## _cr14210 ## Source details **Canonical URL:** [_cr14210](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14210.pdf) ## Other formats - [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14210.pdf.md) - [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14210.pdf.json) --- ### Introduction and objective - The FSAP stress testing exercise comprises a comprehensive analysis of solvency and liquidity risks of the banking sector, using mid-2013 data (cut-off: end-June 2013 / mid-2013). - Purpose: - Determine capacity of the banking sector to absorb realization of key macro-financial risks based on capital adequacy and sufficiency of liquidity under stress. - Examine system-wide resilience over the medium-term, uncover vulnerabilities to rapid macroeconomic deterioration, and identify potential threats to financial stability. - Scope and coverage: - Solvency tests: BU stress test of selected locally incorporated, licensed banks and cross-validation by three TD tests covering nearly all local banks. - Liquidity tests: TD approaches within existing liquidity reporting framework, using supervisory data and parameters specified by the HKMA and the FSAP team. - Sample coverage: up to 19 local banks and eight foreign branches; TD solvency tests cover more than 99 percent of all locally incorporated licensed banks; various stress tests cover around 50 to around 70 percent of the banking sector’s total assets (accounting for up to 78 percent of total deposits and 66 of total loans). ### Stress-test design — solvency (framework and scenarios) - Scenario framework: - Three macro scenarios over a five-year forecast horizon (2014–2018): Baseline; Slow Growth (SG); Severe Adverse (SA). - SG: cumulative negative deviation of about 9.1 percentage points in real GDP growth over five years; average annual growth in HKSAR deteriorates by 1.8 percentage points (to 2.6 percent) relative to baseline average annual growth of 4.4 percent. - SA: cumulative negative deviation of about 12.1 percentage points in real GDP growth over five years; economic growth deteriorates by 2.4 percentage points (to 2.0 percent) on average relative to baseline. - Asset-price shocks under adverse scenarios: property prices decline by 30 and 40 percent respectively; equity prices decline by 50 and 65 percent respectively (headline summaries); HKMA TD maps comparable magnitudes condensed into two years (e.g., property prices fall by 30 percent and 50 percent under “severe” and “more severe” respectively). - Risk channels and components tested: - (i) Impairment charges and mark-to-market valuation changes of fixed income securities (trading and banking book). - (ii) Changes in pre-impairment income, including funding costs. - (iii) Changes in risk-weighted assets (RWAs). - Inclusion of rising risk aversion (market-implied measures), cross-border developments (sovereign risk), foreign currency risk, and Basel III regulatory reforms. - Methodologies: - BU: firms run internal-model BU tests under FSAP guidance (selected local banks). - TD (IMF balance-sheet): panel-satellite models (2SLS/GMM) on quarterly data 1996:Q1–2013:Q2 to map macro-financial linkages (19 local banks). - TD (HKMA balance-sheet): supervisory-element TD test (19 local banks). - TD (IMF SCCA): market-based SCCA (9 publicly listed local banks; daily market data Jan 1997–Aug 2013; five-year rolling aggregation). ### Solvency test results — headline findings - Aggregate resilience: - Stress test results confirm a high degree of resilience; analyses based on prudential data suggest that even a severe economic shock would not result in an aggregate capital shortfall over a five-year forecast horizon. - Banks hold very high levels of capital, are very profitable, and have a low level of asset impairments amid stable funding profiles. - Quantitative impacts (selected exact figures preserved): - Under TD balance-sheet approaches: aggregate CET1 ratio declines at most by 2.7 percentage points and 0.9 percentage points under the two adverse scenarios (HKMA and IMF TD approaches, respectively). - CAR contracts by up to 3.7 percentage points under the SA scenario across all balance-sheet approaches (TD and BU). - Under the IMF SCCA (market-consistent, joint tail risks at a 1-in-200 year probability): aggregate CAR would decrease by 4.5 percentage points; aggregate CET1 ratio would decrease by 4.3 percentage points. - BU exercise: CET1 ratios diverge by as much as 2.3 percentage points from pre-stress capitalization; maximum decline of aggregate capitalization amounts to 3.3 percentage points of CAR. - Bank-level vulnerabilities: - Most large banks exhibit very solid capital buffers and can withstand severe deterioration. - Some smaller banks may be slightly more vulnerable: potential decline in net operating profitability and medium-term capital constraints under severe shock. - Identified potential capital shortfall (one TD approach): HKD 1.3 billion of Tier 1 capital (about 0.2 percent of Tier 1 capital in the sector as of end-June 2013). - In two out of three TD approaches, one to two small banks would fall below hurdle rates toward the end of the forecast horizon; none would fail in two TD approaches at the same time. - SCCA tail-risk outcome: - Under SCCA conditional tail expectation (CTE/expected shortfall) at 95 percent or higher, the sector remains solvent and stays above all hurdle rates over the entire stress horizon, though market-implied capitalization of some banks declines rapidly. ### Stress-test design — liquidity (framework and scenarios) - Focus: - Sudden, sizable withdrawals of funding and sufficiency of existing assets to withstand shocks under stressed conditions; inclusion of funding inflows/outflows, haircuts on assets, and contingent liabilities. - Tools applied: - HKMA regulatory standards and supervisory tests: 7-day test/LOLR analysis; Enhanced Liquidity Stress Test (ELST) spanning one to three months. - FSAP ICF tests: 5-day cumulative and 30-day non-cumulative implied cash flow tests. - Basel III measures: LCR and NSFR (applied to selected banks). - Market risk–contingent liquidity approach (Wong and Hui, 2009) applied to 17 local banks (consolidated). - Coverage: - Liquidity TD stress tests applied to all 19 Group 1 local banks and 8 Group 2 foreign branches on a solo basis; represents 71 percent (70.9 percent noted elsewhere) of sector total assets. ### Liquidity test results — headline findings and exact metrics - General finding: - Liquidity risks are low; banks exceed—and in most cases by a large margin—the minimum liquidity ratios and threshold requirements across stress metrics due to high cash balances and large stocks of liquid assets at low encumbrance levels. - Short-lived shock results: - Average stress test ratio = 279 percent under HKMA’s 7-day/LOLR analysis. - Average stress test ratio = 338 percent under the IMF’s five-day ICF test. - One-month and longer horizons: - Average liquidity ratio = 162 percent under HKMA’s ELST (one-month horizon). - Average liquidity ratio = 235 percent under IMF’s 30-day ICF test. - Basel III standard measures (limited sample, end-June 2013 positions): - Average LCR and NSFR above 100 percent on consolidated and combined bases. - On combined basis, a few banks show potentially insufficient LCR buffers for 30-day horizon, though most report LCR ratios above 60 percent (initial minimum requirement in 2015 transition) and many close to 100 percent. - Most sample banks report NSFR values of 100 percent or higher; a couple show NSFR < 100 percent but shortfall not material. - Wong and Hui (2009) one-year horizon: - No bank triggers cash shortage or default due to liquidity problems over one-year severe asset-price shock horizon; cash inflows sufficient to cover cash outflows for all sample banks. - Funding structure drivers: - Banks generally not reliant on wholesale funding; contractual maturity mismatches increase beyond one year due to long-tenor residential mortgage portfolios and large retail deposit base. - Foreign branches more dependent on wholesale funding and more exposed to intragroup claims; in HKMA ELST local banks’ outflows driven ~80 percent by deposits, foreign branches’ outflows driven ~70 percent by wholesale funding. ### Systemic Contingent Claims Approach (SCCA) — core elements (Box 1) - SCCA combines multivariate contingent claims analysis (CCA) with extreme value theory (EVT) to generate system-wide tail risk. - Two-step / three-step implementation in HKSAR: - Step 1: estimate market-implied expected losses daily using CCA (equity/options + balance sheet) for nine local banks (representing 54 percent of sector assets consolidated as of end-June 2013) over Jan 1997–Aug 2013. - Step 2: forecast individual expected losses using dynamic panel-data estimation (monthly observations over at least 10 years) for the five-year horizon. - Step 3: derive joint expected losses as Expected Shortfall/Tail VaR at 95 percent confidence by aggregating time-varying dependence among banks; compare joint expected losses to sum of individual expected losses. - Key preserved statistics: - 95 percent (statistical confidence), 54 percent (share of assets), January 1997 and August 2013 (data window), five-year horizon, daily frequency, monthly observations over at least 10 years, 5 percent (tail-risk threshold for historical loss estimation), CET1 and CAR metrics. ### Key modeling assumptions, parameter calibrations and exact numeric specifications - RWA and regulatory adjustments: - RWAs for derivatives: +30 percent (Severe), +50 percent (More severe), +0 percent (Baseline). - Overall credit RWA under IRB: ~ +25% (Severe), ~ +35% (More severe). - RWA impact of defaulted loans: defaulted exposures’ RWAs approximated as 2.5 times average RWAs for non-defaulted exposures. - LGD stress specification: - LGD(under stress)=0.3502+2.3408*PD (alternate: 0.4022+2.1535*PD if down-cycle LGDs are through-the-cycle). - Funding-cost add-on (generic linear approximation per year per percentage point of capital ratio below Tier 1 hurdle +2.5 pcp buffer): - retail deposits: 5 basis points - interbank deposits (<1 year or without maturity): 20 basis points - short-term wholesale funding: 10 basis points - long-term wholesale funding: 50 basis points, multiplied by share of short-term liabilities - FX shock specification: - Shock per currency = four times maximum deviation of annualized FX volatility during 2008–11 from long-term FX volatility (>10 years); apply 100% in Year 1 and 50% in Year 2 to net open positions. - Example net open FX position shock increases (Year 1 example mapping): Japanese yen increase by 14.6 percent in Y1 and 7.3 percent in Y2 (specific table values preserved in appendices). - Valuation haircuts for sovereign exposures: - Common interest-rate shock in adverse scenarios = 50bps. - Greece exposures haircut = 100 percent. - Haircuts derived from forward CDS density forecasts (75th percentile for adverse scenario), maturity buckets: 1, 3, 5, 7, 10 years; estimation reference date end-June 2013. - Dividend payout rules: - Minimum payout rules: firms not capital-constrained pay out at least 40 percent (or 45 percent in some templates) of earnings after tax; payout ratio declines by 10 percentage points for each 0.5 percentage point decline in capital buffer; 100 percent retention if buffer below threshold. - Appendix XI pay-out ratio table preserves exact values (e.g., FSAP minimum pay-out >2.5 buffer = 40 percent; Basel III maximum pay-out at various CET1 bands specified). - Hurdle rates (Basel III schedule, Table A11.2 exact entries): - Reg. Minimum Total Capital: 8.0, 8.0, 8.625, 9.25, 9.875 (Y1–Y5) - Reg. Minimum Tier 1 Capital*: 5.5, 6.0, 6.625, 7.25, 7.875 (Y1–Y5) - Reg. Minimum Common Equity Tier 1: 4.0, 4.5, 5.125, 5.75, 6.375 (Y1–Y5) - Conservation Buffer (memo): 0.0, 0.0, 0.625, 1.25, 1.875 (Y1–Y5) - Selected descriptive FSIs and sample totals (IMF TD solvency Table 14): - Total assets: 11,699.4 (In billions of HKD) - Total capital: 991.3 (In billions of HKD) - Regulatory capital to RWAs (average): 16.5 - Tier 1 capital to RWAs (average): 13.3 - CET1 capital to RWAs (average): 13.2 - Loan-to-deposit ratio: 97.2 - Return on assets (average): 0.5 - Return on equity (average): 0.2 - Loss rate (average): 1.2 - Non-performing loan ratio (average): 12.6 - Risk-weighting (In percent): 62.2 - Macroeconomic projections (selected exact items from Appendix II Table A3.1): - Real GDP (annual percentage change): 2013 (est.) 2.96; Baseline average annual growth ~4.4 percent; SG and SA series presented year-by-year for 2014–2018 (numerical table preserved in appendices). - Residential house prices (example entries): 2013 (est.) 7.97; Baseline 2014 -1.68; Severe Adverse 2014 -11.97. - Equity market index (Hang Seng): 2013 (est.) 7.00; Baseline 2014 8.06; Slow Growth 2014 -13.95; Severe Adverse 2014 -16.42. ### Policy implications and recommendations (preserved phrasing) - Supervision and stress-testing integration: - HKMA encouraged to continue integration of risk-based supervision in the development of stress test scenarios for macroprudential policy and surveillance. - Further integration of TD and BU exercises (e.g., cross-validation of results) could enhance supervisory work and financial stability analysis. - Horizon and behavioral modeling: - Extending the stress test horizon would place greater emphasis on potential mitigating effects from profitability and behavioral assumptions of banks, allowing a more comprehensive assessment of impact of different risk drivers over time. - Extend single-period shocks to multiple-period scenarios (possibly with feedback effects) to capture secondary impacts from prolonged financial stress, including higher cost of capital and constrained capital mobility. - Ongoing supervisory practices: - Supervisors should routinely conduct stress tests and modify relevant assumptions periodically to support thematic reviews of identified vulnerabilities against emerging risks. - Incorporate more granular prudential information (e.g., intragroup transactions within conglomerates) into stress testing framework where needed. - HKMA to implement LCR minimum criteria (consultation underway) with a view to implementing minimum criteria by January 1, 2015 and continue ELST as routine liquidity monitoring. ### Limitations, caveats and interpretation notes - Tests based on economic and market conditions as of mid-2013 (cut-off date) and do not incorporate subsequent international capital market developments. - Structural changes to business models and mitigating actions announced/implemented after cut-off date were not considered (e.g., organizational restructuring, dynamic RWA management, contingent capital arrangements). - Tests are static and assume simultaneous escalation of liquidity risk across banks; identified vulnerabilities are conditional on specific stress assumptions and should not be interpreted as actual realized needs in a general stress situation. - HKMA TD exercise relies largely on aggregation of single-factor shocks; macro-financial linkages less prominent than in IMF TD approach. - SCCA adjustments: market data re-scaled for listed entities with activities outside HKSAR could not fully address differences in balance sheet strengths outside HKSAR. *Source: _cr14210 - EXECUTIVE SUMMARY; Sections 5, 24, 39; Box 1; Appendices II–XIII (IMF staff and HKMA stress testing material).* ### EXECUTIVE SUMMARY ___________________________________________________________________________ 5 ### EXECUTIVE SUMMARY ### Introduction and Objective - The FSAP stress testing exercise comprises a comprehensive analysis of solvency and liquidity risks of the banking sector, using mid-2013 data. - Purpose: - Determine capacity of the banking sector to absorb realization of key macro-financial risks based on capital adequacy and sufficiency of liquidity under stress. - Examine system-wide resilience over the medium-term, uncover vulnerabilities to rapid macroeconomic deterioration, and identify potential threats to financial stability. - Coverage and scope: - Solvency tests: bottom-up (BU) stress test of selected locally incorporated, licensed banks (“local banks”) and cross-validation by three top-down (TD) tests covering nearly all local banks. - Liquidity tests: various sensitivity analyses based on different TD approaches within existing liquidity reporting framework, using supervisory data and parameters specified by the Hong Kong Monetary Authority (HKMA) and the FSAP team. - Overall, the various stress tests conducted for FSAP cover around 50 to around 70 percent of the banking sector’s total assets―depending on the type of analysis―accounting for up to 78 percent of total deposits and 66 of total loans). - Two TD solvency stress tests cover more than 99 percent of all locally incorporated licensed banks ("local banks"). - Cut-off and data vintage: - Tests are based on economic and market conditions as of end-2013: Q2, the cut-off date of the exercise. ### Solvency Stress Tests: Design and Scenarios - Scenario framework: - Two adverse macroeconomic scenarios and their deviations from the IMF’s World Economic Outlook (April 2013) baseline over a five-year forecast horizon. - Scenarios comprise a short-lived recession scenario and a prolonged slow growth scenario. - Hurdle rates applied according to the Basel III implementation schedule. - Risk channels and components tested: - (i) Impairment charges (credit losses, other losses from held-to-maturity assets) and mark-to-market (MtM) valuation changes of fixed income securities in both the trading and banking book. - (ii) Changes in pre-impairment income, including changes in funding costs. - (iii) Changes in risk-weighted assets (RWAs). - Incorporation of rising risk aversion in capital markets (via market-implied risk measures) and upcoming regulatory reforms (Basel III). - Specific risk factors included: cross-border developments (sovereign risk) and foreign currency risk. - Methodology modalities: - BU: Firms complete their own stress test according to IMF-developed guidelines, in coordination with the HKMA (selected local banks). - TD (enhanced): IMF staff complete market-based stress test with bank-specific macro-financial linkages ('Systemic CCA') for 9 publicly listed, local banks (in Group 1). - TD (IMF/HKMA): IMF staff complete balance-sheet-based stress test with direct prudential data input from sample banks (19 local banks, complete Group 1). - TD (HKMA): HKMA staff complete balance-sheet-based stress test with supervisory data (19 local banks, complete Group 1). - Reporting basis varies: consolidated basis, combined basis, solo basis, with scaling to HK activities where specified. ### Liquidity Stress Tests: Design and Coverage - Focus: - Sudden, sizable withdrawals of funding and sufficiency of existing assets to withstand shocks under stressed conditions. - Tests include assumptions on in- and outflows of existing and contingent assets and liabilities (“funding liquidity risk”) and application of haircuts to assets (“market liquidity risk”). - Tools and standards applied: - HKMA regulatory standards for liquidity. - FSAP-developed liquidity tests. - Revised Basel III liquidity risk framework: Liquidity Coverage Ratio (LCR) and Net Stable Funding Ratio (NSFR) tests. - HKMA staff conducted additional stress test