## _cr13261

## Source details

**Canonical URL:** [_cr13261](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13261.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13261.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13261.pdf.json)

---

### Executive Summary — Overview
- Banking system described as "well capitalized and liquid" based on stress test results.
- Regulatory stress-testing frameworks:
  - NBP conducts top-down macro and liquidity stress tests semi-annually.
  - KNF conducts annual bottom-up stress tests.
- Structure and recent history:
  - Foreign-owned banks account for "about two thirds" of the system by assets.
  - Top 5 banks account for "about 44 percent" of system assets.
  - Average annual real GDP growth rate in Poland between 2005 and 2011 was "4.5 percent".
- FSAP top-down coverage:
  - Covered "20 institutions" representing "85 percent of the assets" over a "five-year" horizon.

### Top-Down Solvency Tests — Scenarios, Assumptions, and Methods
- Scenarios:
  - Four macroeconomic scenarios: Baseline, V-shape, U-shape, L-shape generated with NBP’s NECMOD over a five-year horizon.
- Capital requirement (Basel III timeline) assumptions:
  - Minimum total capital requirement: "8 percent of RWA" for years 2013–2015; "8.625 percent of RWA" for 2016; "9.25 percent of RWA" for 2017.
  - Minimum Tier 1 capital requirement: "4.5 percent" in 2013; "5.5 percent" in 2014; "6 percent" in 2015; "6.625 percent" in 2016; "7.25 percent" in 2017.
  - Capital conservation buffer: "0.625 percent" in 2016; "1.25 percent" in 2017.
  - Results also calculated without the capital conservation buffer.
- Balance-sheet and behavioral assumptions:
  - Balance sheet composition assumed static.
  - Loans classified into corporate loans, consumer loans, and mortgages.
  - Bank balance sheets assumed to grow at the nominal rate of GDP (if positive) provided minimum capital requirements were met.
- Dividend policy rule (buffer above minimum => maximum dividend payout ratio):
  - "0 - 0.5" => "0" (percent of earnings).
  - "> 0.5 - 1.5" => "20" (percent of earnings).
  - "> 1.5 - 2.5" => "40" (percent of earnings).
  - "> 2.5" => "40 to 100" (percent of earnings); text indicates payout ratio set at "75 percent" in this range.
- Satellite models and RWA:
  - Panel data models for interest income, interest expenses, and credit losses estimated using quarterly data 1997–2012.
  - RWAs calculated following an approximate Basel II standardized approach; initial RWA weights kept fixed and adjusted for losses under the stress horizon.
- Aggregation:
  - Representative banks constructed by dividing sample into six percentile buckets based on CAR distribution; aggregation used due to data confidentiality.

### Top-Down Solvency Tests — Key Quantitative Findings
- NBP top-down tests: up to "30 percent of the analyzed group of banks" may not meet Basel III capital requirements in the recession scenarios; erosion of capital buffers concentrated up to 2015 and affects small banks.
- L-shape scenario capital need:
  - Total capital needs of about PLN "10.5 billion" (of which "2.5 billion" results from capital conservation buffer), equivalent to less than one percent of system assets.
- Representative banks:
  - Bottom three percentile buckets could face problems meeting CAR in first two years of recession scenarios; top three buckets recover after third year due to rebound in net income.
  - These representative banks represent around "30 percent of total loans".
- CAR distribution examples (NBP top-down and IMF top-down snapshots preserved exactly):
  - NBP baseline, 2012: Less than 10% = 10.94; Between 10% and 25% = 11.50; Between 25% and 50% = 12.78; Between 50% and 75% = 14.45; Between 75% and 90% = 16.44; Above 90% >=90% = 18.94.
  - NBP L-shape, 2013: Less than 10% = 7.72; Between 10% and 25% = 10.20; Between 25% and 50% = 11.04; Between 50% and 75% = 13.35; Between 75% and 90% = 15.17; Above 90% >=90% = 18.75.
  - IMF baseline, 2012: Less than 10% = 11.92; Between 10% and 25% = 11.97; Between 25% and 50% = 12.25; Between 50% and 75% = 13.22; Between 75% and 90% = 15.58; Above 90% >=90% = 18.94.
  - IMF V-shape, 2013: Less than 10% = 5.13; Between 10% and 25% = 5.22; Between 25% and 50% = 6.47; Between 50% and 75% = 7.72; Between 75% and 90% = 8.71; Above 90% >=90% = 13.88.
- Market-based top-down tests (EDF from equity prices via structural model):
  - Some banks could face substantial declines in capital-asset ratios; default of a few institutions cannot be ruled out under adverse scenarios.
  - Problems concentrated among smaller institutions.
  - Caveats: EDFs may be unreliable due to limited secondary market liquidity and foreign parent ownership limiting free-float information content.

