to macroeconomic conditions.
  - The presence of “unhedged borrowers” represents a significant inherent risk (difficult to quantify owing to lack of data).
- Interest rate risk:
  - Interest rate risk on interest income seems to be limited.
  - Several loans are essentially floating rate loans, and the resulting gaps between assets and liabilities (based on time-to-repricing buckets) are small.
- Exchange rate and market risks:
  - The direct effects of exchange rate risk seem to be contained, due to the relatively small net open FX positions reported by the banks.
  - Risks related to holdings of securities and trading portfolios are quite small, owing to the limited amount of such holdings in the banks’ balance sheet.
- Concentration risks:
  - Concentration risks are extremely high in a few (mainly domestically-owned) banks, where the default by just a few of their largest exposures would render these banks insolvent.
  - Concentration risks in the rest of the system appear more muted.
- Liquidity:
  - System-wide liquidity appears to be broadly appropriate.
  - A few banks in both entities present relatively low liquidity ratios.
  - Potential system-wide liquidity shortfalls appear to be manageable.
- Contagion and cross-border exposures:
  - Contagion risks through domestic interbank exposures appear to be contained.
  - Most banks present significant cross-border exposures, largely due to funding lines from parent companies (in the case of foreign-owned banks), and deposits for FX transactions in corresponding accounts abroad.
  - A couple of small banks exhibit “round-trip” investments, mainly with neighboring countries.
- Public sector support reliance:
  - Several domestically-owned banks seem to rely on public sector support, including both funding and capital support, either directly or indirectly (e.g., through the IRBRS).

### Conclusion and systemic implications
- System-wide solvency and liquidity indicators appear broadly adequate, but significant vulnerabilities remain in specific segments.
- Stress tests suggest that aggregate stress losses, mainly related to increased provisions in the loan book, remain manageable, although non-negligible.
- System-wide liquidity ratios appear broadly adequate.
- Several primarily small domestically-owned banks exhibit a wide range of significant vulnerabilities, including:
  - low liquidity ratios,
  - large concentration risks,
  - risky cross-border exposures,
  - questionable quality of some of their assets.
- Interconnectedness:
  - These vulnerable banks are somewhat interconnected among themselves but present low direct linkages with other domestic banks, so potential losses are likely to have limited direct spillovers to the rest of the banking system.
  - Indirect contagion risks (through e.g., reputational risks), not assessed in the stress tests, could pose significant risks to the stability of the system.

*Source: IMF staff analysis as presented in the provided chapter excerpt.*

### 56.      On the basis of the stress test analysis, a number of recommendations are warranted

### 56.      On the basis of the stress test analysis, a number of recommendations are warranted

### Key recommended focus areas
- Recommendations are mainly related to the largest sources of risk in the system:
  - credit risk in the loan book;
  - inherent risk from “unhedged borrowers;”
  - potential liquidity shortfalls (including in FX);
  - asset quality of weak banks;
  - etc.
- These recommendations are mostly aimed at addressing the limitations of the existing off-site supervisory framework in terms of monitoring these risks.