incorporating market and credit risk arising from prolonged negative asset price shocks on cash flow projections. - Coverage: - Liquidity TD stress tests cover both local banks and the largest foreign branches in the sector. - Tests applied include: (i) HKMA 7-day/enhanced liquidity tests, (ii) implied cash flow tests as per IMF guidance, (iii) Basel III standard measures (LCR and NSFR), and (iv) market risk-liquidity interaction. ### Key Findings and Results - Overall resilience: - Stress test results confirm a high degree of resilience of the sector, reflecting strong starting positions of banks. - Banks in HKSAR hold very high levels of capital, are very profitable, and have a low level of asset impairments amid stable funding profiles. - Real estate and mortgage cushioning: - Macroprudential measures adopted by the HKMA, high collateralization of mortgages and declining loan-to-value (LTV) ratios absorb the impact from even severe near-term shocks to property prices. - Solvency outcomes: - Analyses based on prudential data suggest that even a severe economic shock would not result in an aggregate capital shortfall over a five-year forecast horizon. - All larger banks exhibit high levels of capitalization and are able to withstand a severe deterioration of economic conditions. - Some smaller banks may be slightly more vulnerable to economic shocks, greater competitive pressures and rising interest rate risks affecting their solvency conditions and funding costs. - These smaller institutions may experience a significant decline in net operating profitability, which might result in capital constraints over the medium-term under a severe economic shock. - Liquidity outcomes: - Bank liquidity tests focused on sudden withdrawals and application of haircuts indicate the sufficiency of existing assets to withstand shocks under specified assumptions (tests include funding inflows/outflows, haircuts, and supervisory parameters). - Tail risks: - One stress test considered the impact of joint tail risks on system-wide solvency conditions, recognizing that institutional viability might be insufficient to maintain financial stability during extreme stress. ### Policy Implications and Recommendations - Supervision and stress-testing integration: - HKMA is encouraged to continue its integration of risk-based supervision in the development of stress test scenarios for macroprudential policy and surveillance. - Further integration of TD and BU exercises (e.g., cross-validation of results) could enhance supervisory work and financial stability analysis. - Horizon and behavioral modeling: - Extending the stress test horizon would place greater emphasis on potential mitigating effects from profitability and behavioral assumptions of banks, allowing a more comprehensive and potentially more realistic assessment of impact of different risk drivers over time. - Ongoing supervisory practices: - Banking supervisors should routinely conduct stress tests and, from time to time, modify relevant assumptions to support thematic reviews of identified vulnerabilities against emerging risks. *Source: _cr14210 - EXECUTIVE SUMMARY* ### 5. The purpose of the stress test exercise differs from that of supervisory stress testing. ### _cr14210 - 5. The purpose of the stress test exercise differs from that of supervisory stress testing. ### Purpose and scope of the FSAP stress test exercise - The multi-period FSAP stress test exercise is designed and completed for surveillance purposes, with a medium-term focus. - The exercise typically involves very severe stress scenarios to assess the overall resilience of the banking sector. - The results of the stress testing exercise have no immediate supervisory implications but provide input into a broader analysis undertaken by the FSAP, forming the basis for policy discussions with the authorities. - This is different from routine capital reviews undertaken by the authorities (Pillar II), which are aimed at identifying potential capital needs and for which management actions may be required. No management action would be expected as a result of the FSAP stress tests. ### Banking sector structure and key balance-sheet features - The sector comprises 201 institutions―156 licensed banks, 21 restricted license banks, and 24 deposit-taking companies―with assets equivalent to 705 percent of GDP. - Assets are concentrated in four banks, which account for almost half of the consolidated assets of the banking sector. - The sector has been growing rapidly, driven by mortgage-related lending and increasing exposure to non-financial corporates in Mainland China. - Lending composition: - Lending to the corporate sector represents around half of the banking system’s total lending. - Property-lending accounts for about a third. - Funding composition and trends: - Banks are primarily funded by customer deposits, which account for 71 percent of local banks’ total liabilities at end-2012. - Foreign branches: customer deposits represent 31 percent of their total liabilities. - Aggregate loan-to-deposit ratio has increased as some banks issued certificates of deposits in the wholesale market to diversify funding. - Liquidity characteristics: - Liquidity risks for banks are generally low. - Asset encumbrance is relatively low, with total unencumbered liquid assets remaining stable relative to the amount of short-term liabilities. - It is desirable for banks to develop alternative term funding sources at longer maturity tenors to augment the large deposit base, which creates considerable maturity mismatches beyond one year. ### Data coverage and methodology - Stress tests used mid-2013 financial data of key institutions and macroeconomic projections and market information available at that time. - Coverage: Up to 19 local banks and eight foreign branches were included in the stress tests. - Granular supervisory information and banks’ internal data were used in addition to publicly available information. - Objective: assess resilience to solvency and funding shocks under different macroeconomic scenarios, including unemployment rise, sharp depreciation of real estate prices, declining lending profitability, rising sovereign risk, and upcoming regulatory reforms. - Solvency tests: - Based on three scenarios (baseline and two adverse scenarios) determined in collaboration with HKMA. - Hurdle rates are applied according to the Basel III implementation schedule. - Cross-border effects: considered in all macroeconomic scenarios; assumptions aligned to allow time-varying patterns of selected macro-financial variables consistent with HKSAR forecasts. - Liquidity tests: - Focus on sudden, sizeable withdrawal of funding (liabilities run-off) and sufficiency of liquidity buffers. - Applied tests: - HKMA implied cash flow (ICF) tests (stress horizons of one week and three months). - FSAP team liquidity tests (stress horizons of one week and one month). - Basel III standard liquidity measures: LCR and NSFR. - A supplementary HKMA staff stress test incorporating market and credit risk from prolonged negative asset price shocks on cash flow projections. - Capital adequacy assessed in accordance with Basel III standard; hurdle rates for total capital, Tier 1, and CET1 follow internationally agreed Basel III schedule. - The capital conservation buffer will come into full effect in 2015 and is applied in the last three years of the five-year forecast horizon. ### Stress-testing framework and approaches - Three-pronged solvency testing approach: - BU balance sheet stress test conducted by banks in collaboration with FSAP and HKMA staff (BU exercise). - Cross-validation by two TD balance sheet stress tests based on FSAP assumptions (IMF TD exercise) and a modified implementation of HKMA supervisory stress test (HKMA TD exercise). - Cross-validation by a TD market-based stress test applying the “Systemic Contingent Claims Approach, SCCA” (IMF SCCA) using multivariate extreme value theory to generate an endogenous measure of aggregate capital adequacy. - Inputs and assumptions: - Projected variables include real GDP, household income, unemployment, inflation, interest rates (3-/12-month HIBOR, term deposit rate, and “best lending rate”), asset swap rates, equity prices, commercial property price index, real estate price index, and credit growth. - Prescriptive assumptions cover loss rates, profitability, fixed income holdings, exchange rates, taxes, valuation haircuts on sovereign exposures, funding costs, behavioral proxies (dividend payout, credit growth, deleveraging), and regulatory changes (capital requirements, RWAs, and definition of capital). - Structural changes to business models and mitigating factors requiring managerial intervention or arising after the cut-off date were not considered. ### Macroeconomic scenarios and calibrated shocks - Baseline scenario: macroeconomic projections based on the World Economic Outlook of April 2013 and the Article IV staff report for Mainland China in June 2013. - Slow Growth (SG) scenario: - Underpinned by a broad-based slowdown of global growth, including Mainland China, triggered by an increase in cost of capital due to markets’ accelerated view on pace of U.S. monetary policy tightening. - Cumulative negative deviation of about 9.1 percentage points in real GDP growth over a five-year period. - Severe Adverse (SA) scenario: - SG scenario aggravated by intensified capital outflows affecting HKSAR as an international financial center. - Shock of two standard deviations of the long-term (30-year average) two-year cumulative growth rate from the IMF-projected baseline. - Cumulative negative deviation of about 12.1 percentage points in real GDP growth over a five-year horizon. - Asset-price and growth impacts under adverse scenarios: - Property prices decline by 30 and 40 percent respectively (under the two adverse scenarios). - Equity prices decline by 50 and 65 percent respectively. - Economic growth deteriorates by 1.8 and 2.4 percentage points (to 2.6 and 2.0 percent, respectively) on average relative to baseline expectations of average annual growth of 4.4 percent. - HKMA TD exercise maps to “severe” and “more severe” scenarios over a shorter two-year period with comparable magnitudes condensed into two years (e.g., property prices fall by 30 percent and 50 percent under “severe” and “more severe” respectively). ### Key findings from solvency and liquidity stress tests - Aggregate resilience: - Stress test results confirm that the banking sector would remain sufficiently capitalized and liquid under the current regime. - Analyses based on prudential data suggest that even a severe economic shock relative to the baseline would not result in an aggregate capital shortfall over a five-year forecast horizon. - Heterogeneity across banks: - All larger banks exhibit high levels of capitalization and can withstand a severe deterioration of macroeconomic conditions. - Some smaller banks might be more vulnerable to economic shocks, greater competitive pressures in a saturated lending market, regulatory changes, and rising interest rate risks affecting solvency and funding costs. - Smaller institutions may experience a significant decline in net profitability, potentially resulting in capital pressures over the medium-term under a severe economic shock. - Liquidity position: - Liquidity test results show banks exceed—and in most cases by a large margin—minimum liquidity ratios and threshold requirements of various stress test metrics due to high cash balances and large stocks of liquid assets at low encumbrance levels. - Starting conditions matter: - Outcome reflects strong solvency at the starting point: banks hold very high levels of capital, are highly profitable, and have a low level of asset impairments amid stable funding profiles. - Surge of property-related lending has not resulted in higher leverage or a rise in RWAs. - Macroprudential measures by HKMA (high mortgage collateralization and declining LTV ratios) can absorb impact from even severe near-term shocks to property prices in a mortgage-dominated sector. ### Limitations and caveats - Stress tests were based on economic and market conditions as of mid-2013 (cut-off date) and did not take into account developments in international capital markets during the completion of the exercise. - Structural changes to business models and potential mitigating actions not announced/implemented by the cut-off date were not considered (e.g., organizational restructuring, changes in business lines, dynamic RWA management, contingent capital arrangements). - The HKMA TD exercise relies largely on aggregation of single-factor shocks, so macro-financial linkages may be less prominent than in IMF TD approach. *Source: PEOPLE’S REPUBLIC OF CHINA––HONG KONG SPECIAL ADMINISTRATIVE REGION, INTERNATIONAL MONETARY FUND (content unit: _cr14210 - 5. The purpose of the stress test exercise differs from that of supervisory stress testing.)* ### 24. The results from both BU and TD exercises show a significant drop in banks’ capital ### _cr14210 - 24. The results from both BU and TD exercises show a significant drop in banks’ capital ### Solvency stress-test headline findings - Both Bottom-Up (BU) and Top-Down (TD) exercises show a significant drop in banks’ capital adequacy ratios under severe stress, but the sector remains sufficiently capitalized under Basel III. - All capital adequacy measures (capital adequacy ratio (CAR), Tier 1 capital and CET1) are materially affected by adverse scenarios, but there is no aggregate capital shortfall across the sector under the balance-sheet approaches. - Under baseline conditions: - Aggregate capitalization increases moderately by up to one percentage point of Tier 1 capital (and less so for CET1 capital) as all banks record net operating profits in the BU exercise. - Under TD balance-sheet approaches: - Aggregate CET1 ratio declines at most by 2.7 percentage points and 0.9 percentage points under the two adverse scenarios (HKMA and IMF TD approaches, respectively). - CAR contracts by up to 3.7 percentage points under the SA scenario across all balance-sheet approaches (TD and BU). - Under the IMF SCCA (market-consistent, joint tail risks at a 1-in-200 year probability) results: - Aggregate CAR would decrease by 4.5 percentage points. - Aggregate CET1 ratio would decrease by 4.3 percentage points. - Under the SCCA conditional tail expectation (CTE/expected shortfall) at a statistical confidence level of 95 percent or higher, the sector remains solvent and stays above all hurdle rates over the entire stress horizon. ### Bank-level vulnerabilities and capital shortfalls - Most large banks exhibit very solid capital buffers; some smaller banks are more vulnerable during the final years of the SA scenario. - Identified potential capital shortfall (one TD approach): - HKD 1.3 billion of Tier 1 capital (the only hurdle rate breached by a bank until the end of the forecast horizon). - This represents about 0.2 percent of Tier 1 capital in the sector as of end-June 2013. - In two out of three TD approaches, one to two small banks would fall below the hurdle rate toward the end of the forecast horizon; none would fail in two TD approaches at the same time. - Market-implied capitalization of some banks declines rapidly under IMF SCCA—suggesting potential need for timely build-up of additional capital buffers over the medium term if the stress scenario materializes. - HKMA has engaged in in-depth discussions with local banks on capital planning for Basel III and setting higher “supervisory trigger” capital add-ons to Pillar II progressively during Basel III implementation. ### BU stress-test specifics and results - BU stress tests were administered jointly with the HKMA; banks used their internal models with FSAP-provided detailed guidelines on assumptions and methodologies. - Each bank submitted a “report card” to the HKMA; aggregated results were provided to the FSAP team to estimate: - Changes in potential losses from asset impairments, - Profitability, - Regulatory impact of Basel III on the definition of capital, - Post-shock RWAs and capital needs (where applicable). - BU exercise findings: - All banks pass the capital hurdle rates under all scenarios. - CET1 ratios diverge by as much as 2.3 percentage points from their pre-stress capitalization. - The maximum decline of aggregate capitalization amounts to 3.3 percentage points of CAR. - The SA scenario is more stringent than the SG scenario; both scenarios’ impacts were milder than initially anticipated, partly due to the interest rate scenario design and mild compression of term spreads given the prevalence of floating rate lending and a large share of term deposit funding. ### TD stress-test framework, SCCA and reconciliation with BU - TD balance-sheet-based framework: - Used financial data as of end-June 2013 provided by all 19 local banks. - Satellite models estimated changes in profitability and credit losses using two-stage least squares panel regressions over quarterly observations from 1996:Q1 to 2013:Q2 (orthogonal deviations transformation; up to five quarter lags of endogenous variables and up to two quarter lags of other variables as instruments). - Modeled profitability components (interest income, interest expenses, fee/commissions income, operating expenses), flow of asset impairments, and stock of NPLs. - IMF SCCA (market-based) framework: - Accounts for dependence among individual banks to estimate joint market-implied expected losses (individual expected losses aggregated using SCCA with a five-year sliding window and monthly updates). - Treats banking sector as a portfolio of implicit put options (individual expected losses) with non-linear dependence structure; links macroeconomic paths to individual and joint expected losses via changes in banks’ market-implied expected losses (monthly implicit put option values). - TD versus BU reconciliation: - TD balance-sheet results are broadly consistent with aggregated BU findings; trends for CET1, Tier 1, and total capital ratios are similar across approaches for baseline and adverse scenarios. - Differences arise from model design and scope: - BU uses firms’ internal models and more granular data for selected local banks (consolidated level). - TD analyses cover a larger sample (HKMA TD on combined basis, IMF TD on consolidated basis) and apply uniform sensitivities implied by panel estimates, producing less diverse loss results across banks. - BU results generally show greater sensitivity to scenario variation (larger increases/decreases in capital ratios in baseline/adverse scenarios) with differences diminishing in later years. ### Risk drivers affecting solvency - Principal risk drivers with strong impact on solvency under stress: - Diminished earnings capacity if economic conditions deteriorate. - Increases in unexpected losses implied by higher RWAs. - Increasing provisions for credit risk and higher loan impairments. - Procyclical impact on credit RWAs. - Impact of regulatory changes to the definition of capital (Basel III). - Current mitigants: - Extremely robust credit conditions and low leverage in the sector. - Notable stress outcomes: - NPL balances in the real estate sector rise considerably in relative terms under stress. - Rising competition could further depress interest income and limit capital buffer build-up after full adoption of new capital requirements. ### Liquidity stress-test design and scope - Liquidity stress tests were carried out by HKMA staff in consultation with the FSAP team using end-June 2013 data. - Objective: assess resilience to sudden, sizable funding withdrawals; tests informative under very severe system-wide distress. - Channels modeled for funding shortfalls: - Scheduled and unscheduled cash outflows. - Cash inflows from maturing assets and assets repo-able or saleable at stressed market values (“market liquidity risk”). - Restricted access to funding markets (“funding liquidity risk”). - Rollover risk from maturity mismatches. - Assumptions influencing severity: declines in asset values, amortization/renewal rates, callback rates on contingent claims, haircuts on assets used as collateral, and swap arrangements. - Three liquidity regimes/tests examined: - HKMA supervisory liquidity stress testing framework: 7-day test/LOLR analysis and Enhanced Liquidity Stress Test (ELST) spanning one to three months. - FSAP team’s ICF tests (IMF cumulative five-day test and non-cumulative 30-day test): simulate gradual outflow over five consecutive days (cumulative) and over 30 days (non-cumulative); underlying assumptions include withdrawal of more than one quarter of unsecured funding within five days and complete run-off of secured funding within one month in the extreme case. - Basel III measures: LCR and NSFR. - Assessment metric: - Ratio of cash inflows (including proceeds from securities lending, repos and asset sales) to cash outflows (stress test ratio). - Stress test ratio > 100 percent implies a liquidity surplus under the stress scenario. - Coverage: - HKMA and IMF liquidity stress tests applied to all 19 banks in Group 1 (local banks) and all eight banks in Group 2 (foreign branches) on a solo basis, representing 71 percent of the banking sector's total assets. *Source: _cr14210 - 24. The results from both BU and TD exercises show a significant drop in banks’ capital (IMF).