### Bottom-Up Solvency Tests (KNF) — Coverage and Findings
- No new bottom-up tests specifically for the FSAP update; KNF provided 2012 annual regulatory bottom-up test distribution.
- KNF coverage: "32 banks" accounting for "90 percent" of assets.
- Scenarios: three 1½-year scenarios (baseline, adverse 1, adverse 2) with macro parameters including unemployment, inflation, exchange rate, interest rates, and residential real estate prices.
- Banks used internal models; KNF applied supervisory judgment in limited instances; banks allowed to factor in mitigating measures and policy responses.
- KNF 2012 bottom-up results:
  - Banking system resilient in adverse scenario.
  - No more than two banks, representing about "1 percent" of total banking system assets, fall below the regulatory minimum capital ratio in adverse scenarios.
  - Results qualitatively similar to NBP December 2012 Financial Stability Report.

### Liquidity Stress Tests — Design, Shocks, and Results
- Coverage: same "20 banks" as top-down solvency tests; supervisory data as of end-2012; baseline date "2012 Q4".
- Methodology: top-down balance-sheet test; output metric: ratio of liquid assets to liquidity needs.
- Shocks and channels (selected, preserved exactly):
  - Exchange rate/FX shocks: "30 percent" depreciation against the euro; increase in domestic currency value of FX liabilities; increased hedging needs.
  - Market/funding shocks: parallel widening of government yield curve by "300 bps"; decline in buffer of liquid assets; foreign capital outflow.
  - Run/confidence shocks: withdrawal of "100 percent" of deposits, "10 percent" of loans and "25 percent" of other liabilities with foreign institutions; partial withdrawal of stable deposits: Households: "5 percent"; Enterprises: "10 percent"; General government entities: "10 percent".
  - Other pressures: increased use of credit lines; activation of "10 percent" of off-balance sheet financing liabilities.
- Results:
  - Average weighted coverage of liquidity needs by liquid assets exceeded "100 percent" at system level.
  - Four banks experienced liquidity shortages amounting to about PLN "29 billion".
  - These four banks controlled slightly more than "10 percent of the total assets" in the system.
  - Forced liquidations: PLN "36 billion" of money bills and PLN "10½ billion" of government bonds.
  - Given average daily government bond turnover of about PLN "30 billion", such liquidation could lead to turmoil in secondary markets.
- Contagion/domino-effect simulation assumptions and outcomes:
  - Insolvency definition: breaching CAR of "4 percent".
  - December 2012 FSR simulation: "three commercial banks" could trigger a domino effect; second-round defaults could comprise up to "14 commercial and cooperative banks"; affected institutions have a share of total assets of less than "1¼ percent".
  - Drivers of limited interconnectedness: small interbank transactions, reduced interbank market usage post-2008, availability of client deposits and intragroup funding from foreign parents.
- Complementary tools:
  - BFG early warning system (EWS) is multidimensional but requires more back-testing to reduce type II errors.

### Sensitivity and Additional Stress Tests — Design and Results
- NBP sensitivity tests covered counterparty risk shocks, credit shocks, FX shocks, housing price shocks, and interest rate shocks.
- Selected sensitivity exercises and outcomes:
  - Counterparty risk shock: failure of five largest borrowers for each bank and failure of the five largest borrowers in the system.
    - Under the most severe counterparty risk shock, banks representing "14 percent of assets" could breach minimum capital requirements.
  - Credit risk shocks:
    - If one out of five performing loans becomes impaired (i.e., "20 percent"), banks representing "30 percent of assets" would not meet the minimum CAR.
    - If LGDs increase to "75 percent" and neither collateral nor profits can be used to cover losses, banks with an asset share of "10 percent" would breach the minimum CAR.
  - FX shock: "30 percent" depreciation against USD, euro and CHF; assessed impact on open FX positions, FX mortgage credit risk, and liquidity needs from FX swaps.
  - Housing price shock: "30 percent" decline in housing prices; impact on mark-to-market values.
  - Interest rate shock: "300 bps" parallel widening of government yield curve; bear flattening: 3-month rates widening "300 bps" and 5-10 year rates widening by "100 bps".
  - Market risk sensitivity tests had little impact on profits/losses and liquidity.
- Conclusion: Sensitivity tests validate solvency stress-test results; problems from recession scenarios could be manageable but concentrated in certain small banks.