### Stress test—overall solvency results (system-wide, 27 banks; actual data as of 31-March-2014)
- Total regulatory capital - before shock: 2,938,843 (thousands of KM)
- Total risk-weighted assets (RWAs): 16,942,976
- Total assets: 21,783,153
- Total regulatory capital-ratio (CAR; in percent of RWAs) - before shock: 17.3 percent
- Stress test estimated capitalization after shock (2015):
  - Baseline (Top-Down): Total regulatory capital - after shock: 3,157,302; CAR after shock: 18.6 percent; Implied capital shortfall: -
  - Baseline (Bottom-Up): Total regulatory capital - after shock: 3,216,311; CAR after shock: 19.0 percent; Implied capital shortfall: 144 (thousands of KM)
  - Adverse Scenario 1 (Top-Down): Total regulatory capital - after shock: 2,753,549; CAR after shock: 16.3 percent; Implied capital shortfall: 26,147 (thousands of KM)
  - Adverse Scenario 1 (Bottom-Up): Total regulatory capital - after shock: 2,916,570; CAR after shock: 17.2 percent; Implied capital shortfall: 2,941 (thousands of KM)
  - Adverse Scenario 2 (Top-Down): Total regulatory capital - after shock: 951,673; CAR after shock: 5.6 percent; Implied capital shortfall: 1,147,991 (thousands of KM)
  - Adverse Scenario 2 (Bottom-Up): Total regulatory capital - after shock: 2,374,494; CAR after shock: 14.0 percent; Implied capital shortfall: 122,731 (thousands of KM)
- Number of banks with CAR below 12 percent:
  - Baseline: Top-Down -, Bottom-Up 1
  - Adverse Scenario 1: Top-Down 5, Bottom-Up 2
  - Adverse Scenario 2: Top-Down 22, Bottom-Up 13
- Number of banks with CAR below "zero":
  - Adverse Scenario 2 (Top-Down): 5

### Top-Down solvency results by entity (FBiH and RS; 31-March-2014)
- FBiH (17 banks) — before shock:
  - Total regulatory capital: 2,100,981
  - RWAs: 12,002,921
  - CAR before shock: 17.5 percent
- RS (10 banks) — before shock:
  - Total regulatory capital: 837,862
  - RWAs: 4,940,055
  - CAR before shock: 17.0 percent
- After shock (Adverse Scenario 2, Top-Down):
  - FBiH: Total regulatory capital - after shock: 641,986; CAR after shock: 5.3 percent; Implied capital shortfall: 859,457 (thousands of KM); Number of banks with CAR below 12 percent: 13; Number of banks with CAR below "zero": 5
  - RS: Total regulatory capital - after shock: 309,686; CAR after shock: 6.3 percent; Implied capital shortfall: 289,948 (thousands of KM); Number of banks with CAR below 12 percent: 9

### Stress losses components (selected, system-wide 2015)
- Stress test estimated losses after shock (2015) — examples (Top-Down / Bottom-Up where shown):
  - Credit risk: Increase in provisions due to loan migration:
    - Baseline TD: (4,067)
    - Baseline BU: (55,062)
    - Adverse 1 TD: (545,431)
    - Adverse 1 BU: (326,040)
    - Adverse 2 TD: (2,019,179)
    - Adverse 2 BU: (829,554)
  - Expected net losses on BIH government bond holding (HTM) - "issuer default risk":
    - Baseline TD: (1,267)
    - Baseline BU: (230)
    - Adverse 1 TD: (2,025)
    - Adverse 1 BU: (537)
    - Adverse 2 TD: (3,127)
    - Adverse 2 BU: (1,071)
  - Expected operational risk losses (examples):
    - Baseline TD: (1,810)
    - Baseline BU: (9,221)
    - Adverse 1 TD: (22,343)
    - Adverse 1 BU: (10,454)
    - Adverse 2 TD: (238,144)
    - Adverse 2 BU: (11,651)
  - Total net expected "stress losses":
    - Baseline TD: 730,724
    - Baseline BU: 668,328
    - Adverse 1 TD: 122,551
    - Adverse 1 BU: 365,411
    - Adverse 2 TD: (1,684,089)
    - Adverse 2 BU: (179,600)
- Other net income after shock (2015) — Total "other net income":
  - Baseline TD: (512,265)
  - Baseline BU: (390,860)
  - Adverse 1 TD: (307,845)
  - Adverse 1 BU: (387,684)
  - Adverse 2 TD: (303,081)
  - Adverse 2 BU: (384,748)