* ### 39. The Basel III liquidity framework is based on two quantitative liquidity standards that ### _cr14210 - 39. The Basel III liquidity framework is based on two quantitative liquidity standards that ### Basel III liquidity framework: LCR and NSFR - The framework is based on two standardized ratios applied to a selection of local banks on both consolidated and combined bases and foreign branches on a solo basis. - LCR—This ratio: - Promotes short-term resilience to potential liquidity disruptions by requiring banks to hold sufficient high-quality liquid assets (HQLA) to withstand the run-off of liabilities over a stressed 30-day scenario specified by supervisors. - Requires banks to hold a sufficient stock of unencumbered, HQLA to cover cash outflows less cash inflows (subject to a cap at 75 percent of total cash inflows) that are expected to occur in times of stress. - In January 2013, the Basel Committee reached agreement on a composition of HQLA and parameters for net cash outflows resulting from deposits and contingent liabilities, and a transition period for introduction of LCR. - An LCR of less than 100 percent indicates a liquidity shortfall. - NSFR—This structural ratio: - A final agreement had not yet been reached by the Basel Committee (BCBS, 2014) at the time of the report. - Based on existing proposals, would require banks to establish a funding profile expected to be stable over a one-year horizon under an extended stress scenario. - Would reflect the proportion of long-term assets funded by stable sources (customer deposits, long-term wholesale funding with maturities of more than one year, and equity; excludes short-term funding). - A value of this ratio of less than 100 percent indicates a shortfall in stable funding. - Basel Committee proposed revisions (January 2014) to NSFR sought to: - Reduce “cliff effects” by introducing a new category of assets and liabilities with remaining maturities between six months and one year. - Align required stable funding weights to the LCR’s definition of HQLAs. - Alter calibration to focus more on shorter-term, potentially more volatile funding sources—revisions align NSFR closer to LCR and expected to be more accommodative to banks’ business models. ### Market risk–contingent liquidity approach (Wong and Hui, 2009) — methodology - Applied to 17 local banks on a consolidated basis (including overseas branches and subsidiaries). - Framework features: - Captures how MtM losses on banks’ holdings of risky assets due to prolonged negative asset price shocks increase solvency risk and reduce ability to generate liquidity from asset sales. - Banks’ default risk is endogenously determined using a Merton-type firm value model, which determines potential retail and wholesale deposit outflows. - With estimated deposit outflows and contingent drawdowns, daily cash outflows of individual banks are derived to assess cash shortage and default risk for given paths of asset price shocks. - Employs a Monte Carlo simulation to generate random asset price changes for major asset classes on banks’ balance sheets. - After 1,000 iterations, the probability of cash shortage and probability of default due to liquidity problem are estimated based on the difference between cumulative inflows and outflows over a one-year horizon. - Severity of shocks used (comparable to HKMA TD “severe” scenario): - At the 90th percentile, credit spreads rise by around 0.4, 1.7 and 1.74 percentage points for corporate bonds rated “AA” or higher, “A” and “BBB”, respectively. - Credit spreads of non-investment grade (and unrated) corporate bonds increase by 10.5 percentage points. - Equity prices decline by 27 percent. - Market value of structured financial securities decline by 32 percent. - Maximum monthly run-off rate for retail deposits set at 42 percent; for wholesale deposit is 100 percent. - 15 percent of committed credit lines assumed to grant to SIVs; drawdown rate assumed negatively correlated with price of structured financial securities. - No risk mitigation measures by parent banks and the central bank assumed in the stress horizon. ### Liquidity stress test results — key findings and statistics - General finding: - Liquidity risks for banks are low; banks exceed—and in most cases by a large margin—the minimum acceptable ratios due to high cash balances and large stocks of liquid assets at low encumbrance levels. - Results are robust to variation in type and magnitude of funding shocks. - Short-lived shock results (HKMA and IMF measures): - Average stress test ratio stood at 279 percent under HKMA’s 7-day/LOLR analysis. - Average stress test ratio stood at 338 percent under the IMF’s five-day ICF test. - Larger banks have better liquidity positions. - One-month and longer horizons: - Average liquidity ratio of 162 percent under HKMA’s ELST. - Average liquidity ratio of 235 percent under the IMF’s 30-day ICF test. - Differences reflect higher (lower) run-off (callback) rates in HKMA’s tests, applied to entire deposit base, whereas IMF tests limit run-off to open maturities and maturity terms not exceeding stress period. - Basel III standard measures (limited sample, end-June 2013 positions): - On both consolidated and combined bases, the average LCR and NSFR are above the 100 percent threshold. - Most institutions clear the 100 percent threshold of each indicator with ease, especially on a consolidated basis. - On a combined basis, a few banks show potentially insufficient liquidity buffers to meet negative net cash outflows over a 30-day horizon according to LCR. - Nonetheless, most of them report LCR ratios above 60 percent, which is the initial minimum requirement set out in the first year (2015) of the transition schedule for adoption of Basel III. - Many of these banks already report LCR ratios close to 100 percent. - Most sample banks report NSFR values of 100 percent or higher; a couple of banks show NSFR ratios lower than 100 percent, but the shortfall was not material. - Wong and Hui (2009) one-year horizon results: - No bank triggers cash shortage or default due to liquidity problems over a one-year stress horizon. - Banks’ default risk remains at a low level under the specified severe asset price shocks. - Cash inflows sufficient to cover cash outflows for all banks in the sample. - Funding structure and drivers: - Banks generally not reliant on wholesale funding, especially term funding beyond one year. - Contractual maturity mismatches increase significantly beyond one year due to long-tenor residential mortgage portfolios and large retail deposit base. - Supplementary longer-term sources of funding could reduce rising cash flow mismatches at maturities of one year and longer. - Foreign branches vs local banks: - Foreign branches exposed to different risk drivers than local banks due to liability-side balance sheet differences. - Main risk driver for local banks: shrinking deposit base; liquidity positions improve substantially if customer deposits (especially retail) remain stable. - Foreign branches more dependent on wholesale funding and show greater susceptibility to outflows from related party lending and contingent claims. - In HKMA ELST, around 80 percent of outflows from local banks caused by withdrawals of deposits; outflows from foreign branches more driven by a run on wholesale funding (around 70 percent of total outflows). - Intragroup claims are an important distinguishing feature in projected cash outflows for foreign branches, accompanied by considerable decline in wholesale funding. ### Comparisons and interpretation of stress tests - HKMA vs IMF liquidity stress tests: - HKMA liquidity stress test is generally less restrictive regarding definition of liquidity buffer (evaluation of HQLA) but applies stricter assumptions on stress scenario and contingent cash outflows. - HKMA’s ELST and the IMF’s 30-day test allow a wider range of assets that might not qualify as HQLA in LCR calculation. - Caveats on stress test interpretation: - Tests are static and assume all banks face escalating liquidity risk simultaneously. - Identified liquidity risk should be interpreted as vulnerability to the specific assumptions used, not necessarily representative of an actual liquidity need in a general stress situation. - Results should be qualified by mitigating considerations, such as likely reallocation of deposits within the banking sector when not all banks experience funding shocks simultaneously, and the assumption that (at least retail) deposits largely remain in the banking system. ### Policy implications and recommendations - HKMA supervisory and monitoring actions: - HKMA has introduced ELST as essential routine liquidity monitoring, including more severe scenarios compared to previous 7-day test. - HKMA undergoing consultation with industry on LCR details with a view to implementing minimum criteria by January 1, 2015. - HKMA introduced a matched term funding requirement in the second half of 2011 for banks experiencing high credit growth; introduced a stable funding requirement in late 2013 to complement matched term funding. - Recommended supervisory and technical enhancements: - Continue integration of risk-based supervision in development of stress test scenarios for macroprudential policy and surveillance. - Incorporate more granular prudential information (e.g., intragroup transactions within conglomerates) into stress testing framework where needed. - Further integrate TD and BU stress tests (e.g., periodic cross-validation of results) to enhance supervisory work and financial stability analysis. - Extend single-period shocks to multiple-period scenarios (possibly with assessment of feedback effects) to: - Allow longer forecast horizons and incorporate macro-financial transmission channels. - Emphasize mitigating effects from profitability and behavioral assumptions over medium and long term. - Capture secondary impacts from prolonged financial stress, including higher cost of capital and constrained capital mobility. *Source: IMF staff analysis in _cr14210 (Basel III liquidity framework and liquidity stress testing results and implications).* ### Box 1. Overview of the Systemic Contingent Claims Approach Framework for Stress Testing ### Box 1. Overview of the Systemic Contingent Claims Approach Framework for Stress Testing and the Implementation in the Context of Hong Kong SAR ### Overview of the Systemic Contingent Claims Approach (SCCA) - SCCA models systemic solvency risk by combining the multivariate extension to contingent claims analysis (CCA) with the concept of EVT to generate a system-wide tail risk (Jobst and Gray, 2013). - The magnitude of risk from multiple firms falling into distress is modeled as a portfolio of individual expected losses calculated from equity market and balance sheet information using an enhanced form of CCA. - CCA is applied to construct risk-adjusted (economic) balance sheets and estimate market-implied expected losses for financial institutions. - Firm-specific distributions of expected losses and their dependence are combined to generate a multivariate distribution that captures extreme joint expected losses. ### Core conceptual elements of CCA within SCCA - CCA treats equity owners in leveraged firms as holders of a call option on firm value after liabilities are paid; bondholders effectively write a European put option to equity owners. - Market-implied potential losses associated with outstanding liabilities are valued as an implicit put option (credit spread above the risk-free rate). - The put option value depends on the duration of total debt, leverage, and volatility of asset value. - The put option is modeled based on a jump diffusion process to address empirical shortcomings of the Merton (1973 and 1974) model. - CCA quantifies default risk by comparing stochastic asset values to a default barrier; higher asset uncertainty relative to the barrier raises default risk. ### Two-step SCCA estimation framework - Step 1: Estimate market-implied expected losses (and associated change in capital) for each sample bank for daily observations of market prices and quarterly balance sheet information over a pre-defined estimation period using advanced CCA. - Step 2: Forecast individual expected losses over the stress horizon using historical sensitivities to firm-specific and macro-financial condition changes; aggregate individual forecasts in a multivariate setup accounting for time-varying dependence to derive joint expected losses. - Reported capital results show changes in CAR and CET1 ratios over the stress horizon for: - individual expected losses, - the sum of individual expected losses (ignoring dependence), - the joint expected losses expressed as conditional tail expectation at 95 percent statistical confidence (Expected Shortfall / Tail VaR). ### Implementation steps in the Hong Kong SAR context (three-step estimation process) - Calculation of market-implied expected losses: - Historical market-implied expected losses estimated at daily frequency using CCA based on equity/equity options data and balance sheet information of nine local banks. - The nine local banks represent 54 percent of total banking sector assets on a consolidated reporting basis as of end-June 2013. - Data between January 1997 and August 2013 are used to estimate the central case (median) market-implied expected losses and losses during extreme market stresses at a statistical probability of 5 percent or less (expressed as “tail risk”). - Specification of macro-financial linkages and estimation of individual expected losses: - Forecast series of expected losses for each sample bank are generated for all three scenarios over a five-year horizon. - Dynamics estimated from a dynamic panel-data estimation (Arellano and Bover, 1995) using monthly observations over at least 10 years. - Individual expected losses applied to capitalization at the starting point (at end-June 2013) to determine potential individual capital shortfall over the stress horizon. - Estimation of joint expected losses and tail risk: - Forecast series of joint expected losses derived as Expected Shortfall/Tail VaR at 95 percent statistical confidence by considering the time-varying dependence structure of forecasted expected losses. - Univariate marginal density functions are assumed to converge to a Generalized Extreme Value (GEV) distribution (as defined in Jobst and Gray, 2013). - Joint expected losses’ impact on aggregate capitalization (CAR and CET1) of all nine sample banks is compared to the impact of the sum of individual expected losses. ### Aggregation methodology and capital measures - Market-implied capital results are calculated by subtracting individual and joint expected losses from individual and aggregate capitalization (CET1 and total regulatory capital) in each year of the forecast horizon, aligned with minimum requirements under Basel III. - The aggregation derives point estimates of market-implied joint expected losses from the multivariate density of each bank’s marginal distribution of market-implied expected losses and their dependence structure. ### Comparative and methodological notes - EVT is used to study tail behavior of heavily skewed data and specify residual risk at high percentile levels through generalized parametric estimation of order statistics. - CCA is a generalization of option pricing theory (Black and Scholes (1973); Merton (1973 and 1974)) and is based on: - (i) values of liabilities derived from assets; - (ii) assets follow a stochastic process; - (iii) liabilities have different priorities (senior and junior claims). - The Merton model has been shown to consistently underpredict spreads. - The SCCA time-varying multivariate distribution approach to point estimates of systemic risk is described as more comprehensive than CoVaR and Marginal Expected Shortfall (MES) expositions. - For sample banks with listed entities including activities outside HKSAR, market data inputs to CCA were re-scaled: implied assets, asset volatility, and default barrier were adjusted to reflect balance sheet size of local entities in HKSAR; adjustments could not fully address potential differences in balance sheet strengths and performance between local entities in Hong Kong and those elsewhere. ### Key statistics and parameters preserved from the source - 95 percent (statistical confidence for Expected Shortfall / Tail VaR) - 54 percent (share of total banking sector assets represented by nine local banks, consolidated, as of end-June 2013) - January 1997 and August 2013 (data window for historical market-implied expected losses) - five-year horizon (forecast horizon for scenario projections) - daily frequency (estimation frequency for historical market-implied expected losses) - monthly observations over at least 10 years (data for dynamic panel estimation) - 5 percent (statistical probability threshold for tail risk definition used in historical loss estimation) - CET1 and CAR (capital metrics used for impact assessment) *Source: _cr14210 - Box 1. Overview of the Systemic Contingent Claims Approach Framework for Stress Testing* ### 5. Regulatory and ### 5. Regulatory and Market-based Standards and Parameters ### Regulatory standards - n.a. (but cash flow monitoring part of the Basel III