### Market-Based Modeling of EDFs and Capital-to-Asset Impacts
- EDF–GDP nonlinear models estimated using sample period "2006 Q3 to 2012 Q3" (some banks 2010 Q3–2012 Q3).
- Model specifications fitted: one-term exponential y = ae^{bx}; rational polynomial y = (ax + b)/(x + c); second degree polynomial y = ax^{2} + bx + c.
- Exponential model coefficients (selected, preserved exactly):
  - Bank 1: a = 10.40 (6.45, 14.36); b = -0.56 (-0.76, -0.36)
  - Bank 2: a = 3.19 (1.92, 4.45); b = -0.45 (-0.62, -0.27)
  - Bank 4: a = 25.49 (10.39, 40.59); b = -0.34 (-0.56, -0.12)
  - Bank 5: a = 5.67 (2.88, 8.46); b = -0.33 (-0.51, -0.15)
  - Bank 8: a = 1.60 (1.09, 2.11); b = -0.52 (-0.67, -0.36)
  - Bank 9: a = 1.24 (0.97, 1.51); b = -0.38 (-0.46, -0.29)
  - Bank 10: a = 9.02 (0.64, 17.41); b = -0.58 (-0.93, -0.23)
  - Bank 12: a = 1.84 (1.28, 2.40); b = -0.42 (-0.55, -0.29)
  - Bank 13: a = 51.90 (15.15, 88.65); b = -0.35 (-0.57, -0.13)
- Rational polynomial and polynomial coefficients reported in source for Bank 6, Bank 7 and other groupings (values preserved exactly in original).
- Macro scenario GDP growth (year-on-year) table (selected rows preserved exactly):
  - 2012 Q3: 1.78, 1.80, 1.80, 1.80
  - 2012 Q4: 0.94, 0.94, 0.96, 0.95
  - 2013 Q1: 1.14, -0.14, 0.36, 0.14
  - 2013 Q2: 1.49, -1.78, -0.52, -1.01
  - 2013 Q3: 1.87, -3.24, -1.28, -1.99
  - 2013 Q4: 2.33, -3.89, -1.63, -2.46
  - 2014 Q1: 2.25, -2.74, -1.47, -2.14
  - 2014 Q2: 2.32, -0.36, -0.52, -1.01
  - 2014 Q3: 2.36, 2.00, 0.33, -0.05
  - 2014 Q4: 2.40, 3.66, 1.02, 0.55
  - 2015 Q1: 2.72, 4.35, 2.03, 1.43
  - 2015 Q2: 2.89, 4.38, 2.60, 1.83
  - 2015 Q3: 2.92, 3.99, 2.97, 2.03
  - 2015 Q4: 2.98, 3.60, 3.26, 2.28
  - 2016 Q1: 3.20, 3.54, 3.42, 2.33
  - 2016 Q2: 3.40, 3.58, 3.58, 2.39
  - 2016 Q3: 3.43, 3.62, 3.61, 2.32
  - 2016 Q4: 3.39, 3.66, 3.55, 2.16
  - 2017 Q1: 3.42, 3.76, 3.59, 2.15
  - 2017 Q2: 3.45, 3.84, 3.64, 2.21
  - 2017 Q3: 3.48, 3.91, 3.68, 2.26
  - 2017 Q4: 3.47, 3.91, 3.64, 2.26
- Projected impacts:
  - Banks generally experience declines in capital-asset ratios in the first two years of recession scenarios, reversed once growth resumes.
  - Under assumption asset values remain flat or decline during negative growth periods, banks could suffer capital losses ranging from "20 percent to 60 percent".
  - Certain banks could face recapitalization needs if stress scenarios realize.
  - Caveat: equity-implied measures sensitive to secondary market liquidity.

### Interconnectedness Risk and Contagion
- No new contagion test specifically for FSAP update; NBP domino-effect simulation (December 2012) results:
  - Only "three commercial banks" could trigger second-round defaults.
  - Banks affected by such contagion would represent "less than 2 percent of the assets" in the system (alternative phrasing elsewhere: affected institutions share less than "1¼ percent"; both figures preserved as reported in source sections).
- Overall assessment: interconnectedness risk limited.

### Conclusions and Policy Recommendations
- Aggregate conclusion:
  - All tests indicate the banking system is well capitalized to withstand large shocks; vulnerabilities concentrated in some small banks.
- Priority recommendation (from Summary of Recommendations):
  - "Enhance the coordination of the KNF’s bottom-up stress tests and the NBP’s top-down tests, and sharing of information."
  - Priority: "Low"
  - Time Frame: "Medium"
  - Time frame footnote: "“Immediate” is within 1 year; “near-term” is 1–3 years; and “medium-term” is 3–5 years."
- Additional recommendations implied by findings:
  - Small banks could benefit from strengthening their capital base.
  - Greater integration of top-down and bottom-up stress testing exercises would improve surveillance and supervisory policy.
  - Develop clearer synergies between stress tests and the EWS through enhanced cooperation and information sharing between departments in each agency responsible for stress testing.
  - KNF could examine whether it is appropriate to conduct on-site inspections more frequently, given that bottom-up stress test information is complemented with on-site supervision information.

*Source: Executive Summary and relevant sections of the IMF FSAP Update (Poland) as presented in the supplied document.*

### Executive Summary ......................................................................................................