### Non-performing loan (NPL) outcomes under Adverse Scenario 2 (Implied NPL ratios, percent of total loans)
- Federation of Bosnia and Herzegovina (FBiH) — Top-Down:
  - 2014: Minimum 5.5; Median 12.2; Maximum 58.0
  - 2015: Minimum 16.7; Median 31.1; Maximum 82.7
  - 2016–2019: Median ranges 31.1 to 33.7 (detailed year values given in table)
- Republika Srpska (RS) — Top-Down:
  - 2014: Minimum 6.0; Median 13.7; Maximum 52.0
  - 2015: Minimum 18.0; Median 34.8; Maximum 77.8
  - 2016–2019: Median ranges 34.8 to 37.6 (detailed year values given in table)
- Bottom-Up series provided separately in the table (showing lower medians in later years for bottom-up compared with top-down in some years).

### Credit concentration risk (Table 9) — implied outcomes
- Default of the largest borrower (assumed recovery rates):
  - Recovery rate 36 percent:
    - System-wide CAR (27 banks): 15.3 percent
    - Implied capital shortfall (millions of BAM): 91
    - Number of banks with CAR < 12 percent: 9
    - Number of banks with CAR below "zero": 1
  - Recovery rate 0 percent:
    - System-wide CAR: 13.9 percent
    - Implied capital shortfall (millions of BAM): 166
    - Number of banks with CAR < 12 percent: 10
    - Number of banks with CAR below "zero": 2
  - Recovery rate 70 percent:
    - System-wide CAR: 16.3 percent
    - Implied capital shortfall (millions of BAM): 39
    - Number of banks with CAR < 12 percent: 6
    - Number of banks with CAR below "zero": -
- Default of the largest 5 borrowers (summary lines shown with higher implied shortfalls and numbers of undercapitalized banks for each recovery assumption).

### Liquidity stress test (Table 10) — system-level metrics
- LCR (including "required reserves"):
  - System-wide LCR (in percent): 291.7 (BiH, 27 banks); 348.6 (FBiH, 17 banks); 205.2 (RS, 10 banks)
  - Implied liquidity shortfall (in millions of BAM): 70.2 (BiH); 49.7 (FBiH); 20.4 (RS)
  - Number of banks with LCR below 100 percent: 5 (BiH); 2 (FBiH); 3 (RS)
- LCR (excluding "required reserves"):
  - System-wide LCR (in percent): 200.5 (BiH); 241.5 (FBiH); 138.1 (RS)
  - Implied liquidity shortfall (in millions of BAM): 324.6 (BiH); 142.7 (FBiH); 181.9 (RS)
  - Number of banks with LCR below 100 percent: 11 (BiH); 6 (FBiH); 5 (RS)
- NSFR:
  - System-wide NSFR (in percent): 123.0 (BiH); 127.4 (FBiH); 114.0 (RS)
  - Implied liquidity shortfall (in millions of BAM): 322.8 (BiH); 209.9 (FBiH); 112.9 (RS)
  - Number of banks with NSFR below 100 percent: 7 (BiH); 4 (FBiH); 3 (RS)

### Cross-border financial spillovers and contagion
- A bank is considered “undercapitalized” when its CAR falls below the 12 percent minimum requirement.
- Vakufska bank (FBiH) and Pavlovic international bank (RS) were excluded from spillover exercises because their pre-shock CARs are already below the minimum requirement.
- Figures show number of undercapitalized banks and capital impairment (in percent of total regulatory capital) under credit shocks and combined credit and funding shocks (detailed charts in source).

### Spillovers from banks to insurance sector (Adverse Scenario 2)
- Capital impairment to insurance sector measured in percent of total capital of insurance sector and by individual insurance company (detailed charts and per-company values provided in source).
- Note: An insurance company loses all its capital when capital impairment is larger than 100 percent.