liquidity risk framework) - Basel III ratios: LCR and NSFR. ### Reporting Format for Results / Output presentation - Hurdle metrics: distribution of ratios, number of failed banks (i.e., ratio < 100%), liquidity shortfall relative to unencumbered assets. - Hurdle metrics: distribution of ratios, number of failed banks (i.e., ratio < 100%), liquidity shortfall relative to unencumbered assets. ### Stress Test Matrix (STeM) — Banking Sector: Solvency — Institutional perimeter and data - Institutions included: - Selected local banks. - All large locally incorporated licensed banks (16), excluding those which are subsidiaries of another licensed bank. - All large locally incorporated licensed banks (19). - 9 publicly listed, locally incorporated licensed banks. - Market share: - About 50 percent of total banking sector assets on a consolidated basis (as of end-June 2013). - 63.0 percent of total banking sector assets on a consolidated basis (as of end-June 2013). - 58.0 percent of total banking sector assets on a combined basis (as of end-June 2013). - 53.8 percent of total banking sector assets on a consolidated basis (as of end-June 2013). - Data and baseline date: - Source: institutions’ own granular data. - Date: end-June 2013 (projected to end-2013). - Scope: consolidated banking group. - Source: supervisory data. - Scope: combined basis. - Source: institutions’ own granular data and capital market data. - Scope: consolidated banking group. ### Channels of risk propagation — Measurement approaches - Banks’ internal models. - BU guidance (IMF, 2011a, 2012a and 2013b). - Valuation haircut model for sovereign risk (Jobst and others, forthcoming; IMF, 2013b). - Balance sheet-based model (IMF, 2010, 2011 and 2012). - Balance sheet-based model (HKMA, 2013). - Systemic CCA model (Gray and Jobst, 2010; Jobst and Gray, 2013). - Valuation haircut model for sovereign risk (Jobst and others, forthcoming; IMF, 2013b). ### Satellite models for macro-financial linkages - Bottom-up by banks: - Macro-financial linkages estimated based on firm’s internal models to forecast the profitability components (interest income, interest expenses, fee/commissions income, trading income and operating expenses) as well as the flow of asset impairments and charge-offs; each material loan category includes an increase of loss-given-defaults (LGDs) under stress according to the increase of default risk (PD), after controlling down-cycle LGDs that are based on a long-term average, i.e., “through the cycle.” - Top-down by FSAP Team (balance sheet-based): - Macro-financial linkages estimated based on dynamic 2SLS panel data regression (using GMM with orthogonal deviations over quarterly observations between 1996: Q1 and 2013: Q2) to forecast the profitability components (interest income, interest expenses, fee/commissions income, and operating expenses) as well as the flow of asset impairments and charge-offs; each material loan category includes an increase of LGDs under stress according to the increase of default risk (PD), after controlling down-cycle LGDs that are based on a long-term average, i.e., “through the cycle”; the change in trading income was mapped to nominal GDP growth. - Top-down by HKMA (balance sheet-based): - Macro-financial linkages estimated for asset impairments/charge-offs for certain loan categories (“Nonbank Mainland China Exposures (NBMCE)”, other loans (excl. residential mortgages, property investment loans, and credit card lending), and off-balance sheet credit exposures derived from their historical sensitivity to changes in real GDP at specified statistical confidence level; other shocks to default probabilities (for stock-related lending, credit card lending, off-balance sheet derivatives exposures, debt securities) and asset prices by means of statistical mapping to specified historical severity (90th or 99th percentile); expert judgment on other risk factors (interest rate shocks, floors on asset impairments/charge-offs, procyclical impact on credit RWA). - Top-down by FSAP Team (market-based, ‘Systemic CCA’): - Macro-financial linkages estimated based on dynamic 2SLS panel data regression (→ other TD test by FSAP team). - Key macroeconomic and financial variables include GDP (nominal and real), household income, unemployment, inflation, interest rates/asset swap rates (short-term and long-term), equity prices, commercial property price index, and real estate price index, and were projected using the HKMA’s macro model and IMF staff estimates based on G20MOD model. - Cross-border effects are not explicitly considered in any macro scenario but are implicit in nominal GDP growth. - Estimated dependence between market-implied expected losses and their impact on capital adequacy. ### Macroeconomic variables and cross-border considerations - Key macroeconomic and financial variables include GDP (nominal and real), household income, unemployment, inflation, interest rates/asset swap rates (short-term and long-term), equity prices, commercial property price index, real estate price index, and credit growth, and were projected using the HKMA’s macro model and IMF staff estimates based on G20MOD model. - Cross-border effects are considered in all macro scenarios: IMF staff provided estimates for real GDP growth rates consistent with the macroeconomic forecast for HKSAR under both baseline and adverse scenarios for all relevant countries (Mainland China, Australia, France, Germany, India, Japan, South Korea, Singapore, the United Kingdom, and the United States). - Cross-border effects considered in relation to Mainland China: real GDP growth rate for Mainland China is incorporated in projections of loan losses. - Sovereign risk: - Assessed by applying valuation haircuts on all direct and indirect net exposures to sovereign risk (including home country) and financial bonds (i.e., bonded debt issued by financial institutions) in trading book as well as available-for-sale (AfS) and hold-to-maturity (HtM) assets over the entire time horizon; cash at central banks as well as repos or asset swaps where there is no economic interest in the security (for instance, instruments held against assets pledged to the HKMA) are excluded. - For HKMA top-down: assumption of exposure-weighted average duration of three years. ### Stress test horizon - 2014–2018 (five years). - 2014–2018 (five years). - 2014–2015 (two years) but single-period set-up. - 2014–2018 (five years). ### Tail shocks — Scenario analysis - Three different macroeconomic scenarios: - (i) a baseline scenario with projections in line with the World Economic Outlook (April 2013); - (ii) a “slow growth” scenario, underpinned by a broad-based slowdown of global growth triggered by markets’ accelerated view on the pace of tightening in U.S. monetary policy; - (iii) a “severe adverse” scenario, where the contraction of economic activity (ii) is further aggravated by a severe intensification of capital outflows in emerging market countries, impacting Hong Kong’s major trading partners. - Under the two adverse scenarios, economic growth in HKSAR deteriorates to 2.6 and 2.0 percent on average relative to baseline expectations of average annual growth of 4.4 percent (which is slightly above the long-term (30-year) annual growth rate of 4.1 percent). - The magnitude of these shocks—at cumulative negative deviations of about 9.1 and 12.1 percentage points in real GDP, which equate to about 1.6 and 2.1 standard deviations of the average cumulative two-year growth rate over the last 30 years, respectively—is in line with the spectrum of economic adversity considered in the context of other stress testing exercises. - Top-down by FSAP Team (market-based, ‘Systemic CCA’) scenarios: - Scenario A (“Severe”) and Scenario B (“More Severe”), which include single factor shocks to credit losses, market risk (interest rates, equity prices), profitability (net interest income and non-interest income), asset risk-weighting, and interest in land & buildings at the 90 percent (Scenario A) and 99 percent (Scenario B) statistical confidence levels (Table 6). ### Sensitivity analysis - FX shock: - Shock to FX net open positions (both through the P&L and on RWAs): Chinese renminbi, euro, Japanese yen, pound sterling, Singapore dollar, and other material currencies for the firm vis-à-vis the U.S. dollar; - For each currency should be four times the maximum deviation of the annualized FX volatility during the 2008–11 period from the long-term FX volatility (>10 years) for the adverse scenarios (not baseline) and impact the trading book in 2014 (100 percent) and 2015 (50 percent) only. - RWAs increase not only for derivatives but all credit risk-sensitive exposures. - Variability of statistical confidence level of risk measure. - Quantification of contribution of individual banks to systemic risk. ### Risk factors and balance sheet assumptions - Risks/factors assessed: - Credit risk (households and corporates, domestic and foreign exposures, local currency and FX). - Counterparty risk of off-balance sheet exposures in the banking book. - Sovereign risk of all government bonds, indirect sovereign exposure as well as all financial bonds. - Funding risk (additional add-on to interest expenses, contingent on Tier 1 capitalization). - Market risk: interest rates and FX. - Tax rate: 16.5 percent. - Credit risk (loans and advances (incl. NBMCE, residential mortgages and property investment loans, credit card lending, stock-related lending, other loans, off-balance sheet exposures) and debt securities (non-securitization/securitization)). - Market risk: interest rates and equity prices. - Balance sheet assumption: - Static balance sheet with constant credit growth (i.e., lending and funding increases in line with nominal GDP (if positive)) without changes in the funding structure, subject to a “deleveraging rule” (i.e., credit growth decreases by 2 percentage points for each decrease in Tier 1 capital by 1 percentage point once the buffer is less than 2.5 percentage points; no asset disposals/divestments after cut-off date; defaulted loans are not replenished). - Static balance sheet without any adjustments. ### Treatment of dividends - Dividend payout is limited if the firm reports profits over the past year (and exhibits sufficient Tier 1 capitalization) but falls below 10.5 percent (which reflects the magnitude of the CAR and “capital conservation buffer” under Basel III); however, firms that are not capital constrained will have to pay out at least 40 percent of earnings after tax each year. - Dividend payout is limited if the firm reports profits over the past year (and exhibits sufficient Tier 1 capitalization) but CAR falls below a certain threshold (which reflects the average Pillar 2 minimum CAR requirement imposed by the HKMA on the local banks under Basel III and the historical capital buffer of local banks over that requirement (3.2 percent)). Firms that are not capital constrained will have to pay out at least 45 percent of earnings after tax each year. The payout ratio declines by 10 percentage points for each 0.5 percentage point decline in the capital buffer; 100 percent retention if capital buffer is lower than the threshold. - Dividend payout: discounted from the current share price when deriving implied asset values. *Source: _cr14210 - 5. Regulatory and (PDF chapter).* ### 5. Regulatory and Market-based Standards and Parameters ### 5. Regulatory and Market-based Standards and Parameters ### Calibration of risk parameters - PDs and LGDs: internal models for point-in time-PDs and down-cycle LGDs (or adjustment to long-term average LGDs); in absence of an internal model, LGDs - PDs and LGDs: PDs (not LGDs) change over time; through-the-cycle PDs based on prudential data and stressed PDs based on the change in the stock of NPLs (via - Credit loss rates, PDs and LGDs: change in response to shock (either based on estimated sensitivity to macro-financial variables (in the case of NBMCE, - Market-implied expected losses (with endogenously determined PD and LGD) inform the amount of capital shortfall (if any) - LGD stress specification: LGD(under stress)=0.3502+2.3408*PD. In case of over-collateralization (or supervisory LGDs below the intercept value), the increase of LGDs is limited to the trend coefficient (beta) of the LGD elasticity to PDs, which allows banks to calibrate the LGD at the starting point to their own LGD estimate. - RWAs estimation: in accordance with IRB method under Basel III using through-the-cycle PDs, plus adjustments for loan portfolio concentration and changes in default risk. - Alternative RWAs estimation: in accordance with AIRB method under Basel III, plus adjustments for loan portfolio concentration and changes in default risk (i.e., RWAs for credit risk are reduced by the RWAs of defaulted exposures, which are assumed to be 2.5 times the average RWAs for non-defaulted exposures (accounting for the fact that risk-weights for defaulted exposures were higher prior to default)). ### Stress test RWA and loss assumptions - RWAs for derivatives remain constant under the baseline scenario but increase by 30 and 50 percent under the “severe” and “more severe” scenario, respectively. - Increase in IRB credit RWAs based on IRB banks’ historical data and stress test estimates, external data and internal analysis. - RWAs for operational and market risk remain constant throughout the forecast period (in some top-down approaches). - RWAs for credit risk are sensitive to the regulatory impact due to Basel III based on firm’s own data; credit RWAs are subject to the Basel I floor and sensitive to both changes in PDs and portfolio concentration, including: (a) nonlinear effect of changes in PDs and (b) concentration risk impact on RWAs. - RWA impact of defaulted loans: The risk-weights for credit risk are subsequently reduced by the RWAs of defaulted exposures, which are calculated by banks’ internal models or approximated by taking... ### Regulatory, accounting and market-based standards applied - Full Basel III transition schedule. - Capital definition according to the Basel III framework; includes phase-in of capital deductions and the phase-out of non-eligible forms of capital, without consideration of grandfathering, during the forecast horizon. - Phase-in of adjustments to common CET1 capital: reduction of CET1 capital (such as goodwill, deferred tax assets and minority interests that exceed the permissible limit) deducted at a rate of 20 percent p.a. between 2014 and 2018; in the BU exercise, firms must document deductions if the total amount is less than 20.1 percent (which is the average value for Group 2 (small banks) according to the results from the latest Basel III monitoring exercise as of 30 June 2012 (BCBS, 2013b)). - Phase-out of non-Tier 1 and Tier 2 capital elements: the higher of either (i) 10 percent (per annum) of the amount of capital to be phased-out based on results of the quantitative impact study (QIS) (BCBS, 2010) at 16.6 percent ( 1.7 percent p.a.) or (ii) the amount of capital maturing each year of the stress test horizon between 2014 and 2018 (BCBS, 2010a). - Basel III hurdle rate for 2015. - Capital definition according to the Basel III framework, including phase-in of adjustments to common CET1 capital and phase-out of non-eligible forms of capital, without consideration of grandfathering, during the forecast horizon (same as “Bottom-up by Banks”). - Under the concept of market-implied capital adequacy (MCAR) of the Systemic CCA model, the implied asset value of a firm corresponds to Pillar 1 RWAs. *Source: _cr14210 - 5. Regulatory and Market-based Standards and Parameters* ### 2.5 times the average ### 2.5 times the average ### RWA and Capital Assumptions - RWAs for market and operational risk remain constant throughout the forecast period. - RWAs for credit risk change due to higher PD and LGD for the major asset classes under the IRB approach. - RWA impact of defaulted loans: no replenishment of loan portfolio. - Basel III applied; banks are subject to the Basel I floor. - Regulatory impact due to Basel III based on HKMA’s own data. - Procyclical effect (IRB banks only): - Overall credit RWA under IRB: ~ +35% (More severe) - Overall credit RWA under IRB: ~ +25% (Severe) - Increase of RWA for credit risk: +17% (Baseline, according to the credit growth projection under the baseline scenario same as IMF TD solvency test) - Derivatives RWA: - RWA +50% (More severe) - RWA +30% (Severe) - RWA +0% (Baseline) ### Reporting Format and Capital Adequacy Outputs - Basel III metrics reported for each year of the risk horizon: Common Equity Tier 1, Tier 1, Total Capital, plus conservation buffer. - Firms reported capital adequacy each year over the forecast horizon based on an output template; in case of a capital shortfall, recapitalization needs were calculated. - Staff determined capital adequacy for each year; major risk drivers identified: profitability, haircuts on sovereign debt holdings, capital phase-in/phase-out and increases of RWAs according to Basel III. - Dividend payout policy under scenarios: - Dividend payout is limited if the firm reports profits over the past year (and exhibits sufficient Tier 1 capitalization) but CAR falls below a certain threshold which reflects the average Pillar 2 minimum CAR requirement imposed by the HKMA on the local banks under Basel III and the historical capital buffer of local banks over that requirement (3.2 percent). - Firms that are not capital constrained will have to pay out at least 45 percent of earnings after tax each year. - The payout ratio declines by 10 percentage points for each 0.5 percentage point decline in the capital buffer; 100 percent retention if capital buffer is lower than the threshold. - Basel III capital definition applied in line with reported values in line with Basel III transition schedule. ### Scenario Assumptions (HKMA Solvency Top-down Detailed Assumptions — Table 6) - Non-bank Mainland China exposures: - More severe: 99% confidence level according to GDP-based model (with floor at 4.0%) - Severe: 90% confidence level according to GDP-based model (with floor at 2.0%) - Baseline: 0.3% (according to GDP-based model with the baseline scenario same as IMF TD solvency test) - RML & property investment loans: - More severe: 1.24% - Severe: 0.50% - Baseline: 0.01% - Other loans: - More severe: 99% confidence level according to GDP-based model (with floor at 3.0%) - Severe: 90% confidence level according to GDP-based model (with floor at 1.8%) - Baseline: 0.17% (according to GDP-based model with the baseline scenario same as IMF TD solvency test) - Off-balance sheet exposures drawdown rate and loss: - More severe: drawdown rate of 20%; same loss rate as for "other loans" - Severe: drawdown rate of 15%; same loss rate as for "other loans" - Baseline: drawdown rate of 10%; same loss rate as for "other loans" - Debt securities - securitization/non-securitization: - More severe: 99th percentile historical loss rate based on Moody's data - Severe: 90th percentile historical loss rate based on Moody's data - Baseline: latest available two-year loss rates provided by Moody's reports - Credit card lending: - More severe: 99th percentile of individual banks' historical loss (floor at 99th percentile of industry's historical loss (20.9%)) - Severe: 90th percentile of