### Executive Summary

### Overview
- The Polish banking system is described as "well capitalized and liquid" based on stress test results.
- Top-down and bottom-up regulatory stress testing frameworks are in place: the National Bank of Poland (NBP) conducts top-down macro and liquidity stress tests semi-annually; the Financial Supervision Authority (KNF) conducts annual bottom-up stress tests.
- The banking system structure and recent history:
  - Foreign-owned banks account for "about two thirds" of the system by assets.
  - The top 5 banks account for "about 44 percent" of system assets.
  - The average annual real GDP growth rate in Poland between 2005 and 2011 was "4.5 percent".
- The FSAP top-down exercises covered "20 institutions" representing "85 percent of the assets" in the system over a "five-year" horizon.

### Top-down solvency tests (NBP and FSAP)
- Scenarios:
  - Four macroeconomic scenarios were used: a baseline scenario and three recession scenarios labeled V-shape, U-shape, and L-shape. Scenarios were generated with the NBP’s NECMOD model over a five-year horizon.
- Key capital requirement assumptions (Basel III implementation timeline used):
  - Minimum total capital requirement: "8 percent of RWA" for years 2013–2015; "8.625 percent of RWA" for 2016; and "9.25 percent of RWA" for 2017.
  - Minimum Tier 1 capital requirement: "4.5 percent" in 2013; "5.5 percent" in 2014; "6 percent" in 2015; "6.625 percent" in 2016; and "7.25 percent" in 2017.
  - Capital conservation buffer: "0.625 percent" in 2016; and "1.25 percent" in 2017.
  - Results were also calculated without the capital conservation buffer.
- Balance-sheet and behavioral assumptions:
  - Balance sheet composition was assumed static during the stress tests.
  - Loans were classified into three categories: corporate loans, consumer loans, and mortgages.
  - Bank balance sheets were assumed to grow at the nominal rate of GDP (if positive) provided they met minimum capital requirements.
- Dividend policy rule used in tests (Table 2):
  - Buffer above minimum capital ratio "0 - 0.5" => Maximum dividend payout ratio "0" (percent of earnings).
  - Buffer above minimum capital ratio "> 0.5 - 1.5" => Maximum dividend payout ratio "20" (percent of earnings).
  - Buffer above minimum capital ratio "> 1.5 - 2.5" => Maximum dividend payout ratio "40" (percent of earnings).
  - Buffer above minimum capital ratio "> 2.5" => Maximum dividend payout ratio "40 to 100" (percent of earnings) — note text also indicates in the range of buffer over 2.5 the dividend payout ratio was set at "75 percent".
- Main findings:
  - Both top-down tests (NBP and FSAP) indicated that the banking system is resilient in aggregate even under severe adverse scenarios.
  - Some small banks could fail to meet Basel III capital requirements under recession scenarios; these small banks together represent up to "30 percent of the assets" in the system (FSAP top-down balance-sheet tests using representative banks).
  - Caution: the FSAP balance-sheet top-down tests were conducted using representative banks due to data confidentiality, so impacts on representative banks may not map precisely to individual banks.
  - FSAP market-based solvency stress tests showed some smaller banks could face substantial declines in their capital asset ratios.

### Bottom-up solvency tests (KNF)
- No new bottom-up solvency tests were conducted specifically for the FSAP update.
- KNF provided the distribution of results from its 2012 annual regulatory bottom-up stress testing exercise.
  - The KNF bottom-up tests used different scenarios and a shorter horizon ("1½ years"), but results were consistent with top-down solvency tests.

### Liquidity stress tests
- NBP conducted a top-down liquidity stress test including simultaneous realization of several shocks (withdrawal of foreign funding, domestic deposit outflows, margin calls on FX hedges, and others).
- Main findings:
  - The banking system could withstand large liquidity shocks in aggregate.
  - Some small banks, representing "10 percent of assets" in the system, could face difficulties covering liquidity needs.
  - Forced sale of liquid assets could put pressure on the secondary market for government bills and bonds.

### Sensitivity and additional stress tests
- NBP conducted comprehensive sensitivity tests including counterparty risk shocks, credit shocks, FX shocks, housing price shocks, and interest rate shocks.
- Key result:
  - Counterparty risk shocks were the most severe, with banks breaching minimum capital requirements representing around "14 percent of assets" in the system.
  - The impact of market risk shocks on credit losses and/or liquidity needs was small.

### Interconnectedness risk
- No new contagion test was conducted specifically for the FSAP update.
- NBP shared results of a domino-effect simulation from its December 2012 report:
  - Only "three commercial banks" could trigger second-round defaults in that simulation.
  - The banks affected by such contagion would represent "less than 2 percent of the assets" in the system.
- Overall assessment: interconnectedness risk in the system is limited.