### Macroeconomic scenarios used in stress tests (note on provenance)
- These scenarios are based on projections made in July 2014 and are period averages. The file notes: “Therefore, these numbers (in particular the projections for 2014) need to be interpreted in that context. In addition, the latest baseline projections from the IMF’s Area Department team might differ slightly from those presented here.”
- Baseline scenario — Real GDP growth (%) projections (selected years):
  - 2014: 0.7
  - 2015: 3.5
  - 2016: 3.7
  - 2017: 4.0
  - 2018: 4.0
  - 2019: 4.0
- Alternative scenario 1 and Alternative scenario 2 projections for Real GDP growth and other macro variables are provided in the table (detailed year-by-year series included in source).

### Data access and coverage for top-down stress tests
- The FSAP team’s access to supervisory data for top-down stress tests (availability status shown in table):
  - Data at the individual bank level: Fully available
  - Data aggregated along groups of banks: Partially available
  - Data aggregated at the banking system level: Fully available

*Source: IMF staff and authorities supervisory data as presented in the stress test chapter.*

### Appendix I. Preliminary Risk Assessment Matrix

### Appendix I. Preliminary Risk Assessment Matrix

### Summary of Main Risks and Staff Assessments
- Time horizon considered: Next 1–3 Years.
- Overall format: For each threat, the matrix lists (a) Staff assessment of overall level of concern, (b) Likelihood of severe realization of threat in the next 1–3 years (high, medium, or low), and (c) Expected impact on financial stability if threat is realized (high, medium, or low).

- Risk 1 — Bond market stress from a reassessment in sovereign risk in the euro area
  - Staff assessment (overall level of concern): Low.
  - Key channels and concerns:
    - Financial stress in the euro area could re-emerge and bank-sovereign-real economy links could re-intensify as a result of stalled or incomplete delivery of policy commitments.
    - Euro area corporate and bank deleveraging as well as fiscal drag could affect the growth outlook for the euro and lead to heightened turmoil in financial markets.
  - Likelihood of severe realization in next 1–3 years: Medium.
  - Expected impact on financial stability if realized: Medium.
  - Transmission to Bosnia and Herzegovina (BiH):
    - Euro area countries are BiH’s largest trading partners; export activity and revenues will be severely disrupted.
    - Remittances from euro area countries are sizeable and a large source of foreign exchange and banks’ liquidity and deposits.
    - FDI and other capital flows may also come to a halt, further deteriorating BiH’s growth outlook.
    - Given currency board arrangement, lower availability of FX will constrain base money.
    - Banks will likely suffer from funding retrenchment and asset quality deterioration due to the increase in NPLs.
  - Treatment in stress testing: Incorporated through an adverse macroeconomic scenario in the solvency stress test.

- Risk 2 — Geopolitical tensions surrounding Russia/Ukraine
  - Staff assessment (overall level of concern): Medium.
  - Key channels and concerns:
    - Geopolitical tensions create significant disruptions in global financial, trade and commodity markets.
    - Disruptions in commodity production or transport raise oil and gas prices in Europe and neighboring countries.
    - Increased uncertainty and lower confidence could trigger a permanent increase in risk aversion that reduces global equity prices.
  - Likelihood of severe realization in next 1–3 years: Low.
  - Expected impact on financial stability if realized: Low.
  - Transmission to BiH:
    - Russian banks own subsidiaries in European countries, which themselves are the parent companies for banks in BiH.
    - Banks would suffer from a potential funding retrenchment from the “grand-parent” banks.
    - Demand for BiH exports from Russia and neighboring countries would fall.
  - Treatment in stress testing: Incorporated through an adverse macroeconomic scenario in the solvency stress test.