individual banks' historical loss (floor at 90th percentile of industry's historical loss (18.8%)) - Baseline: 50th percentile of individual banks' historical loss - Stock-related exposures — investment in listed shares: - More severe: 99th percentile of max drop in HSI within two yrs (-59.6%) - Severe: 90th percentile of max drop in HSI within two yrs (-47.4%) - Baseline: +16.64% (according to the baseline scenario same as IMF TD solvency test) - IPO margin lending: - More severe: 3.00% - Severe: 1.00% - Baseline: 0.0% - Non-IPO margin lending: - More severe: 4.09% - Severe: 1.55% - Baseline: 0.0% - Other loans to stock brokers, other loans to non-stock brokers: - More severe: 99th percentile of individual banks' historical loss (floors specified in table) - Severe: 90th percentile of individual banks' historical loss - Baseline: 50th percentile of individual banks' historical loss - Property prices: - More severe: -50% - Severe: -30% - Baseline: -6% (according to the baseline scenario same as IMF TD solvency test) - HKD interest rate shocks: - Best lending rate: +400bps (More severe); +200bps (Severe); +10bps (Baseline) - Savings deposit rate: +400bps; +200bps; +0bps - Time deposit rate: +500bps; +250bps; +3bps - HIBOR: +500bps; +250bps; +18bps (Baseline value per IMF TD solvency test) - Net interest income: - More severe: the worse of -5% or individual banks' decrease in 97/98 - Severe: the worse of -3% or individual banks' decrease in 97/98 - Baseline: Projected figures by individual banks - Non-interest income: - More severe: the worse of -20% or individual banks' decrease in 97/98 - Severe: the worse of -10% or individual banks' decrease in 97/98 - Baseline: Projected figures by individual banks ### Liquidity Stress Testing: Frameworks and Key Parameters - Implied Cash Flow (ICF) tests (IMF): - 5-day ICF test: cumulative inflow and outflow over 5 consecutive days; liquid financial asset haircut ranges given by asset type (e.g., haircut: 0 percent for cash); cumulative cash inflows include call-back rate: 20 percent per day for credit extension without collateral. - 30-day ICF test: non-cumulative; specific inflow/outflow rates specified (e.g., non-cumulative cash inflows related to credit extension without liquid collateral [100]). - HKMA 7-day deposit run-off and LOLR analysis: - Haircut for Exchange Fund bill and notes (EFBNs): 5 percent. - Negotiable certificates of deposit (NCDs) and negotiable debt instruments (NDIs) haircuts: 40 percent. - Net due from unconnected banks within one month call-back rate: 40 percent. - Average overall (non-cumulative) deposit outflow of 17 percent in 7 days (resulting from discount factors: retail deposits discount factor 2 percent per day). - Enhanced Liquidity Stress Test (ELST) assumptions: - Liquid asset inflows and call-back rates provided (e.g., due from connected/unconnected banks within three months [80]). - Non-cumulative outflows include LCR QIS-consistent deposit run-off for retail/other banks [floor at 20/30]. - Liquidity stress test with shocks to market and credit risk (one-year horizon): - At the 90th percentile: credit spreads of corporate bonds rated 'AA' or higher, 'A' and 'BBB' rise by around 0.4 percent, 1.7 percent and 1.74 percent, respectively. - Corporate bonds with speculative grade ratings or unrated rise by around 10.5 percent. - Market value of equities and structured financial assets decline by 27 percent and 32 percent, respectively. - Interest rate hike: around 125 basis points. - Classified loan ratio of banks increases by 200 basis points. - Maximum monthly run-off rate for retail deposits: 42 percent; wholesale deposits: 100 percent. - Cash inflows from income reduce by 10 percent. - HKMA LCR and NSFR (Basel III standard measures): - LCR: stock of HQLA to cover 30-day net cash outflows; cash inflows capped at 75 percent of cash outflows; Level 2A limited to 40 percent and Level 2B to 15 percent of total HQLA stock. - LCR haircuts and inflow/outflow percentages specified by instrument (e.g., high-quality corporate bonds rated 'AA-' and higher [15]; corporate bonds (rated 'A+' to 'BBB-') [50]). - NSFR: required stable funding calculations with RSF and ASF weights specified (e.g., capital and long-term debt >1 year [100]; 'stable deposits' of retail and small business customers (< 1 year) [90]). - HKMA notes: no consideration of access to HKMA liquidity via the LOLR facility on the basis of non-LCR asset buffer eligible HQLAs; draw-down rate assumptions and national discretions specified (e.g., 5 percent run-off rate for retail stable deposits). ### Sample Coverage and Key Descriptive Statistics - Sample coverage and reporting bases: - Group 1: Locally incorporated, licensed banks (19) — combined and consolidated reporting used in various tests. - Group 2: Foreign branches (8). - Total sample assets across reporting bases shown in tables and figures (e.g., Total assets in Table 14: 11,699.4; Total capital and RWA comparisons across reporting bases shown in figures). - IMF Top-down Solvency Test descriptive statistics / FSIs (Table 14): - Total assets (In billions of HKD): 11,699.4 - Total capital (In billions of HKD): 991.3 - Regulatory capital to risk-weighted assets (average): 16.5 - Tier 1 capital to risk-weighted assets (average): 13.3 - CET1 capital to risk-weighted assets (average): 13.2 - Loan-to-deposit ratio: 97.2 - Return on assets (average): 0.5 - Return on equity (average): 0.2 - Loss rate (average): 1.2 - Non-performing loan ratio (average): 12.6 - Risk-weighting (In percent): 62.2 - Table 4 historical FSIs (2007–2013H1) include precise series (examples): - Return on average assets: 1.4, 0.6, 0.8, 0.9, 0.8, 0.9, 1.2 - Return on average equity 2/: 23.2, 13.4, 15.4, 15.2, 15.8, 15.3, 15.1 - Non-performing loans as percent of gross loans 5/8/: 0.7, 1.2, 1.6, 0.8, 0.7, 0.6, 0.5 - Regulatory capital to risk-weighted assets 2/4/1: 13.4, 14.7, 16.8, 15.8, 15.8, 15.7, 15.9 - Total assets (in percent of GDP) 9/: 627.0, 629.8, 641.0, 691.7, 709.8, 727.7, 704.8 ### Solvency and Liquidity Test Outputs and Key Risk Drivers - Solvency test outputs: - Aggregate capital ratios (Total Capital Ratio and CET1 Ratio) are presented across approaches (Bottom-up, Top-down IMF, Top-down HKMA, IMF SCCA) and scenarios (Baseline, Slow Growth, Severe Adverse) with evolution over 2014–2018 shown in figures. - The Systemic CCA analysis yields significant declines in capital ratios even under the baseline scenario due to the high statistical confidence level of joint expected losses affecting capital levels over the forecast horizon. - IMF TD solvency test risk drivers (Figure 16; cumulative changes in CAR over 2014-2018): - Net Operating Income (after taxes) - Basel III: Deductions from Capital - RWA (Procyclicality and Basel III) - Valuation haircuts (Govt Debt and Financial Bonds) - Liquidity test outputs: - Top-down liquidity ratios and ICF stress tests applied on all sample banks (19 Group 1 banks and 8 Group 2 banks) representing 70.9 percent of the banking sector's total assets. - HKMA 7-day test/LOLR analysis and Enhanced Liquidity Stress Test (ELST) results indicate medians and percentiles plotted over time; none of the sample banks would fail the 7-day/LOLR test if potential cash inflow from LOLR measures were ignored in the augmented test. - Maximum monthly run-off rates and one-year liquidity stress assumptions outlined above materially affect implied liquidity ratios (Wong-Hui (2009) implied liquidity ratio analysis included). ### Empirical Modeling and Satellite Estimates - Satellite model estimation (IMF TD solvency approach — Table 13) — signs and significance of estimated relationships: - Change (Δ) in interest income to total assets in %: coefficients with signs and significance indicated (e.g., -***, +***, etc.); lag structures included (e.g., +(-1)***). - Δ interest expenses to total assets in %: coefficients with signs and significance indicated (e.g., -**, +***, etc.). - Δ net fee and commission income to total assets in %: -*** and +*** entries across lags. - Δ stock of loan loss provisions to customer loans in % and Δ flow of loan loss provisions, Δ non-performing loans to customer loans in %: coefficients and significance levels shown; interactions with dummy for four largest local banks (G) documented. - “+” and “-” indicate sign of coefficients; ***, **, and * denote significance at 1, 5 and 10 percent levels respectively. *Source: _cr14210 - 2.5 times the average (IMF staff and HKMA stress testing material as presented in the supplied PDF).* ### INTRODUCTION ### _cr14210 - INTRODUCTION ### A. Background - The FSAP Update stress testing exercise for HKSAR comprises: - BU (bottom-up) solvency stress tests by selected local banks in HKSAR. - Cross-validation through several TD (top-down) tests, undertaken jointly by HKMA staff and the FSAP team. TD tests cover about 70 percent of the banking sector. - TD liquidity stress tests using supervisory data and parameters specified by both the HKMA and the FSAP team; these cover both local banks and foreign branches in the system. - Solvency tests use mid-2013 unaudited financial results as the baseline. - Hurdle rates are applied according to the Basel III implementation schedule. - Liquidity tests apply Basel III LCR and NSFR and various ICF tests (over one-week and one-month periods) to determine short- and medium-term resilience, independent of access to central bank liquidity. ### B. Objective - Purpose of the BU solvency stress testing component: - Assess capital adequacy of the banking sector based on end-2013: Q2 financial results. - Determine capacity of the banking sector to absorb realization of key macro-financial risks under a baseline and two adverse scenarios. - The stress test assesses system-wide vulnerabilities of selected local banks under different adverse macroeconomic scenarios and capital market conditions, considering three key channels of stress from a creditor perspective: - (i) Impairment charges (credit losses, other losses from held-to-maturity assets) and MtM valuation changes of fixed income securities (financial and government bonds) in trading and banking books. - (ii) Changes in pre-impairment income, including changes in funding costs. - (iii) Changes in RWAs. - Specific risk factors incorporated: cross-border developments (such as sovereign risk), foreign currency risk, impact of current regulatory reforms, and behavioral assumptions. - The multi-period FSAP stress test is for surveillance with a medium-term focus; typically involves very severe stress scenarios. Results have no immediate supervisory implications and no management action would be expected as a result of the FSAP stress tests. - Sample coverage and limitations: - The sample includes selected local banks representing about 50 percent of the banking sector. - Firms with separate regulated businesses (e.g., insurance) that are ring-fenced may exclude those businesses from the stress test. - Key limitations are acknowledged and reflected as caveats. ### Consistent assumptions and reporting - Macroeconomic and financial variables projected for input into solvency stress tests include: inflation, unemployment, housing prices, commercial property prices, interest rates (short-term interest rates and asset swap rate curve (up to 10 years)), net household income, and equity prices (Appendices III and IV). - Prescriptive assumptions provided cover: - (i) Risk factors: loss rates, market risk impact on fixed income holdings due to credit risk/FX risk, taxes, and funding costs. - (ii) Behavioral adjustments: balance sheet and credit growth, dividend pay-out, asset disposal, and capital raising. - (iii) Regulatory changes: capital requirements, RWAs, and the definition of capital. - Exclusions from consideration: - Managerial interventions announced/implemented after the cut-off date (end-June 2013), such as ongoing de-risking/de-leveraging through restructuring, run-offs and divestments. - Potential mitigating factors (managerial actions, contingent capital arrangements, bail-in provisions) are not considered; results should be viewed as the most conservative lower bound. - Timeline: Firms requested to conduct BU stress tests using end-2013: Q2 data and report final results to HKMA by November 11, 2013. HKMA to perform due diligence and report aggregate findings to FSAP team. ### Macro Scenarios - Three scenarios analyzed: one baseline and two adverse scenarios. Variables provided for each scenario: real GDP, household income, unemployment rate, price and cost developments (consumption prices, residential house prices, commercial real estate prices, equity market index), and interest rates (short-term interest rates and asset swap rate curve (up to 10 years)) (Appendices III and IV). - Baseline scenario aligned with IMF World Economic Outlook as of April 2013. - Adverse scenarios represent negative deviations from the central growth forecast and account for perceived risk and revisions. - Adverse scenario descriptions and key numeric projections: - “SG” scenario (long-term): - Underpinned by a broad-based slowdown in global growth driven by an increase in the cost of capital. - Growth in Mainland China and the United States of 6.6 and 3.0 percent on average, respectively. - Growth in HKSAR deteriorates by an average of 1.8 percent a year (to 2.6 percent) relative to baseline expectations of average annual growth of 4.4 percent. - Baseline average annual growth of 4.4 percent is slightly above the long-term (30-year) annual growth rate of 4.1 percent. - Cumulative negative deviation of about 9.1 percentage points in real GDP over a five-year period. - This equals more than one and a half standard deviations of the long-term (30-year average) two-year cumulative growth rate (5.6 percent). - “SA” scenario (short-term): - Investment shock aggravated by severe intensification of capital outflows impacting Hong Kong’s major trading partners. - Results in average growth of about 2 percent a year (or 2.4 percent on average lower than the baseline). - This is broadly in line with experience during the Asian financial crisis. - Cumulative negative deviation of about 12.1 percentage points in real GDP over a five-year horizon. - Magnitude equates to more than two standard deviations of the two-year cumulative growth rate (5.6 percent). - Growth in key trading partners: Mainland China 6.0 percent on average, United States 2.8 percent on average. - Cross-border effects: time-varying patterns aligned with forecasts for HKSAR under baseline and adverse scenarios, allowing for temporary or permanent shocks affecting countries where banks operate (Mainland China, Japan, the United Kingdom, the United States). - Interest rate projections: - Short-term interest rate projections supplemented by asset swap rate curve tenors to isolate credit spreads (Appendix IV). - Three-step estimation process: - Determining the asset swap rate curve at maturities between three months and 10 years based on the median of observations from January 1, 2013 to September 6, 2013 (“historical term structure of interest rate swaps”). - Fitting the sovereign yield curve at maturities between three months and 10 years using the arbitrage–free generalized Nelson-Siegel-Svensson term structure model, with discrete and continuously compounded spot rates obtained for the most liquid bonds of each sample country (“benchmark bonds”). - Re-calibrating the historical term structure to the asset swap rates based on the variation of the difference between the interest rate path and swap rates across the entire maturity. - Missing observations can be interpolated as the weighted average of the interest rate swap rates at four lower maturity terms closest to the maturity term of interest or via cubic splines. ### Satellite Models - Firms required to use satellite models (or expert judgment aligned with satellite outputs) to specify macro-financial linkages of firm performance over the forecast horizon. Satellite models should: - Cover at least the last five years. - Include lagged terms, GDP growth, interest rates, other macroeconomic variables, and firm-specific variables such as leverage, loan-to-asset ratio, and the funding gap (loans less deposits divided by total earning assets). - Appendix V provides possible satellite specifications. - Specific modeling guidance: - Credit losses in the banking book: - Forecasted from separate models for write-downs (and write-ups) derived from estimated PD and LGD specific to each sector (corporate, retail, public sector, other financial institutions). - Lending growth: - Assumed to grow broadly in line with nominal GDP or forecasted using a satellite model with changes in real GDP, short-term interest rates, headline inflation, and other relevant macro variables. - Operating profits: - Estimated using separate models for interest income, interest expenses, net fee and commission income, and operational expenses. - Net interest margin should broadly follow the term spread between the short-term interest rate and the lending rate. - Lending rates may increase (decrease) in excess of overall changes due to credit risk of certain exposures, provided portfolio-based assessment preserves overall narrowing/widening of net interest margins. - Projected net interest income should consider foregone interest from impaired assets but ignore pass-through effects of higher lending rates to borrowers with deteriorating credit quality. - Income and capital gains taxes are assumed to be 16.5 percent for firms recording a profit, and zero otherwise. Tax credit accumulated over the forecast period only can be applied during profitable years (before dividend payout). - Funding costs: - Estimated as a separate component of changes in interest expenses, including nonlinear sensitivity of funding costs to changes in solvency conditions. - Trading income under stress: - Should be aligned with changes in nominal GDP based on historical data. - Economic growth under each scenario and year can be matched to the corresponding GDP growth rate during the last 15 years (the growth rate closest to the simulated one). - Firms with exceptionally high historical trading losses may model the probability distribution of trading income and match point estimates to percentile levels of projected GDP growth relative to past volatility. *Source: _cr14210 - INTRODUCTION* ### 16. As a general rule, satellite models need to be clearly documented and back-tested in ### _cr14210 - 16. As a general rule, satellite models need to be clearly documented and back-tested in ### Overview and general requirements - Satellite models must be clearly documented and back-tested to demonstrate proper identification and sufficient robustness. - The HKMA, together with the FSAP team, will require full disclosure of the various satellite models and expert judgments on: - earnings capacity; - market and credit losses; - change in funding conditions under the various scenarios. ### Key assumptions (purpose and consistency) - Purpose: establish a consistent implementation of key risk drivers so stress test results can be compared across all banks within the sample. - Institutions must demonstrate a clear link between: - risk appetite; - business strategy; - capital planning relative to outcomes of different macro scenarios. - Institutions should