### Conclusions and policy recommendations
- Aggregate conclusion:
  - All tests indicate the banking system is well capitalized to withstand large shocks; vulnerabilities are concentrated in some small banks.
- Recommendation (from Summary of Recommendations):
  - "Enhance the coordination of the KNF’s bottom-up stress tests and the NBP’s top-down tests, and sharing of information."
  - Priority: "Low"
  - Time Frame: "Medium"
  - Time frame footnote: "“Immediate” is within 1 year; “near-term” is 1–3 years; and “medium-term” is 3–5 years."
- Additional recommendation implied by findings:
  - Small banks could benefit from strengthening their capital base.
  - Greater integration of top-down and bottom-up stress testing exercises would improve surveillance and supervisory policy.

*Source: Executive Summary of the IMF FSAP Update (Poland).*

### 11.      For the 2013 FSAP Update, the NBP used its stress testing framework to conduct

### _cr13261 - 11.      For the 2013 FSAP Update, the NBP used its stress testing framework to conduct

### Top-Down Solvency Stress Tests — framework and models
- The stress tests included the 20 largest banks comprising about 85 percent of the total assets in the banking system.
- Scenarios: a baseline scenario and three adverse scenarios (V-shape, U-shape, L-shape).
- Satellite models mapped scenarios into credit losses and net income:
  - Panel data models for interest income on loans, interest expenses, and credit losses (expressed as a ratio of loans, interest bearing liabilities, and credit portfolios respectively).
  - Models estimated using quarterly data for the period 1997–2012.
  - Loan portfolio divided into three categories: commercial loans, mortgage loans, and consumer loans.
  - Model projections subject to judgment-based adjustment to account for regulatory and market changes in lending standards.
- RWAs:
  - Changes in RWA calculated following an approximate Basel II standardized approach.
  - The standardized approach determined initial RWA weights, which were kept fixed during the stress scenarios.
  - RWAs adjusted to account for losses under the stress test horizon.
- Aggregation and representative banks:
  - Representative banks constructed by dividing sample into six percentile buckets based on the regulatory capital adequacy ratio (CAR) distribution at the end of the NBP stress test and aggregating capital, credit risk exposures, and RWAs of banks in each bucket.
  - Aggregation used owing to data confidentiality; could potentially underestimate risks.

### Top-Down Solvency Stress Tests — key quantitative findings
- In the NBP top-down stress tests, up to 30 percent of the analyzed group of banks may not meet Basel III capital requirements in the recession scenarios.
- Erosion of capital buffers is mainly concentrated in the period up to 2015 and affects small banks.
- The most severe scenario (L-shape recession) leads to total capital needs of about PLN 10.5 billion (out of which 2.5 billion would result from capital conservation buffer kicking in), equivalent to less than one percent of the assets in the banking system.
- The FSAP top-down solvency test methodology:
  - Capital one period ahead = current capital + current net income − credit losses − dividend payouts.
  - PD approximated as the ratio of the flow of nonperforming loans (NPLs) to total loans in the category.
  - LGD estimates calculated as NPL coverage by loan loss provisions; NPL flows provided by the central bank.
- CAR distribution snapshots (selected figures preserved exactly as in source tables):
  - NBP top-down CAR distribution (percentile buckets) — examples:
    - Baseline scenario, 2012: Less than 10% = 10.94, Between 10% and 25% = 11.50, Between 25% and 50% = 12.78, Between 50% and 75% = 14.45, Between 75% and 90% = 16.44, Above 90% >=90% = 18.94.
    - L-shape scenario, 2013: Less than 10% = 7.72, Between 10% and 25% = 10.20, Between 25% and 50% = 11.04, Between 50% and 75% = 13.35, Between 75% and 90% = 15.17, Above 90% >=90% = 18.75.
  - IMF top-down CAR distribution (percentile buckets) — examples:
    - Baseline scenario, 2012: Less than 10% = 11.92, Between 10% and 25% = 11.97, Between 25% and 50% = 12.25, Between 50% and 75% = 13.22, Between 75% and 90% = 15.58, Above 90% >=90% = 18.94.
    - V-shape scenario, 2013: Less than 10% = 5.13, Between 10% and 25% = 5.22, Between 25% and 50% = 6.47, Between 50% and 75% = 7.72, Between 75% and 90% = 8.71, Above 90% >=90% = 13.88.
- Representative banks corresponding to the bottom three buckets could face problems meeting CAR in the first two years of the recession scenarios; representative banks in top three buckets recover due to a strong rebound in net income after the third year.
- These representative banks represent around 30 percent of total loans.

### Top-Down Market-Based Stress Tests
- Market-based tests used expected default frequencies (EDF) implied from equity prices (Moody’s analytics) mapped via a structural credit risk model to changes in capital-asset ratios.
- Results:
  - Some banks could face substantial declines in capital-asset ratios; default of a few institutions cannot be ruled out under adverse scenarios.
  - Problems concentrated among smaller institutions.
  - High capital adequacy ratios in the system suggest that despite large declines, capital adequacy ratios could remain above regulatory minimum levels.
- Caveats:
  - Equity-price-based EDFs may be unreliable due to lack or scarcity of secondary market liquidity.
  - Foreign parent banks often hold controlling shares of listed Polish banks, limiting the information content of free-float equity prices.