- Risk 3 — External funding: Parent banks remove support of local subsidiaries and/or decide to withdraw from BiH market
  - Staff assessment (overall level of concern): Medium.
  - Key channels and concerns:
    - Euro area banks addressing balance sheet strains may opt to limit presence in the region as well as BiH.
    - AQR exercise may result in additional recapitalization needs and further balance sheet strains.
    - With limited capital available, BiH banks are at a disadvantage given complex regulatory and institutional system.
  - Likelihood of severe realization in next 1–3 years: Medium.
  - Expected impact on financial stability if realized: Medium.
  - Transmission to BiH:
    - BiH banking sector dominated by foreign banks, accounting for over 90 percent of the sector’s assets.
    - Subsidiaries of foreign banks remain reliant on parent bank support: while banks are becoming more reliant on domestic deposits, parent bank loans and deposits are still sizeable.
    - Parent bank capital is one of few options to raise fresh capital in BiH.
    - Repatriation of parent bank deposits may severely constrain liquidity in some large foreign-owned banks.
    - Confidence effects of foreign banks withdrawing from BiH could be significant; depositors still trust the foreign bank brand.
    - Domestic banks too small to compensate for intermediation loss.
  - Treatment in stress testing: Incorporated through the liquidity stress test.

- Risk 4 — Further deterioration in the health of commercial banks and confidence loss by bank depositors
  - Staff assessment (overall level of concern): Medium.
  - Key channels and concerns:
    - Slow economic activity could increase NPLs and require bank recapitalization.
    - A depositor confidence loss could lead to banking system liquidity shocks.
    - Bank recapitalization could be problematic in the absence of parent bank support.
    - Ongoing deterioration of asset quality: At about 15 percent to total loans, the high stock of NPLs constitutes a looming contingent liability for the state/public sector.
    - Authorities unlikely to have the resources/framework to deal with problems in a large bank.
    - Liquidity strains may turn into solvency problems.
    - Currency Board Arrangement and lack of LOLR facilities limit the range of options available.
  - Likelihood of severe realization in next 1–3 years: High.
  - Expected impact on financial stability if realized: High.
  - Treatment in stress testing: Incorporated through (i) an adverse macroeconomic scenario and (ii) a liquidity stress test.

- Risk 5 — Unavailability of official budget financing and sovereign debt restructuring
  - Staff assessment (overall level of concern): Low.
  - Key channels and concerns:
    - A challenging political and economic climate and difficult policy coordination could derail progress under the SBA and lead to unavailability of (or delay in) official budget support.
    - Entity governments have limited sources of alternative financing, already having saturated the domestic market.
    - Political uncertainty surrounding this year’s general elections could pose further fiscal policy risks.
  - Likelihood of severe realization in next 1–3 years: Low.
  - Expected impact on financial stability if realized: Low.
  - Transmission to BiH:
    - Sluggish growth and weak tax revenue may complicate servicing of domestic debt and a run-up in arrears, leading to additional austerity measures and weaker domestic demand, thus impacting banks’ credit activity.
    - Both foreign and domestic banks have accumulated government securities through purchases of entity debt; a potential restructure (or default) of entity government’s debt would somewhat impact banks’ balance sheets.
    - The official sector may be forced to accelerate the withdrawal of their banks’ deposits, compromising banks’ liquidity position.
  - Treatment in stress testing: Incorporated mainly through sensitivity analysis.

### Appendix II — Stress Test Matrix (STeM) for Solvency, Liquidity, and Contagion Risks

- General
  - Solvency top-down stress tests to be done jointly by the FSAP team and the staff at the CBBH.
  - Time horizon for stress tests: 5 years (2015–2019).