assess ability to remain above regulatory minimum capital requirements through the period of stress, consistent with stated risk appetite. - Firms are encouraged to conduct additional solvency stress tests without the stated restrictions to assess aggregate impact of business strategies, behavioral adjustments under stress, and other idiosyncratic assumptions. - Stress test results should be augmented by explanations of credible mitigating actions firms could adopt if capital deteriorates significantly (e.g., management action, changes in business strategy, reinforcing the capital base, contingency plans). ### A. Balance Sheet Growth - General assumption: balance sheets are static except credit growth, which increases in line with nominal GDP. - Credit balance evolves with general credit growth experienced during the business cycle. - Overall lending (without contemporaneous asset impairments) and commensurate funding increase with a positive change in nominal GDP each year of the forecast horizon. - Growth rates may vary across loan portfolios if overall credit balance growth remains consistent with nominal GDP assumption. - Lending growth is matched by a proportionate increase in liabilities supporting the loan book without changes in funding mix (leverage ratio unchanged all else equal). - Assumption consistent with EBA stress test, which assumed a static balance sheet (except for pre-agreed disposals). - Specific rules: - Loan assets increase only if GDP growth in the current period is positive; otherwise, there is no new lending. - Exposures going into default are not replaced in the performing portfolio and generate no interest income in the period they become impaired (foregone interest must be taken into account when assessing net interest income under stress). - The rate of increase of lending and funding is applied to values reported at the end of the previous period, without considering impact of defaulted exposures on stock of outstanding loans at start of current period. - Deleveraging under stress: - Credit growth starts declining once a firm’s capital adequacy falls below a threshold of 2.5 percentage points above the minimum Tier 1 capital ratio applicable over the forecast horizon (example: 5.5 percent in 2014 (Y1) in the transition to Basel III). - If a firm falls below the threshold, credit growth declines by twice the capital shortfall in percentage points. - Example: for a Tier 1 capital ratio of half a percentage point below the regulatory minimum (and capital buffer), credit growth declines by one percentage point. - Each adjustment is made immediately after the period during which the potential for deleveraging is assessed (sequential reaction). - Defaulted loans are not replaced and impact portfolio growth calculation after the period during which they are realized. - Asset disposals and acquisitions: - Asset disposals and acquisitions over time should not be considered, except where agreed with legally binding commitments prior to the cut-off date of the exercise (end-June 2013). - Any interim (and approved) capital increase until end-2013 can be considered in calculations. - Note: prepaid mortgages are assumed to be refinanced (trigger no change in credit volumes and RWAs and only impact interest rate income). ### B. Risk Measurement and Riskiness of Assets - Internal capital adequacy models should be used; IRB capital requirements calculated from an economic capital model. - Account for point-in-time (PIT) level of RWAs and changes in RWAs under stress in economic terms. - Minimum coverage: credit risk and market risk. - Ideally tests based on credit-specific PD and LGD at firm and/or portfolio level; proxies permitted if unavailable (e.g., provisions, NPLs in lieu of PDs, country-level LGDs). - Market risk assessed via changes in interest and credit risks impacting valuation of investments; exposures included in credit risk assessment for banking book assets. - LGD under stress specifications for estimation of credit losses: - LGD(under stress)=0.3502+2.3408*PD (Moody’s, 2009); or - LGD(under stress)=0.4022+2.1535*PD (if down-cycle LGDs represent long-term averages). - In case of overcollateralization (or supervisory LGDs below the intercept value), the increase of LGDs is limited to the trend coefficient (beta) of the LGD elasticity to PDs. - Market risk valuation: assess changes in FX rates, short-term interest rates, and credit spreads implied by proposed valuation haircuts for fixed income securities issued by sovereigns and financial institutions. - RWAs and regulatory changes: - Higher RWAs due to regulatory changes and deteriorating credit quality should be taken into account using economic capital models and/or expert judgment. - If internal models cannot derive changes in risk weights, prescriptive standards apply. - RWAs for operational risk remain constant throughout the forecast period. - RWAs for credit risk and market risk are subject to the Basel I floor and sensitive to regulatory impact due to Basel 2.5 and Basel III. - Firms encouraged to estimate regulatory impact on risk weights using internal models (as of end-June 2013) or select minimum increases in risk weights based on general QIS results: - risk weights should increase by at least 5.13 percent [credit risk] and 0.47 percent [market risk] (independent of asset growth) in 2014 and 2015, respectively (assuming that 1/3 of the change in RWAs has already been absorbed during implementation of Basel 2.5 and III in 2013). - Alternative: where Basel III risk weights for some exposure types are difficult to estimate, firms can select greater of minimum increase in risk weights for certain sub-categories due to Basel 2.5 and Basel III (e.g., securitization) and double original Basel II weights for those sub-categories. - Lower values for changes in risk weights based on internal models need documentation and HKMA approval. - Sensitivity of credit RWAs to PD changes and concentration: - Firms encouraged to estimate economic impact of stress on risk weights using internal models (as of end-June 2013) in addition to regulatory increases. - Alternatively, additive increase of risk weights under stress using general assumptions: - Nonlinear effect of changes in PDs on RWAs determined by fixing asset correlations to lowest level of PDs (Aaa-rating) and LGD to 45 percent. Marginal increase formula: - delta_RWA(in percent)=0.12*delta_PD^2-0.049*delta_PD+0.006 - Impact of concentration risk on RWAs at portfolio level (PD=0.4 percent): - delta_RWA(in percent)=100*(0.02+12.6*HHI Parameter) - delta_RWA increases the formula above by 1+(PD of bank portfolio/0.4%-1)*0.1 for PDs>0.4 percent for each 0.4 percentage point increase of the PD if PD>0.4. - Example: a very large portfolio with low concentration (HHI ≤0.0006) and an EAD-weighed PD of 0.8 percent would be expected to experience an increase of RWAs by 3.0 percent due to concentration risk on unexpected losses. - Impact of defaults on RWAs: - Reduce total credit risk RWAs by the RWAs of defaulted exposures, approximated by taking 2.5 times the average RWAs for non-defaulted exposures. - Reporting and data standards: - Use granular data: credit and market risk parameters, sectoral credit information, securities in trading and banking book, regulatory data on capital and capital adequacy. - PDs and LGDs are assumed to be “through-the-cycle” (TTC); relevant tests may be run based on “PIT risk parameters by calibrating parameters to accommodate the severity of the chosen scenario”. - No feedback effect of firms’ lending on macro variables is assumed. - No changes to portfolio allocation to reduce RWAs or to firms’ lending standards beyond change in credit balance according to the constant loan book assumption. ### C. Dividend Payout Rule / Retained Income - Assessment of potential capital shortfall conditional on dividend payout assumptions and any repayment of public sector support (if applicable). - Dividend rules: - Dividend payouts are payable out of the previous year’s operating profit and cannot result in an ex post drop below any minimum capital requirements. - Only well-capitalized firms (i.e., firms that meet the minimum capital requirement) that generate positive earnings after taxes are assumed to pay dividends. - Dividends paid only by firms that satisfy all three measures of capital adequacy (total capital, Tier 1, and CET1 capital ratios) and exhibit a leverage ratio of no less than three percent in a given year (after creating adequate provisions for impairment of assets and transfer of profits to staff benefits and statutory reserves). - Dividend payout specifics: - Dividend payout ratio = percentage of “dividend payable in a year” to “net profit during the year”. - Minimum payout of profitable firms set to 40 percent of profits. - If firm meets minimum total CAR of 8.0 percent (after envisaged dividend payout and, at the same time, exhibits sufficient Tier 1 and CET1 capitalization) but falls below 10.5 percent threshold during previous period, it is considered capital-constrained and follows a schedule of fixed dividend payouts during the same period. ### D. Additional Elements Impacting Profits and Losses — Funding Risk - Treatment of stress-induced funding costs: - Explicit as additional interest expenses but applies only to undercapitalized firms. - Applies only if Tier 1 capital ratio (after stress) falls below applicable hurdle rate (including a historical capital buffer of 2.5 percent). - Preferably non-linear increase of funding costs for firms experiencing high degrees of short-term stress should be based on proposed approaches or reasonable internal models calibrated on historical/theoretical data. - Estimation conventions: - Annual increase of funding costs unaffected by possible balance sheet deleveraging and assumes a constant funding structure. - In each year, estimated impact of shocks to firm’s balance sheet on cost of funding during the previous year is applied (avoids simultaneity between contemporaneous losses and higher costs of capital). - Funding structure is fixed unless there are well-specified funding plans discussed and agreed at board level before end-2013: Q2. - Change in overall funding costs for all liabilities with residual maturity up to one year is calculated each year. - All short-term debt (excluding deposits) is funded at the new funding rate. - Only long-term debt due in each year is re-priced at the new rate. - Deposit rate moves in proportion to change of overall funding costs, weighted by levels of liabilities with residual maturity up to one year and all other (longer term) debt. *Source: _cr14210 - 16. As a general rule, satellite models need to be clearly documented and back-tested in* ### 29. An empirical approach can be used to estimate the annual increase of funding costs ### _cr14210 - 29. An empirical approach can be used to estimate the annual increase of funding costs ### Estimating annual increase in funding costs (empirical/satellite model) - Use the average historical sensitivity of interest expenses to changes in capitalization to estimate the annual increase of funding costs over the forecast horizon. - A satellite model can link short-term funding costs to one-period lagged risk-weighted capital ratios (and/or leverage)—possibly conditional on changes in loan loss provisions and the funding gap—to simulate a nonlinear effect with respect to default risk. - The marginal change in funding costs from the satellite model should be added to the estimated (general) interest rate expense. ### Generic add-on factor (linear approximation) for short-term funding costs - If firms cannot meet the satellite-model precondition, apply a generic (linear) formula for an “add-on factor” to interest expenses to approximate macro-financial linkages of short-term funding costs in stress situations (stylized specification in Appendix V). - The following additional costs (in basis points or bps) are applied each year of the forecast horizon for each percentage point of the forecasted capital ratio below the Tier 1 hurdle rate (+2.5 percentage point buffer): - retail deposits (all maturities): 5 basis points - interbank deposits with a contractual maturity of less than one year or without maturity: 20 basis points - short-term wholesale funding (i.e., wholesale deposits and debt securities): 10 basis points - long-term wholesale funding: 50 basis points, multiplied by the share of short-term liabilities in total liabilities - Note: Only a fraction of the long-term liabilities will need to be renewed over the short term. ### Alternative: aggregate funding risk model (Moody’s KMV EDFs) - Firms can apply results from an aggregate funding risk model based on the historical relation between Moody’s KMV Expected Default Frequencies (EDFs) and (weighted-average) funding costs of banks, accounting for variations in the relative importance of different funding sources for specific countries. - The general funding-cost elasticity specification infers risk-based capital ratios for rating grades from the Basel II capital model by using the confidence level corresponding to the EDFs; the implicit sensitivity of the economic capital ratio to observed (average) funding costs determines an “add-on” to expected interest expenses (Appendix VII). - The method is heavily based on empirical data and determines changes in the cost of debt for the average banking sector. ### Valuation changes to fixed income holdings (sovereign and financial bonds) - The stress test must include comprehensive assessment of sovereign risk covering trading and banking books, focusing on projection of valuation haircuts for fixed income holdings (and long derivative positions) of sovereign debt and financial bonds. - Firms are asked to adopt IMF estimates of valuation haircuts (Appendix VIII), which are based on an assumed increase of sovereign credit spreads consistent with market expectations and a constant common shock to interest rates at all maturities of 50 basis points in the adverse scenarios. - Firms may apply implied changes in credit spreads (rather than valuation haircuts) together with the interest rate term structure to fully reevaluate holdings and derivatives positions (trading book and banking book). - Calculation method: - Valuation haircuts are based on valuation of government bonds using forward-looking information from CDS markets; future sovereign bond prices are calculated using end-year risk-free rate and a density forecast of expected default risk derived from historical variation of forward rates on sovereign CDS contracts at different maturities (Appendix IX). - Severity assumptions are contingent on scenario choice: baseline (current market expectations) and adverse (75 percent confidence level density forecast) with a common 50 basis point interest rate shock. - Application: - Haircuts should be applied each year of the forecast horizon to all relevant sovereign and financial sector debt exposures in the investment book (HtM) as well as AfS and trading accounts, including local government. - Exposures include all direct and indirect sovereign exposures to all significant countries, including bonds issued by financial institutions in those countries; net exposure = gross (long) exposures net of cash (short) positions (without derivative hedges such as CDS), including on- and off-balance sheet assets and claims. - Exclusions: cash at central banks and repos or asset swaps where there is no economic interest in the security (e.g., instruments held against assets pledged to the HKMA or the People’s Bank of China). - Direct derivatives: subject to fair value adjustments based on relevant shock; indirect derivatives: treated similarly, subject to fair value adjustments and credit value adjustment. - Haircuts are applied to adjusted (MtM) balance sheet values. For investment book exposures, additional market value adjustment to historical cost until end-2013: Q2 should be added to theoretical valuation losses attributable to changes in sovereign risk. - Haircut loss amounts for AfS and HtM exposures should be fully reflected in the income statement (not passed through revaluation reserve for AfS). - Size-weighted maturity profile of sovereign exposures determines choice of valuation haircut for relevant maturity term (Appendix VIII). - Example: HKSAR government bonds (average maturity of five years) haircuts by year: - baseline scenario: –1.67% (2014), –2.64%+1.67%=-0.97% (2015), –3.60%+2.64%=-0.96% (2016), –4.33%+3.60%=-0.73% (2017), –4.81%+4.33%=-0.48% (2018) - adverse scenario(s): –5.60% (2014), –6.53%+5.60%=-0.93% (2015), –7.43%+6.53%=-0.90% (2016), –8.05%+7.43%=-0.62% (2017), –8.56%+8.05%=-0.51% (2018) - Aggregation simplification: firms can combine exposures into buckets and apply valuation haircuts on weighted-average basis using relative orders of magnitude as scaling factor (Domestic; Developed Asia; Emerging Asia; Core Europe; Peripheral Europe; the United States; Other countries). ### Valuation changes to foreign exchange positions (FX shocks) - Firms report aggregate impact of FX shocks on net open positions and FX assets (P&L and RWAs) in terms of appreciation of renminbi, euro, Japanese yen, pound sterling, Singapore dollar, and other material currencies versus the U.S. dollar. - Shock specification: - The shock for each currency = four times the maximum deviation of the annualized FX volatility during the 2008–11 period from the long-term FX volatility; impacts P&L and RWAs of the trading book in 2014 (at 100 percent of calibrated shock) and 2015 (at 50 percent of calibrated shock) only (Appendix X). - Example: net open FX positions denominated in Japanese yen should increase by 14.6 percent in Y1 (2014) and 7.3 percent in Y2 (2015). - Hong Kong dollar-U.S. dollar exchange rate moves within peg range: HK$7.75 to HK$7.85. - Impact applies only to adverse scenarios (not baseline) and should not generate knock-on effects on other elements of the stress test. ### Capital assessment (Basel III transition and hurdle rates) - Solvency assessed in accordance with BCBS changes of September and December 2010 (“Basel III”); FSAP hurdle rates follow the graduated Basel III schedule (Appendix XII). - Basel III changes include: - higher minimum capital requirement ratios (Tier 1 and CET1) - more restrictive definition of eligible capital (“capital deductions”) - higher asset risk-weightings - restriction on financial leverage via a minimum leverage ratio - As of January 1, 2014, minimum capital requirements relative to RWAs: - 4.0 percent common equity/RWAs (up from 2.0 percent prior to Basel III) - 5.5 percent Tier 1 capital/RWAs (up from 3.0 percent) - existing CAR of 8.0 percent total capital/RWAs remains - minimum Tier 1 leverage ratio of 3.0 percent - Graduated regulatory adjustments of CET1 (deductions and prudential filters), including amounts above the aggregate 10/15 percent limit for investments in financial institutions, mortgage servicing rights, and deferred tax assets from timing differences, scheduled to begin on January 1, 2014. - Definition of capital at end-2013: Q2 should be consistent with Basel III phased implementation (Appendices XII and XIII): - Starting point for CET1 and Tier 1: official definitions as laid out by the HKMA. - Phase-in of total regulatory adjustments to CET1: 20.1 