### Bottom-Up Solvency Stress Tests — coverage and scenarios
- No specific bottom-up stress tests were conducted for the FSAP update; authorities shared the latest annual regulatory test results.
- The KNF bottom-up test covered 32 banks accounting for 90 percent of assets in the system.
- Three 1½-year scenarios: baseline (central bank projections), adverse 1 (negative external shocks), adverse 2 (domestic shocks).
- Additional macro parameters specified: unemployment rate, inflation, exchange rate, short and long term interest rates, residential real estate prices.
- Banks used internal models to map scenarios to credit risk parameters; KNF in limited instances amended responses using supervisory judgment.
- Banks allowed to factor in mitigating measures and authorities’ policy responses (e.g., in the second scenario interest rates raised steeply to stem an outflow of capital).

### Bottom-Up Solvency Stress Tests — key quantitative findings
- KNF 2012 bottom-up stress tests indicate banking system resilience in an adverse scenario.
- No more than two banks, representing about 1 percent of total banking system assets, fall below the regulatory minimum capital ratio in the adverse scenarios.
- Results qualitatively similar to NBP December 2012 Financial Stability Report.

### Sensitivity Tests — design and outcomes
- NBP semiannual sensitivity tests complemented top-down stress tests: counterparty risk shocks, credit risk shocks, sharp depreciation, housing price shocks, interest rate shocks.
- FSAP-specific sensitivity tests compared to NBP tests (selection from Table 7):
  - Counterparty risk: Failure of five largest financial or non-financial borrowers for each bank; failure of the five largest borrowers in the banking system.
  - Credit risk shocks: Maximum increase in the PD of performing loans that does not breach CAR; increases in LGD to 50 percent and 75 percent of impaired loans.
  - Foreign exchange shock: Depreciation of 30 percent against the US dollar, the euro and the swiss franc; impact on open FX positions; credit risk of FX mortgage loans; liquidity needs of banks using FX swaps and cross-currency swaps.
  - Housing price shock: 30 percent decline in housing prices; impact on mark-to-market value of banks' assets.
  - Interest rate shock: 300 bps parallel widening of government yield curve and corresponding impact on interest rate swaps; bear flattening: 3-month rates widening of 300 bps and 5-10 year rates widening by 100 bps.
- Sensitivity test results:
  - Under the most severe counterparty risk shock, banks representing 14 percent of total assets could breach minimum capital requirements.
  - Credit risk shocks:
    - If one out of five performing loans becomes impaired (i.e., 20 percent), banks representing 30 percent of assets would not meet the minimum CAR.
    - If LGDs increase to 75 percent and neither collateral nor profits can be used to cover losses, banks with an asset share of 10 percent would breach the minimum CAR.
  - Market risk sensitivity tests had little impact on profits and losses and liquidity.
- Conclusion: Sensitivity tests validate the solvency stress test results, suggesting problems arising from recession scenarios could be manageable.

*Source: NBP; KNF; Staff calculations as presented in the FSAP Update text.*

### 28.      The central bank conducted a liquidity stress test. The test included the same

### The central bank conducted a liquidity stress test

### Liquidity stress test design and scope
- Institutions included: the same twenty banks analyzed in the top-down stress tests; names undisclosed.
- Baseline and data timing:
  - Supervisory data as of end-2012.
  - Baseline date: 2012 Q4.
- Methodology:
  - Top-down, separate test using balance sheet data.
  - Output metric: ratio of liquid assets to liquidity needs.
- Insolvency definition in contagion simulations: breaching a capital adequacy ratio of 4 percent.
- Contagion simulation coverage: commercial banks and cooperative banks; netting of mutual assets and liabilities applied; only commercial banks allowed to fail in the first round.

### Shocks and impact channels (Table 8 summary)
- Exchange rate and FX-related shocks:
  - 30 percent depreciation of the currency against the euro,
  - Increase in the domestic currency value of FX liabilities (the US dollar and the Swiss franc),
  - Increase in the value of domestic currency funds needed for hedging long FX positions in the balance sheet.
- Market and funding shocks:
  - Parallel widening of the government yield curve by 300 bps (government bonds),
  - Decline in the value of the buffer of liquid assets (government bonds),
  - Foreign capital outflow.
- Run and confidence shocks:
  - Withdrawal of 100 percent of deposits, 10 percent of loans and 25 percent of other liabilities with foreign institutions,
  - Drop in mutual confidence among domestic financial institutions; closing of domestic interbank market; complete settlement of all short-term deposits,
  - Drop in confidence of banks' clients with partial withdrawal of stable (core) deposits: Households: 5 percent; Enterprises: 10 percent; General government entities: 10 percent.
- Other liquidity pressures:
  - Increased use of credit lines extended by banks,
  - Activation of 10 percent of off-balance sheet financing liabilities.
- Credit risk sensitivity (graphic elements preserved): LGD increase to 50%; LGD increase to 75%; scenario notes: no collateral and profits can be used to cover losses.