- Banking sector: Solvency risk
  - Institutional perimeter:
    - Institutions included: All banks (27 banks).
    - Market share covered: Percentage of total sector assets: 100 percent.
    - Data and baseline date: Supervisory data; banks’ own data; publicly available data.
  - Channels of risk propagation / Methodology:
    - Combination of banks’ own internal models and pre-defined benchmarks.
    - CBBH’s stress testing framework supplemented with the IMF balance sheet stress testing framework.
    - IMF balance sheet stress testing framework (tailor-made for the Bosnia and Herzegovina FSAP).
  - Satellite models for macro-financial linkages:
    - Banks’ own models for credit losses, pre-impairment income; expert judgment.
    - CBBH’s models supplemented with IMF’s econometric models for credit losses, pre-impairment income; expert judgment.
    - IMF’s econometric models for credit losses, pre-impairment income; expert judgment.
  - Tail shocks / Scenario analysis:
    - Shocks based on GDP trajectories, and translated in a consistent manner to all other variables in the macro-scenarios.
    - Three scenarios: baseline scenario; moderate external shock scenario; severe external shock scenario (implying output losses larger than those recorded in recent relevant historical crisis).
  - Sensitivity analysis:
    - Single-factor shocks: interest rate; exchange rate; sovereign securities loss/haircut.
    - Credit concentration risk.
  - Risks and buffers assessed:
    - Credit risk: credit risk on loan book; issuer default risk on government and corporate bond and other debt instrument holdings.
    - Market risk: interest rate risk impact on net interest income, government and corporate bond and other debt instrument holdings; FX risk.
    - Equity investment-related risk (includes both credit and market risk components).
    - Operational risk.
  - Behavioral adjustments:
    - Evolution of total assets and RWAs based on constant balance sheet assumption.
    - No management actions considered.
    - Other net income items, dividends, and taxes based on macroeconomic scenarios and pre-determined rules.
  - Regulatory and market-based standards and parameters:
    - Expected losses or loan migration (downgrades) and changes in provisions based on banks’ internal models (bottom-up) and satellite models (top-down).
    - Estimation of expected gains/losses on government and corporate bond holdings, real estate and equity investments based on banks’ internal models (bottom-up) and satellite models including gap and duration analysis (top-down).
    - Hurdle rates based on regulatory minimum for total regulatory capital (i.e., CAR of 12 percent).
    - Basel I rules applied.
  - Reporting format for results:
    - CAR, shortfall (if applicable).
    - Pass or fail; number of “undercapitalized” banks (i.e., with a CAR below 12 percent).
    - System-wide and by entity.

- Banking sector: Liquidity risk
  - Institutional perimeter:
    - Institutions included: All banks (27 banks).
    - Market share covered: Percentage of total sector assets: 100 percent.
    - Data and baseline date: Supervisory data; banks’ own data.
  - Channels of risk propagation / Methodology:
    - Basel III LCR-type proxy.
    - Basel III NSFR-type proxy.
  - Risks and buffers:
    - Risks: Market liquidity; Maturity mismatches.
    - Buffers: Counterbalancing capacity (HQLA, ASF).
  - Tail shocks:
    - Haircuts and run-off rates as defined in Basel III for LCR and NSFR.
  - Regulatory and market-based standards and parameters:
    - LCR proxy should exceed 100 percent (not a legal/regulatory requirement).
    - NSFR proxy should exceed 100 percent (not a legal/regulatory requirement).
  - Reporting format for results:
    - Pass rate, remaining buffers, and liquidity shortfall (if applicable).
    - System-wide and by entity.

- Banking sector: Contagion risk
  - Institutional perimeter:
    - Top-down by FSAP Team: All banks (27); All insurance companies (25).
    - Market share covered: Percentage of total sector assets: 100 percent.
    - Data and baseline date: Supervisory data; banks’ own data; publicly available data.
  - Channels of risk propagation / Methodology:
    - Network analysis, using Espinosa-Vega and Solé (2010) methodology.
  - Tail shocks:
    - Stress scenario with a credit shock: a severe stress in a bank or a banking system, causing a default on all of its liabilities to domestic institutions or foreign banks.
    - Stress scenario with a joint credit and funding shock when the default of a bank or a banking system also leads to a liquidity squeeze for those institutions funded by the defaulting bank or banking system.
  - Reporting format for results:
    - Capital impairment to domestic banking system, number of failed banks, and remaining buffers (at both banking-system level and bank level).
    - Capital impairment to domestic insurance sector, number of failed insurance companies, and remaining buffers (at both sector-wide level and company level).

*Source: IMF staff.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2015/_cr15213.pdf_