percent (per annum) of CET1 capital deducted between 2014 and 2018; firms must document deductions if amount is less than 20.1 percent (~4.0 percent p.a.). (Note: 20.1 percent is average value for Group 2 (small banks) per Basel III monitoring as of 30 June 2012.) - Phase-out of non-Tier 1 and Tier 2 capital elements: the higher of either 10 percent (per annum) of amount of capital to be phased-out based on QIS results for Group 2 at 16.6 percent (~1.7 percent p.a.) or the amount of capital maturing each year of 2014–2018. - Existing capital instruments are not grandfathered until they mature for the tier in which they currently belong. - Firms should exclude the Tier 1 share of a material holdings deduction for securitization at the consolidated level. ### Output, reporting, and supervisory engagement - Firms assess capital adequacy under stress on a consolidated basis and report to the HKMA all capital measures for each year over the forecast horizon using the output template in Appendix XIII. - Reported metrics: (i) total capital, (ii) Tier 1 capital, and (iii) CET1 capital for each year of the forecast time horizon. - Firms should disclose composition of capital in each period and show calculated recapitalization needs if a capital shortfall arises. - Firms should report major risk drivers (profitability, credit/trading losses) and show marginal impact of including haircuts on sovereign exposures. - Firms may report alternative stress test results without considering restrictions on behavioral adjustment of banks as separate output. - Documentation requirements: - Firms should document estimation of important stress testing elements (funding costs, supervisory standards (risk-weights), macro-financial linkages (“satellite models” and/or expert judgment)) and demonstrate compliance with IMF-provided minimum standards. - Results should show RWAs for credit, market and operational risk, and specifications of macro-financial linkages affecting profitability and credit loss forecasts. - HKMA supervision and IMF interaction: - HKMA staff will engage with firms on stress testing to ensure consistency of assumptions, supervisory appropriateness, and suitability of models prior to final submission; banks are encouraged to involve line supervisors early for preliminary assessment and possible “pre-approval” of internal model-generated estimates. - HKMA staff will check results against historical experience, firms’ supervisory stress tests, and a TD version of the stress test exercise by IMF staff (jointly with HKMA staff) using satellite models estimated on aggregate data. - Publication: - IMF will publish stress test results only after consulting with the HKMA and subject to existing confidentiality agreements and IMF statutes governing data confidentiality; publication focus will be system-wide vulnerabilities and evolution of overall capital adequacy, and use information on hurdle-rate changes and Basel III capital treatment to support interpretation of results. ### Timeline for completion of solvency BU stress test (Appendix I) - September 16, 2013: Firms receive stress testing guidelines from HKMA - September 3-17, 2013: Technical follow-up with participating banks during first IMF FSAP mission - November 11, 2013: Firms report final results and HKMA prepares output for IMF FSAP team - November 18, 2013: HKMA communicates results to IMF FSAP team - November 14-26, 2013: Discussion of results by HKMA staff and IMF FSAP team during second IMF FSAP mission *Source: _cr14210 - 29. An empirical approach can be used to estimate the annual increase of funding costs* ### Appendix II. Key Bottom-up Solvency Stress Test Parameters ### Appendix II. Key Bottom-up Solvency Stress Test Parameters ### Scenarios - Three scenarios over five years (forecast horizon): (i) Baseline, (ii) “Slow Growth” (SG), (iii) “Severe Adverse” (SA). - Macroeconomic/financial variables conditioned on specific scenario: GDP (nominal and real), household income, unemployment, inflation, interest rates/asset swap rates (short-term and long-term), equity prices, commercial property price index, real estate price index, and credit growth. - Macroeconomic models run by HKMA and IMF staff for all variables with the exception of credit growth. - Cross-border effects considered in all macro scenarios; IMF staff provided estimates for real GDP growth consistent with the macroeconomic forecast for HKSAR and Mainland China under both baseline and adverse scenarios for all relevant countries affecting bank performance abroad. - Aim to ensure consistency with HKMA stress test and other FSAPs. ### Risk factors assessed - Categories assessed: Credit risk (loss rates), Market risk, Profitability, FX shock, Taxes. - Credit losses: - Based on satellite models depending on the selected scenario. - All sample banks report under IRB; use own internal estimates rather than the prescribed PDs/LGDs according to Basel II. - Estimates specific to each material loan category and include an increase of LGDs under stress: LGD(under stress)=0.3502+2.3408*PD (Moody’s, 2009) or LGD(under stress)=0.4022+2.1535*PD (if the down-cycle LGDs are based on a long-term average, i.e., “through the cycle”). - In case of overcollateralization (or supervisory LGDs below the intercept value), the increase of LGDs is limited to the trend coefficient (beta) of the LGD elasticity to PDs. - Market risk: - (a) FX shock to U.S. dollar (vis-à-vis most important currencies). - (b) Haircuts on holdings of both sovereign and all financial sector debt securities in both trading and investment books based on market expectations over five years after controlling for changes of market valuation using density forecasts of forward contracts on sovereign CDS spread over an estimation using the methodology developed by IMF staff. - Profit: - Components: interest income, interest expenses, net fee and commission income, and operating expenses. - Based on IMF/firm’s own satellite models (or expert judgment). - Reported net profit before tax at sample cut-off (end-June 2013) should be adjusted for extraordinary income/losses to avoid misleading results when extrapolating operational performance until year-end. - “Other income” changes with nominal GDP. - Trading income: - Based on satellite model or statistical matching of both trading income and GDP growth using a parametric fit of their historical distribution (i.e., a decline in GDP growth is assumed to result in lower trading income). - Funding costs: - Based on IMF/firm’s own satellite model for additional interest expenses under stress, including (preferably) a nonlinear effect (with the IMF-suggested funding cost increase establishing an absolute minimum). - Changes in funding costs are unaffected by possible balance sheet deleveraging. - Sovereign risk: - Haircut on direct and indirect sovereign exposures, including financial bonds, in the banking and trading books based on market expectations over five years after controlling for changes of market valuation between January 2009 and June 2013 as developed by IMF staff. - Cash at central banks as well as repos or asset swaps where there is no economic interest in the security (for instance, instruments held against assets pledged to the HKMA or the People’s Bank of China) are excluded. - FX shock: - Firms report aggregate impact of FX shock on FX net open positions (both through the P&L and on RWAs) for currencies vis-à-vis the U.S. dollar: Chinese Renminbi, Euro, Japanese yen, Pound sterling, Singapore dollar, and other material currencies for the firm. - The shock for each currency should be four times the maximum deviation of the annualized FX volatility during the 2008-11 period from the long-term FX volatility (>10 years) for the adverse scenarios (not baseline) and impact the trading book in 2014 (100 percent) and 2015 (50 percent) only. - Tax assumption: - 16.5 percent in case of net operating profits, zero otherwise. - Tax credit after the first year of the stress period is taken into account. ### Behavioral adjustment of banks - Balance sheets: - Assumed to be static but credit growth is constant (i.e., lending and commensurate funding requirements increase in line with nominal GDP (if positive)), subject to a “deleveraging rule”; defaulted loans are not replenished. - Dividend pay-out rules (similar to Basel III minima): - Dividend payout depends on capitalization under stress. - Minimum dividend payout occurs only if the firm reported profits during the previous period. - If the total capital ratio is above 8.0 percent (after the envisaged dividend payout) but below 10.5 percent (which reflects the magnitude of the CAR and “capital conservation buffer” under Basel III), the firm is considered capital-constrained and needs to follow a payout schedule as displayed in Appendix XI. - Firms that are not capital constrained will have to pay out at least 40 percent of earnings after tax each year. - Credit growth: - In line with nominal GDP for banks with a Tier 1 capital buffer of 2.5 percentage points above the regulatory minimum (for Tier 1). - Credit growth decreases by 2 percentage points for each decrease in Tier 1 capital by 1 percentage point once the buffer is less than 2.5 percentage points. - Hence, growth becomes negative when capitalization is at minimum capital ratio. - Other business strategy considerations: - Asset disposals or acquisitions over time should not be considered, except where legally binding commitments existed prior to the end-date of data input (end-June 2013). - Maturing exposures are assumed to be replaced. - Any interim capital-raising until end-2013 can be considered in calculations. ### Regulatory standards - Capital requirements (‘hurdle rates’): - Hurdle rates for CET1, Tier 1 capital (T1), and total capital (CAR) according to the Basel III schedule (i.e., increasing from 2013 onwards with application to the five-year stress testing horizon from 2014 to 2018). - Changes in risk-weighted assets (RWAs): - RWAs for operational risk remain constant throughout the forecast period. - RWAs for credit risk and market risk are subject to the Basel I floor and sensitive to the regulatory impact due to Basel 2.5 and Basel III, which should increase by at least 5.13 percent [credit risk] and 0.47 percent [market risk] (independent of asset growth) in 2014 and 2015, respectively (assuming that 1/3 of the change in RWAs has already been absorbed during the implementation of Basel 2.5 and III in 2013). - Credit RWAs are sensitive to changes in PDs and portfolio concentrations: - Calculate the nonlinear effect of changes in PDs on RWAs: assume impact of LGDs on RWAs is linear; extract elasticity of unexpected losses leading to changes in RWAs from the Basel II IRB formula for corporate loans after fixing asset correlations to the lowest level of the PDs (a level corresponding to a “Aaa”-rating) and the LGD to 45 percent. - Marginal increase formula: delta_RWA(in percent)=0.12*delta_PD^2-0.049*delta_PD+0.006 (which assumes an initial elasticity of 0.6 of RWAs to PDs at the lowest level of PD). - Control for concentration risk impact on RWAs: delta_RWA(in percent)=100*(0.02+12.6*HHI Parameter) at portfolio level in the banking book, with an increase of delta_RWA by 1+(PD of bank portfolio/0.4%-1)*0.1 for PDs>0.4 percent, where HHI is the Herfindahl-Hirschman concentration measure (Appendix VI). - Firms may apply lower values for the initial increase of RWAs for credit risk if documented. - Where calculation of Basel III risk weights for some exposure types (e.g., counterparty credit risk) are difficult to estimate, risk weights are double those of the Basel II weights. - RWA impact of defaulted loans: - Risk-weights for credit risk are reduced by the RWAs of defaulted exposures, which should be approximated by taking 2.5 times the average RWAs for non-defaulted exposures (accounting for the fact that risk-weights for defaulted exposures were higher prior to default). - Capital definition: - According to the Basel III framework. - During the forecast horizon it has to comply with the envisaged phase-in of capital deductions and the phase-out of non-eligible forms of capital elements without consideration of grandfathering. - Phase-in of total regulatory adjustments to CET1 capital: - 20.1 percent of CET1 capital (e.g., goodwill, deferred tax assets and minority interests that exceed the permissible limit) is to be deducted between 2014 and 2018; firms must document deductions if the amount is less than 4.0 percent p.a. (Note: 20.1 percent is the average value for Group 2 (small banks) according to the results from the latest Basel III monitoring exercise as of 30 June 2012 (BCBS, 2013a)). - Phase-out of non-Tier 1 and Tier 2 capital elements: - The higher of either (i) 10 percent (per annum) of the amount of capital to be phased-out based on QIS results (BCBS, 2010a) for Group 2 (large banks) at 16.6 percent ( 1.7 percent p.a.) or (ii) the amount of capital maturing each year of the stress test horizon between 2014 and 2018. ### Outcome: reporting of results and additional outputs - Output template: - Firms report capital adequacy under stress based on the common capital measures (total capital, Tier 1 capital and CET1) for each year over the forecast horizon using the suggested output template. - In case of a capital shortfall, recapitalization needs are calculated. - Firms should report the major risk drivers (profitability, credit/trading losses, risk-weights) and show the marginal impact of including: - (i) haircuts on sovereign debt holdings; - (ii) capital phase-in/phase-out according to Basel III; and - (iii) FX shocks. - Firms may report alternative stress test results without considering the restrictions on the behavioral adjustment of banks as separate output. ### Macroeconomic projections for HKSAR under different scenarios (Table A3.1 highlights) - Table covers 2013 (est.) and projections 2014–2018 under Baseline, Slow Growth (SG), and Severe Adverse (SA) scenarios. - Real GDP (annual percentage change): - 2013 (est.): 2.96 - Baseline: 2014 4.39; 2015 4.37; 2016 4.46; 2017 4.46; 2018 4.51 - Slow Growth: 2014 1.78; 2015 2.66; 2016 2.70; 2017 2.99; 2018 2.95 - Severe Adverse: 2014 1.32; 2015 0.63; 2016 2.38; 2017 2.91; 2018 2.84 - Nominal household income (annual percentage change): - 2013 (est.): 4.49 - Baseline: 2014 4.67; 2015 5.13; 2016 4.56; 2017 4.69; 2018 4.82 - Slow Growth: 2014 1.53; 2015 2.62; 2016 2.33; 2017 1.82; 2018 2.06 - Severe Adverse: 2014 0.88; 2015 0.14; 2016 1.38; 2017 1.33; 2018 1.48 - Unemployment rate (% of labor force): - 2013 (est.): 3.30 - Baseline: 2014 3.46; 2015 3.57; 2016 3.70; 2017 3.82; 2018 3.84 - Slow Growth: 2014 4.15; 2015 4.78; 2016 5.54; 2017 6.18; 2018 6.75 - Severe Adverse: 2014 4.32; 2015 5.54; 2016 6.43; 2017 7.09; 2018 7.69 - Consumption prices (average CPI): - 2013 (est.): 3.95 - Baseline: 2014 3.67; 2015 3.57; 2016 3.54; 2017 3.53; 2018 3.53 - Slow Growth: 2014 3.09; 2015 2.21; 2016 1.81; 2017 1.65; 2018 1.58 - Severe Adverse: 2014 3.04; 2015 1.59; 2016 0.82; 2017 0.91; 2018 1.05 - Residential house prices: - 2013 (est.): 7.97 - Baseline: 2014 -1.68; 2015 -0.70; 2016 -1.20; 2017 -1.49; 2018 3.28 - Slow Growth: 2014 -9.54; 2015 -6.10; 2016 -6.51; 2017 -6.24; 2018 -1.84 - Severe Adverse: 2014 -11.97; 2015 -12.58; 2016 -7.06; 2017 -5.98; 2018 -1.97 - Commercial real estate prices: - 2013 (est.): -1.15 - Baseline: 2014 -9.35; 2015 -0.26; 2016 2.03; 2017 1.98 - Slow Growth: 2014 7.95; 2015 -17.05; 2016 -7.19; 2017 -4.99; 2018 -3.69; 2018 1.20 (table presents multiple entries across scenarios) - Severe Adverse: 2014 -19.26; 2015 -14.47; 2016 -6.96; 2017 -3.47; 2018 1.02 - Equity market index (Hang Seng index): - 2013 (est.): 7.00 - Baseline: 2014 8.06; 2015 7.94; 2016 8.00; 2017 8.00; 2018 8.04 - Slow Growth: 2014 -13.95; 2015 -9.13; 2016 -9.39; 2017 -7.89; 2018 -8.32 - Severe Adverse: 2014 -16.42; 2015 -20.00; 2016 -11.13; 2017 -8.28; 2018 -8.89 - Selected interest rates: - 3-month U.S. Treasury rate: - 2013 (est.) 0.10 - Baseline: 2014 0.17; 2015 0.13; 2016 0.26; 2017 0.46; 2018 0.57 - Slow Growth: 2014 0.10; 2015 0.20; 2016 0.90; 2017 2.10; 2018 3.10 - Severe Adverse: 2014 0.35; 2015 0.53; 2016 1.23; 2017 2.43; 2018 3.90 - 3-month HIBOR rate (%): - 2013 (est.) 0.38 - Baseline: 2014 0.41; 2015 0.51; 2016 0.62; 2017 0.72; 2018 0.82 - Slow Growth: 2014 0.66; 2015 1.10; 2016 2.24; 2017 3.39; 2018 4.53 - Severe Adverse: 2014 0.66; 2015 1.10; 2016 2.24; 2017 3.39; 2018 4.53 - 12-month HIBOR rate (%): - 2013 (est.) 0.87 - Baseline: 2014 1.00; 2015 1.10; 2016 1.20; 2017 1.30; 2018 1.40 - Slow Growth: 2014 1.25; 2015 1.68; 2016 2.84; 2017 3.99; 2018 5.15 - Severe Adverse: 2014 1.25; 2015 1.68; 2016 2.84; 2017 3.99; 2018 5.15 - Best lending rate (%): - 2013 (est.) 5.00 - Baseline: 2014 5.00; 2015 5.10; 2016 5.20; 2017 5.30; 2018 5.40 - Slow Growth: 2014 5.20; 2015 5.57; 2016 5.94; 2017 6.31; 2018 6.68 - Severe Adverse: 2014 5.12; 2015 5.41; 2016 5.71; 2017 6.00; 2018 6.30 - Memo item: Real GDP (P.R. China) - 2013 (est.): 7.80 - Baseline: 2014 7.70; 2015 7.60; 2016 7.50; 2017 7.50; 2018 7.50 - Slow Growth: 2014 5.60; 2015 5.52; 2016 6.96; 2017 7.50; 2018 7.44 - Severe Adverse: 2014 5.23; 2015 3.73; 2016 6.32; 2017 7.32; 2018 7.27 ### Projected real GDP for other countries under different scenarios (Table A3.2 highlights) - Real GDP growth, year-on-year (selected entries): - HKSAR: Baseline 2014 4.4; 2015 4.4. Deviation from Baseline: Slow Growth 2014 -2.6; 2015 -1.7. Severe Adverse 2014 -3.1; 2015 -3.7. - China: Baseline 2014 7.7; 2015 7.6. Deviation from Baseline: Slow Growth 2014 -2.1; 2015 -2.1. Severe Adverse 2014 -2.5; 2015 -3.9. - Australia: Baseline 2014 3.3; 2015 3.1. Deviation from Baseline: Slow Growth 2014 -0.8; 2015 -0.4. Severe Adverse 2014 -0.9; 2015 -0.9. - France: Baseline 2014 0.9; 2015 1.5. Deviation from Baseline: Slow Growth 2014 -0.4; 2015 -0.2. Severe Adverse 2014 -0.5; 2015 -0.5. - Germany: Baseline 2014 1.5; 2015 1.3. Deviation from Baseline: Slow Growth 2014 -0.8; 2015 -0.5. Severe Adverse 2014 -3.1; 2015 -3.7. - India: Baseline 2014 6.2; 2015 6.6. Deviation from Baseline: Slow Growth 2014 -2.0; 2015 -1.7. Severe Adverse 2014 -2.4; 2015 -3.2. - Japan: Baseline 2014 1.4; 2015 1.1. Deviation from Baseline: Slow Growth 2014 -1.5; 2015 -1.5. Severe Adverse 2014 -1.8; 2015 -2.8. - South Korea: Baseline 2014 3.9; 2015 4.0. Deviation from Baseline: Slow Growth 2014 -1.3; 2015 -0.9. Severe Adverse 2014 -1.6; 2015 -1.8. - Singapore: Baseline 2014 5.1; 2015 4.2. Deviation from Baseline: Slow Growth 2014 -0.5; 2015 -0.5. Severe Adverse 2014 -0.6; 2015 -0.9. - United Kingdom: Baseline 2014 1.5; 2015 1.8. Deviation from Baseline: Slow Growth 2014 -0.7; 2015 -0.4. Severe Adverse 2014 -0.9; 2015 -0.9. - United States: Baseline 2014 2.7; 2015 3.5. Deviation from Baseline: Slow Growth 2014 -0.9; 2015 -0.8. Severe Adverse 2014 -1.0; 2015 -1.5. *Source: _cr14210 - Appendix II. Key Bottom-up Solvency Stress Test Parameters* ### Appendix IV. Estimated Asset Swap Rate Curve ### Appendix IV. Estimated Asset Swap Rate Curve ### Estimated Asset Swap Rates — United States, P.R. China, and Hong Kong SAR (In percent, end of period) - Maturity / Term / USD CNY HKD (first scenario block) - 3M: 0.23 1.69 0.03 0.33 1.79 0.13 0.43 1.89 0.24 0.53 1.99 0.34 0.63 2.09 0.44 - 6M: 0.35 2.30 0.05 0.45 2.40 0.15 0.55 