### Liquidity stress test results and market implications
- System-level coverage:
  - Average weighted coverage of liquidity needs by liquid assets exceeded 100 percent.
- Banks facing shortages:
  - Four banks experienced liquidity shortages amounting to about PLN 29 billion.
  - These four banks controlled slightly more than 10 percent of the total assets in the system.
- Forced asset liquidations under the scenario:
  - Banks would be forced to liquidate PLN 36 billion of money bills and PLN 10½ billion of government bonds.
  - Given the average daily market turnover for government bonds is about PLN 30 billion, asset liquidation could lead to some turmoil in secondary markets.

### Interconnectedness and contagion simulation findings
- Semiannual top-down contagion/domino-effect simulations use detailed interbank claims and liabilities (counterparty names included).
- Simulation assumptions:
  - Netting of mutual assets and liabilities.
  - All types of loans and interbank deposits accounted for regardless of maturity.
  - Insolvency defined as breaching a capital adequacy ratio of 4 percent.
- Simulation outcomes (December 2012 FSR results):
  - Three commercial banks could trigger a domino effect.
  - Second-round defaults could comprise up to 14 commercial and cooperative banks.
  - These affected institutions have a share of total assets in the banking system of less than 1¼ percent.
  - The failure of one trigger bank did not affect other trigger banks.
- Drivers of limited interconnectedness risk:
  - Small transactions in the interbank market,
  - Reduced interbank market usage post-2008 crisis,
  - Availability of client deposits and intragroup funding from foreign parent entities.

### Complementary monitoring tools and limitations
- BFG early warning system (EWS):
  - Multidimensional design using a granular and flexible database.
  - Combines indicators via a methodology drawing on reliability theory.
  - More back-testing required to reduce the risk of type II errors and enhance credibility for payout preparation.

### Conclusions and policy recommendations
- System resilience and strengths:
  - The domestic banking system is well capitalized; capital levels are well above those required by Basel III and of high quality.
  - Regulatory guidelines limiting dividend repayments have contributed to substantial capital buffers.
  - Banks show high ROE and ROA, supported by strong recent Polish economic growth and relatively low banking penetration.
- Stress-test implications:
  - Adverse scenarios would affect mainly small banks accounting for a limited share of assets.
  - Low interconnectedness limits second-round defaults even if a large bank fails.
  - Liquidity buffers appear adequate, but realization and ensuing sale of liquid assets could create volatility and downward price pressure on government securities markets.
- Recommended enhancements:
  - Develop clearer synergies between stress tests and the EWS through enhanced cooperation and information sharing between departments in each agency responsible for stress testing.
  - KNF could examine whether it is appropriate to conduct on-site inspections more frequently, given that bottom-up stress test information is complemented with on-site supervision information.

### Market-based top-down stress test (annex highlights)
- Purpose:
  - Complement balance-sheet top-down and NBP tests using market estimates of default risk to infer implied changes in capital-asset ratios under macro scenarios.
- Methodology overview:
  - Uses the link between probability of default and capital structure (Merton, 1974).
  - For each bank, the one-year probability of default corresponded to the one-year Expected Default Frequency (EDF) calculated by Moody’s Analytics.
- Key findings:
  - Some banks could face substantial declines in capital-asset ratios under adverse scenarios.
  - Default of a few institutions cannot be ruled out under adverse scenarios, but problems are concentrated among smaller institutions and high system capital adequacy ratios suggest capital ratios could remain above regulatory minimum levels despite large declines.

*Source: NBP; supervisory and FSAP team analysis as reported in the referenced document.*

### 42.      Simple nonlinear models were estimated to link the behavior of the EDF to year-

### Simple nonlinear models were estimated to link the behavior of the EDF to year-on-year changes of real GDP

### Model estimation and sample
- Sample period: 2006 Q3 to 2012 Q3.
- For some banks, models were estimated using only data for the period 2010 Q3–2012 Q3.
- The more sophisticated one-factor modeling approach of Hamerle, Jobst, Knapp, and Lerner (2008) was also used but it was not possible to fit the models to all the banks. The results for banks fitted with one-factor models were very similar to those reported here.
- For two banks, model fit was somewhat counterintuitive for growth rates exceeding 7 percent; these growth values did not realize in the stress scenarios.

### Model specifications
- Three different models were fitted to link banks’ expected default frequencies (EDFs) to year-on-year real GDP changes:
  - One-term exponential model: y = ae^{bx}
  - Rational polynomial: y = (ax + b)/(x + c)
  - Second degree polynomial: y = ax^{2} + bx + c
- The 95 percent confidence levels are shown within brackets in the coefficient tables.