2.50 0.25 0.65 2.60 0.35 0.75 2.70 0.45 - 9M: 0.40 2.76 0.08 0.50 2.86 0.18 0.60 2.96 0.28 0.70 3.06 0.38 0.80 3.17 0.48 - 1Y: 0.42 3.11 0.13 0.52 3.21 0.23 0.62 3.31 0.33 0.72 3.42 0.43 0.82 3.52 0.53 - 2Y: 0.37 3.72 0.39 0.47 3.82 0.49 0.57 3.92 0.59 0.68 4.02 0.69 0.78 4.12 0.79 - 3Y: 0.38 3.72 0.65 0.48 3.82 0.75 0.58 3.92 0.85 0.68 4.02 0.95 0.78 4.12 1.05 - 4Y: 0.50 3.63 0.87 0.60 3.73 0.98 0.70 3.83 1.08 0.80 3.93 1.18 0.90 4.03 1.28 - 5Y: 0.69 3.62 1.07 0.79 3.72 1.17 0.89 3.82 1.27 0.99 3.92 1.38 1.09 4.02 1.48 - 6Y: 0.91 3.69 1.26 1.01 3.79 1.36 1.11 3.89 1.47 1.21 3.99 1.57 1.31 4.10 1.67 - 7Y: 1.11 3.80 1.46 1.21 3.90 1.56 1.31 4.00 1.66 1.42 4.10 1.76 1.52 4.20 1.86 - 8Y: 1.29 3.89 1.67 1.39 3.99 1.77 1.49 4.09 1.87 1.59 4.19 1.97 1.69 4.29 2.07 - 9Y: 1.43 3.90 1.89 1.53 4.00 1.99 1.63 4.10 2.09 1.73 4.20 2.19 1.84 4.30 2.30 - 10Y: 1.54 3.80 2.14 1.64 3.90 2.24 1.74 4.00 2.34 1.84 4.10 2.44 1.94 4.20 2.54 - Maturity / Term / USD CNY HKD (second scenario block) - 3M: 0.48 1.94 0.28 0.91 2.37 0.72 2.06 3.52 1.86 3.21 4.66 3.01 4.35 5.81 4.15 - 6M: 0.60 2.54 0.29 1.03 2.98 0.73 2.18 4.13 1.88 3.33 5.28 3.02 4.48 6.43 4.17 - 9M: 0.65 3.01 0.33 1.08 3.45 0.76 2.24 4.60 1.91 3.39 5.75 3.07 4.54 6.90 4.22 - 1Y: 0.67 3.36 0.38 1.10 3.80 0.81 2.26 4.95 1.97 3.42 6.11 3.12 4.57 7.26 4.28 - 2Y: 0.62 3.97 0.64 1.06 4.40 1.08 2.23 5.57 2.24 3.40 6.74 3.41 4.57 7.91 4.58 - 3Y: 0.63 3.97 0.90 1.07 4.41 1.34 2.25 5.59 2.52 3.43 6.77 3.70 4.61 7.95 4.88 - 4Y: 0.75 3.88 1.13 1.19 4.32 1.56 2.38 5.51 2.76 3.58 6.71 3.96 4.77 7.90 5.15 - 5Y: 0.94 3.87 1.32 1.38 4.31 1.77 2.59 5.52 2.97 3.80 6.73 4.18 5.01 7.94 5.39 - 6Y: 1.16 3.94 1.52 1.60 4.39 1.96 2.83 5.61 3.18 4.05 6.83 4.40 5.27 8.05 5.63 - 7Y: 1.37 4.05 1.71 1.81 4.50 2.16 3.05 5.74 3.39 4.28 6.97 4.63 5.52 8.21 5.86 - 8Y: 1.54 4.14 1.92 1.99 4.59 2.37 3.24 5.84 3.62 4.49 7.09 4.87 5.74 8.34 6.12 - 9Y: 1.69 4.15 2.15 2.13 4.60 2.59 3.40 5.86 3.86 4.66 7.13 5.12 5.93 8.39 6.39 - 10Y: 1.79 4.05 2.39 2.24 4.50 2.84 3.52 5.78 4.12 4.80 7.06 5.40 6.08 8.34 6.68 - Scenario labels appearing in table: - Baseline 2014 2015 2016 2017 2018 - Slow Growth (SG)/Severe Adverse (SA) 2014 2015 2016 2017 2018 ### Methodology for Estimated Asset Swap Rate Curve - Table A4.2 description and methodological notes: - Source: Bloomberg LP and staff calculations. - Methodology: a hybrid approach for calibration of a swap spread curve based on observable interest rates under different macro scenarios of the stress test. - Estimation follows a three-step process, building on a conventional term structure model of government bond yields (Nelson-Siegel-Svensson, 1994) and incorporates dynamics of the historical interest rate swap spread curve to generate a discrete interest rate path for the three major currencies of the aggregate sector balance sheet. ### Figures and Components (descriptive labels preserved) - NSS-model fitted yield curve based on benchmark govt bonds - Observable Interest Rate Swap Curve (U.S. dollar) - Historical Median (2012) - Observable Interest Rate Swap Curve (Renminbi) - Historical Median (2013) - Observable Interest Rate Swap Curve (HK dollar) - Historical Median (2013) - Synthetic Interest Rate Swap Curve — Total, Comp 1, Comp 2, Comp 3, Comp 4 - Axes labels shown: "In percent", "Maturity Tenor (In years)" ### Appendix V. Possible Satellite Model Specification — Key dependent variables and regressors (labels preserved) - Dependent Variable (examples of regressands shown): - Lagged Term - Total customer loans to total assets in % - 10-year sov. yield - 3-month interest rate (effective) - Real GDP growth (y-o-y) - Total assets (logarithm of total assets, lagged) - Leverage ratio (equity to total assets) in %, lagged - Nonperforming loans to customer loans in %, lagged - Funding gap (difference between customer loans and deposits in % of total assets, lagged) - Other macro variables: headline inflation, unemployment, and asset prices (real estate/equity markets) - Constant, R² - Modelled changes (examples of dependent changes): - Change (Δ) in interest Income to total assets in % - Δ interest expenses to total assets in % - Δ net fee and commission income to total assets in % - Δ operating expenses to total assets in % - Δ loan loss provisions (LLP) (write downs in lending business in % of customer loans) - Additional term: - "+ funding cost add-on per one percentage point (pcp) of capital (in percent)*([x]-Tier 1 capital ratio (after stress) in percent)*defined liabilities (see above)), where [x] represents the hurdle rate (e.g., 6%) for Tier 1 in each forecast period and a 2.5pcp capital buffer) 1/" - Note 1/: "This term represents an adjustment of interest expenses by additional funding costs (in basis points) at a level of capitalization consistent with the applicable hurdle rate Tier 1 capital in the stress test and the economic capital ratio approximation in Appendix VII." ### Appendix VI. Concentration Impact on RWAs under Stress (labels and chart axes preserved) - Chart labels preserved: - Increase in RWAs (In percent) - Portfolio PD (In percent) - Portfolio Size - HHI - Multiple - Example numeric sequences extracted from chart grid (rows preserved as presented): - large portfolio: 0.00 0.6 2.8 2.8 2.9 3.0 3.0 3.1 3.2 3.2 3.3 3.4 3.4 3.5 3.6 - medium-sized portfolio, low concentration: 0.00 1.0 3.3 3.3 3.4 3.5 3.6 3.7 3.7 3.8 3.9 4.0 4.1 4.2 4.2 - small portfolio, low concentration: 0.00 4.0 7.0 7.2 7.4 7.6 7.7 7.9 8.1 8.3 8.4 8.6 8.8 9.0 9.2 - small portfolio, concentrated: 0.01 10 15.9 16.3 16.7 17.0 17.4 17.8 18.2 18.6 19.0 19.4 19.8 20.2 20.6 - very small portfolio: 0.03 10 41.1 42.1 43.1 44.1 45.2 46.2 47.2 48.2 49.3 50.3 51.3 52.3 53.4 ### Appendix VII. Minimum Funding Cost: Empirical Estimation of Nonlinear Change (table of ratings and metrics) - Column headings (preserved labels): Rating scale (S&P, Fitch) | EDF or PD (one-year, in percent) | Funding costs (spread above T-bills, bps) | Economic capital ratio (Basel II (quasi-IRB)) | Change of funding spread (CAR elasticity) - Table rows (exact values preserved): - AAA: 0.0000 48.7 28.1 n.a. - AA+: 0.0000 68.7 27.3 0.0000 - AA: 0.0001 8.7 26.2 0.0000 - AA-: 0.0018 9.8 21.2 0.0002 - A+: 0.0029 0.0 19.7 0.0008 - A: 0.026 11.9 14.3 0.0055 - A-: 0.032 12.7 13.9 0.0180 - BBB+: 0.121 1.0 11.7 0.0386 - BBB: 0.139 25.9 11.0 0.0806 - BBB-: 0.291 44.6 9.9 0.1464 - BB+: 0.682 92.7 8.5 0.3541 - BB: 0.728 98.4 8.4 0.5738 - BB-: 1.791 229.4 7.1 1.0269 - B+: 2.453 310.5 6.7 2.0109 - B: 3.827 480.2 6.2 3.1611 - Note: "Funding cost exclude the cost of equity. The economic capital ratio includes a capital buffer above the hurdle rate of 2.5 percentage points." *Source: _cr14210 - Appendix IV. Estimated Asset Swap Rate Curve* ### Appendix VIII. Valuation Haircuts and Implied Credit Spread ### Appendix VIII. Valuation Haircuts and Implied Credit Spread ### Estimated valuation haircuts: scope and key rules - Exposures to Greece (if applicable) receive a valuation haircut of 100 percent. - Tables provide "Valuation Haircut for Sovereign Exposures (relative to end-June 2013), In percent" by: - maturity buckets: 1, 3, 5, 7, and 10 years (the most liquid bonds in ±0.5 years assumed representative of banks’ holdings); - forecast years: 2014, 2015, 2016, 2017, 2018; - scenarios: Baseline Scenario (current expectations 1/) and Adverse Scenarios (75th percentile value 2/); - alternative assumption: country-specific shock without (with) constant common shock (50bps) to interest rate level in the baseline (adverse) scenario. - "Current expectations" reflect market expectations implied by forward contracts (with maturity terms between 1 and 5 years) on 1-, 3-, 5-, 7-, and 10-year credit default swaps (CDS) at end-June 2013 (without considering past CDS dynamics). - The adverse scenarios were generated from the historically derived density forecast at the 75th percentile of variations in the forward CDS spreads between 2009 and 2013. ### Estimated credit spread shocks: scope - Table A8.2 reports "Implied Credit Spread Increase of Direct and Indirect Sovereign Exposures (relative to end-June 2013), In percent" by: - maturity buckets: 1, 3, 5, 7, and 10 years; - forecast years: 2014, 2015, 2016, 2017, 2018; - scenarios: Baseline Scenario (current expectations 1/) and Adverse Scenarios (75th percentile value 2/); - adverse scenarios include a common shock (50bps) to the interest rate level. - "Current expectations" and the adverse scenario generation match the definitions used for valuation haircuts above. ### Notable numerical parameters and scenario design - Greece exposures: 100 percent haircut. - Common interest-rate shock used in adverse scenarios: 50bps. - Statistical confidence level used for adverse scenario density forecast: 75th percentile. - Forecast maturities considered: 1, 3, 5, 7, and 10 years. - Estimation reference date for market-implied expectations: end-June 2013. - Forward-contract maturity window used to infer market expectations: between 1 and 5 years. ### Estimation methodology for sovereign risk valuation haircuts (Appendix IX): conceptual steps - Objective: calculate haircuts on sovereign debt exposures under different adverse macro scenarios based on valuation of government bonds in response to rising credit spreads using forward-looking information from CDS markets. - Representative bonds: for each country, the most liquid bonds in maturity buckets of one, three, five, seven and ten years (±0.5 years) are assumed representative of banks’ bond holdings. - Two pricing approaches are used to project future bond prices: - Zero-coupon (quasi-zero coupon) equivalent derived from coupon-bearing bond discounted cash-flow formula. - Direct discounted cash-flow pricing for coupon-bearing bonds (an approach consistent with European stress-test practice). ### Methodological building blocks (equations and mappings used) - Coupon-bearing bond discounted cash-flow pricing reconciled with zero-coupon bond pricing to project future bond prices contingent on idiosyncratic risk; coupon frequency denoted by n. - Quasi-zero-coupon price specification incorporates: - cumulative probability of default (PD), - constant loss-given-default (LGD), - risk-free rate r_f,t. - Cash CDS spread linkage (in basis points) to cumulative PD is expressed (see text) with scaling by 10,000 in the mapping between CDS spreads and PD (e.g., expressions contain factors "10, 000"). - When calculations are performed before year-end, valuation adjusts for the change in market valuation due to yield changes between the estimation window end-point t and forecasting start point t+τ; extrapolated year-end yield denoted r*_t is used to reflect valuation effects. - Future bond price with and without a common shock to the interest-rate term structure is computed using the estimated risk-free rate and i-period forward sovereign CDS spread f^+_{kjt,i}. - The i-period forward CDS rate f^+_{kjt,i} is derived as a density forecast from the historical dynamics of forward CDS spreads. The historical series of forward-spread observations is parametrically fitted to the generalized extreme value (GEV) distribution to account for large (nonlinear) fluctuations in sovereign CDS spreads. - GEV parameters used: shape parameter ξ (kappa), scale parameter σ, location parameter μ; greater absolute ξ implies heavier tail weight and slower tail convergence. - Density-forecast quantile selection for adverse scenarios: the 75th percentile (quantile at statistical confidence level a = 0.75). - The specification produces two projected prices per bond per year: one under current expectations and one under the adverse (75th percentile density) scenario, for each pricing method. ### Haircut computation and aggregation - For each bond b of country j and forecast year i the haircut is computed as the percentage drop from the base-year price: - Haircut = (1 - P_hat_{t+i} / P_{t}) × 100, where P_hat_{t+i} is the projected price under the scenario. - Two haircut series are produced corresponding to the two pricing methods. - The general haircut for each sovereign is an issuance size–weighted average of individual projected haircuts applied to q-number of bonds outstanding (weights are outstanding amounts bAmt). - Haircuts are constrained to be non-negative (no negative haircuts for price appreciation); sovereign exposure gains are ignored (i.e., no negative haircuts). - The aggregate trading valuation change or loss across all relevant direct and indirect sovereign exposures is computed as the exposure-weighted sum of the haircuted exposures across countries. ### Practical notes, limitations, and links to practice - The methodology uses market-implied forward CDS information observed at the end of the estimation period at time t for current expectations. - For adverse scenarios, point estimates of expected changes in default risk are chosen from the historical distribution of forward CDS spreads; haircuts under the adverse scenario reflect the volatility of market expectations via the 75th percentile density forecast. - The approach ignores second-order (convexity) effects of interest-rate changes on future bond prices in haircut determination. - Haircut estimation approach may overstate haircuts because projected yields are imposed on an unchanged set of bonds over the projected horizon (i.e., assumes no new government issuance and time-invariant coupons), thereby not accounting for possible changes in the sample of bonds and remaining maturities. *Source: _cr14210 - Appendix VIII. Valuation Haircuts and Implied Credit Spread; Appendix IX. Estimation Methodology for Sovereign Risk (tables and notes as provided in the source content).* ### Appendix X. Foreign Currency Shock ### Appendix X. Foreign Currency Shock ### Negative Shock to Net Open Foreign Currency Positions (In percent) - Currency pairs (with U.S. dollar as reference): CNY, EUR, JPY, GBP, CHF, SGD, AUD - Table values: - Maximum (2008-11): CNY 8.0, EUR 14.6, JPY 14.7, GBP 15.2, CHF 13.6, SGD 9.4, AUD 18.4, Avg. 13.4 - Long-term average: CNY 4.6, EUR 11.0, JPY 11.0, GBP 10.3, CHF 11.0, SGD 6.6, AUD 12.8, Avg. 9.6 - Multiple (4x): CNY 13.4, EUR 14.1, JPY 14.6, GBP 19.7, CHF 10.4, SGD 10.9, AUD 22.5, Avg. 15.1 - Note observations and application: - Sources: Bloomberg LP and staff calculations. - "Maximum (2008-11)" shows the highest implied annualized volatility of the respective currency pair (with the U.S. dollar as reference currency). - The long-term average was determined over an estimation period from Jan. 1, 2005 to Sept. 6, 2013. - Banks should apply 100% and 50% of the percentage shock to net open FX positions in Year 1 and Year 2 of the stress test horizon, respectively. - For net open FX positions to currencies other than those shown above, the simple average (rightmost column) should be applied. --- ### Appendix XI. Pay-out Ratio and Hurdle Rates ### Table A11.1. Pay-out ratio Conditional on Capitalization under Stress (In percent) - Capital buffer (In percent) | FSAP (minimum dividend pay-out ratio based on total capital ratio) | Basel III (maximum pay-out ratio based on CET 1 ratio) - 0-0.5: FSAP 5, Basel III 0 - 0.5-1: FSAP 10, Basel III 20 - 1-1.5: FSAP 15, Basel III 20 - 1.5-2: FSAP 20, Basel III 40 - 2-2.5: FSAP 30, Basel III 40 - >2.5: FSAP 40, Basel III 40 to 100 ### Table A11.2. Hurdle Rates (2014-2018) (In percent) - Forecast Year: Y1 (2014), Y2 (2015), Y3 (2016), Y4 (2017), Y5 (2018) - Hurdle Rates (under Basel III definition of capital): - (1) Reg. Minimum Total Capital: 8.0, 8.0, 8.625, 9.25, 9.875 - (2) Reg. Minimum Tier 1 Capital*: (assumption of conservation buffer applying to Tier 1 capital) 5.5, 6.0, 6.625, 7.25, 7.875 - (3) Reg. Minimum Common Equity Tier 1: 4.0, 4.5, 5.125, 5.75, 6.375 - Memo item: - Conservation Buffer: 0.0, 0.0, 0.625, 1.25, 1.875 ### Basel III Transition Schedule (Key phased values as presented) - Year progression and selected ratios/phase-ins: - Minimum Common Equity Capital Ratio: 3.50 (2011), 4.00 (2012), 4.50 (2013–2014/2015?), 4.50 (continues as listed) - Capital Conservation Buffer: 0.625 (2016 listed start), 1.25, 1.875, 2.50 - Minimum Common Equity + Capital Conservation Buffer: 3.50, 4.00, 4.50, 5.125, 5.75, 6.375, 7.00 - Phase-In (incl. amounts > CET1 limit for DTAs, MSRs & financials): 20, 40, 60, 80, 100, 100 - Minimum Tier 1 Capital: 4.50, 5.50, 6.00, 6.00, 6.00, 6.00, 6.00 - Minimum Total Capital: 8.00 (consistent across years listed) - Minimum Total Capital + Capital Conservation Buffer: 8.00, 8.00, 8.00, 8.625, 9.25, 9.875, 10.50 - Liquidity Coverage Ratio (LCR) introduction and phase-in: - Introduce minimum standard starting at 60, then 70, 80, 90, 100 (noted schedule) - Note in text: the LCR will be introduced as planned on 1 January 2015, but the minimum requirement will begin at 60 percent, rising in equal annual steps of 10 percentage points to reach 100 percent on 1 January 2019. - Net Stable Funding Ratio (NSFR): observation and introduction phases noted; phased out over a 10-year horizon beginning 2013 for certain items. --- ### Appendix XIII. Output Format for Reporting Firms to Hong Kong Monetary Authority (end-June 2013 → Forecast Years Y1–Y5) ### Reporting structure and scenarios - Reporting basis: solo, combined, consolidated - Macro scenario options: baseline, slow growth, severe adverse ### Hurdle Rates (with conservation buffer) as presented in reporting template - Hurdle Rate Total Capital (with conservation buffer): 8.0% (Y1 2014), 8.0% (Y2 2015), 8.6% (Y3 2016), 9.250% (Y4 2017), 9.875% (Y5 2018) - Hurdle Rate Tier 1 Capital (with conservation buffer): 5.5% (Y1), 6.0% (Y2), 6.6% (Y3), 7.250% (Y4), 7.875% (Y5) - Hurdle Rate Common Equity Tier 1 Capital (with conservation buffer): 4.0% (Y1), 4.5% (Y2), 5.125% (Y3), 5.750% (Y4), 6.375% (Y5) ### Example pre-populated balance and ratios (illustrative cell formatting values) - Total capital adequacy ratio (In percent): 12.1% - Tier 1 capital ratio (In percent): 8.6% - Common/core Tier 1 ratio (In percent): 6.1% - Total capital (in '000): 23,565,678 - Tier 1 capital (in '000): 16,957,185 - Common/core capital (in '000): 13,508,413 - Leverage (capital/assets): 3.3% - Return on total regulatory capital: 6.3% - Dividend yield (dividend paid/equity): 7.2% - Percentage of profits retained: 40.0% ### Credit risk and RWAs (Reporting template stress parameters) - PD/NPL ratio (EAD-weighed average): 0.50% - LGD (EAD-weighed average): 25.0% - Asset correlation (EAD-weighed average): 30.0% - Credit growth: 2.0% - Change of Credit Risk RWAs: 2.0% - Change of Market Risk RWAs: 2.0% - Change of Operational Risk RWAs: 2.0% - Notes on reporting fields: - Operating profit (before losses and impairments but after taxes) with breakdowns (interest income, interest expenses, fee income, trading income, operating expenses) - Credit losses/net impairments from lending - Losses/gains from trading and investment activities (incl. valuation haircuts and FX shock) - Net profit (after dividends paid and tax, if applicable) - Risk-weighted assets (RWAs) and changes in RWAs (In percent) - Capital needs to recapitalize bank (In HKD millions) and relative to total assets (In percent) - Sensitivity check: like "Main Results" above but without valuation haircuts --- *Source: IMF staff (appendices X–XIII as presented).* --- _Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/scr/2014/_cr14210.pdf_