### Exponential model coefficients (as reported)
- Bank 1: a = 10.40 (6.45, 14.36); b = -0.56 (-0.76, -0.36)
- Bank 2: a = 3.19 (1.92, 4.45); b = -0.45 (-0.62, -0.27)
- Bank 4: a = 25.49 (10.39, 40.59); b = -0.34 (-0.56, -0.12)
- Bank 5: a = 5.67 (2.88, 8.46); b = -0.33 (-0.51, -0.15)
- Bank 8: a = 1.60 (1.09, 2.11); b = -0.52 (-0.67, -0.36)
- Bank 9: a = 1.24 (0.97, 1.51); b = -0.38 (-0.46, -0.29)
- Bank 10: a = 9.02 (0.64, 17.41); b = -0.58 (-0.93, -0.23)
- Bank 12: a = 1.84 (1.28, 2.40); b = -0.42 (-0.55, -0.29)
- Bank 13: a = 51.90 (15.15, 88.65); b = -0.35 (-0.57, -0.13)

### Rational polynomial coefficients (as reported)
- a = -53.56 ( -759.41, 652.29 ); -19.56 ( -112.67, 73.55 )
- b = 249.71 ( -3019.78, 3519.21 ); 93.75 ( -341.96, 529.47 )
- c = 29.82 ( -385.67, 445.31 ); 8.65 ( -40.16, 57.4665 )

(Note: table entries list paired coefficient columns corresponding to Bank 6 and Bank 7; values are preserved as presented.)

### Polynomial curve coefficients (as reported)
- For one bank grouping:
  - a = 0.08 (0.02, 0.13)
  - b = -0.82 (-1.26, -0.38)
  - c = 2.26 (1.42, 3.10)
- For the other bank grouping:
  - a = 0.15 (0.06, 0.24)
  - b = -1.66 (-2.47, -0.85)
  - c = 4.62 (3.06, 6.17)

### Macro scenarios (GDP growth year-on-year) — Table 2
- Scenarios: Baseline; V-shaped recession; U-shaped recession; L-shaped recession.
- 2012 Q3: 1.78, 1.80, 1.80, 1.80
- 2012 Q4: 0.94, 0.94, 0.96, 0.95
- 2013 Q1: 1.14, -0.14, 0.36, 0.14
- 2013 Q2: 1.49, -1.78, -0.52, -1.01
- 2013 Q3: 1.87, -3.24, -1.28, -1.99
- 2013 Q4: 2.33, -3.89, -1.63, -2.46
- 2014 Q1: 2.25, -2.74, -1.47, -2.14
- 2014 Q2: 2.32, -0.36, -0.52, -1.01
- 2014 Q3: 2.36, 2.00, 0.33, -0.05
- 2014 Q4: 2.40, 3.66, 1.02, 0.55
- 2015 Q1: 2.72, 4.35, 2.03, 1.43
- 2015 Q2: 2.89, 4.38, 2.60, 1.83
- 2015 Q3: 2.92, 3.99, 2.97, 2.03
- 2015 Q4: 2.98, 3.60, 3.26, 2.28
- 2016 Q1: 3.20, 3.54, 3.42, 2.33
- 2016 Q2: 3.40, 3.58, 3.58, 2.39
- 2016 Q3: 3.43, 3.62, 3.61, 2.32
- 2016 Q4: 3.39, 3.66, 3.55, 2.16
- 2017 Q1: 3.42, 3.76, 3.59, 2.15
- 2017 Q2: 3.45, 3.84, 3.64, 2.21
- 2017 Q3: 3.48, 3.91, 3.68, 2.26
- 2017 Q4: 3.47, 3.91, 3.64, 2.26
- Source of scenarios: IMF staff estimates.

### Projected impacts on capital-to-asset ratio and implications
- Once the nonlinear models were estimated, projections of changes in the capital-to-asset ratio under the macro scenarios were produced.
- In general, banks experience a decline in their capital asset ratios in the first two years of the recession scenarios.
- The declines are reversed after the economy starts posting positive growth rates.
- Under the assumption that asset values remain flat or decline during the negative growth periods in the scenarios, the analysis shows that banks could suffer substantial capital losses ranging from 20 percent to 60 percent.
- It cannot be ruled out that certain banks could face recapitalization needs if the stress scenarios were to realize.
- Caution: As with any market-based model, secondary liquidity in the stock market could affect the reliability of equity-implied measures of default risk.

### Visual summaries (as presented)
- Figure 1: Fitted nonlinear EDF-versus-GDP growth curves and fitted EDF points for Banks 1–13 (graphs preserved in original).
- Figure 2: Changes in the capital-to-asset ratio of Polish banks under different macro scenarios relative to 2012 Q3 levels (percent deviations by bank and scenario; charts preserved in original).
- Source for figures and model inputs: Moody’s Analytics; and author’s calculations.

*Source: Moody’s Analytics; and author’s calculations.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2013/_cr13261.pdf_