## 1hkgea2021002

## Source details

**Canonical URL:** [1hkgea2021002](https://www.imf.org/-/media/files/publications/cr/2021/english/1hkgea2021002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2021/english/1hkgea2021002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2021/english/1hkgea2021002.pdf.json)

---

### Executive summary — overview and macroeconomic impact
- HKSAR is a small and open economy and a major international financial center with extensive linkages to Mainland China.
- Over the past two years, HKSAR’s economy and financial sector were adversely impacted by domestic social unrest, US-China tensions, and the global COVID-19 pandemic, resulting in an unprecedented two consecutive years of negative economic growth.
- Sound macroeconomic and prudential policies provided buffers to cope with the COVID-19 shock.
- The Linked Exchange Rate System (LERS) remains well-anchored and resilient, underpinned by large foreign exchange reserves.
- Key macroeconomic indicators:
  - GDP: decline of 1.2 percent in 2019; GDP declined by 8.9 percent in 2020 Q1 year-on-year; full year 2020 GDP declined by 6.1 percent.
  - Unemployment: rose from 3.3 percent at end-2019 to 6.6 percent at end-2020.
  - Growth began to recover in the second half of 2020 with a projected rebound in 2021 and 2022.

### Main macro-financial vulnerabilities
- Identified vulnerabilities:
  - Extensive linkages to Mainland China.
  - Stretched real estate valuations.
  - Exposure to shifts in global market and domestic risk sentiment.
- Key risk manifesting channels:
  - A prolonged COVID-19 pandemic.
  - A further slowdown in Mainland China economy.
  - A sharp rise in global risk-premia compounded by escalating US-China tensions.
  - A sharp housing market correction.
- Severe scenario: the four risks above could materialize simultaneously and amplify each other’s effects.

### Bank solvency and stress test findings
- Banking sector condition:
  - Remained well capitalized and profitable; NPL ratios remained low during the COVID-19 shock period.
- Solvency stress test outcomes (pre-COVID-19 data):
  - Even under a severely adverse scenario, banks’ average CET1 ratio decreases by about 4.5 percentage point to 11.5 percent in 2022—remaining well above regulatory requirements for all banks in the sample.
  - Average CET1 ratio in sample: 15.9 percent in 2019 → 11.5 percent in 2022 (adverse scenario).
  - IRB banks: CET1 ratio from 16.4 percent in 2019 to 11.7 percent in 2022 (decline of 4.6 percentage points).
  - Standardized banks: CET1 ratio from 14.4 percent in 2019 to 10.8 percent in 2022.
  - Average leverage ratio: declines from 7.8 percent in 2019 to 6.7 percent in 2022.
- Concentration risks:
  - Assuming default of the five largest NFC exposures in each bank, the average CET1 ratio decline would be close to 6.5 percentage points.
  - Non-financial corporate debt-at-risk and default risk could increase significantly under an adverse scenario, particularly in the Mainland China real estate sector.
- Recommendation:
  - Periodically stress test banks’ large exposures separately from their total loan books, considering potential cash flows from collateral in case of default of large borrowers. — ST, HKMA

### Liquidity risk analysis and stress test findings
- Regulatory liquidity indicators (as of 2020Q3 / June 30, 2020 where specified):
  - Average LCR for Category 1 institutions: just under 157 percent (2019YE: 160 percent).
  - Average Liquidity Maintenance Ratio (LMR) for Category 2 institutions: about 56 percent (regulatory minimum: 25 percent).
  - NSFR ratio: around 134 percent (2019YE: around 132 percent; regulatory minimum: 100 percent).
- Cash-flow based stress test key findings:
  - Short-term contractual outflows within the first 4 weeks amount to about 82 percent of total assets (including current account deposits 38 percent of TA and derivatives outflows 22 percent of TA); excluding derivatives outflows, short-term contractual outflows are 60 percent of TA.
  - Contractual inflows in the first month amount to about 62 percent of total assets (derivatives inflows 22 percent of TA; readily monetizable debt securities 17 percent of TA).
  - Cumulated contractual maturity mismatch over the first month: 25 percent of total assets; over the first 6 months mismatch increases to 31 percent.
- Horizon-specific resilience (CFLST results):
  - 7-day horizon: even under the most severe scenario CCBC = 15.1 percent of total banking assets of the sample; no bank with negative CCBC across 7-day scenarios.
  - 1-month horizon: CCBC average ~12 percent under most scenarios; under the most severe scenario CCBC = 10.4 percent (system-wide); only one foreign branch faces negative CCBC in the most extreme scenario (shortfall 1 percent of TA).
  - 3-month horizon: CCBC decreases substantially; under the most extreme severe market run-offs scenario system CCBC = 4.9 percent (local banks CCBC = 6.2 percent; foreign branches CCBC = -0.7 percent); up to 9 banks with negative CCBC under most severe scenarios.
- Structural observations:
  - Local banks: strong deposit funding base; loans to non-banks dominate assets.
  - Foreign branches: higher reliance on unsecured interbank funding and corporate deposits; greater USD funding share.
- Recommendations (selected):
  - Integrate all bank liquidity stress tests, streamline reporting, and enhance monitoring of liquidity position of foreign branches that appear more vulnerable. — MT, HKMA
  - Analyze early redemption risk for time deposits and banks’ measures to address such early redemption risk (thematic review). — ST, HKMA
  - Incentivize banks to diversify funding sources, lengthen and stagger funding tenors. — MT, HKMA
  - Support development of broad and deep repo markets and safe assets (e.g., a covered bond regime) to reduce shortage of HQLA. — MT, HKMA

### Interconnectedness, contagion, and cross-border analysis
- Balance sheet and market-based contagion analyses:
  - HKSAR banking sector among highest levels of cross-border exposures globally; total cross-border claims exceeded liabilities by close to 30 percent in 2019.
  - Cross-border contagion (balance-sheet approach):
    - Scenario 1 (credit shock): average impact of other banking systems’ failure on HKSAR ≈ loss of 13 percent of initial capital (Index of Vulnerability).
    - HKSAR hypothetical failure → average loss of about 6 percent of capital for other systems (Index of Contagion).
    - Adding funding shock (Scenario 2) increases HKSAR average losses by less than six percent of base losses.
  - Market-based contagion (Diebold-Yilmaz methodology):
    - D-SIBs in HKSAR are key recipients of spillovers from G-SIBs; HKSAR banks generally net recipients of shocks from G-SIBs in the US and Euro Area.
- Domestic interconnectedness:
  - Large fund flows from local banks to foreign branches and from foreign branches to NBFIs.
  - For sample of largest banks: foreign branches receive >53 percent of local banks’ intra-financial system lending; foreign branches’ intra-financial lending: 64 percent goes to NBFIs.
  - Interbank lending maturities: >50 percent of total domestic interbank lending maturity ≤ one month; >70 percent ≤ three months.
- Data gaps and NBFI build-up:
  - Lack of who-to-whom exposures across banking, NBFIs, corporates, households, government, and rest of world.
  - NBFI loans-to-GDP ratio increased by close to 20 percentage points since 2012, reaching 34 percent recently.
- Recommendations:
  - Enhance oversight over banking groups with both foreign branches and local subsidiaries in HKSAR; heighten monitoring of liquidity risk at group and entity level. — ST, HKMA
  - Enhance authorities’ data collection of intersectoral claims across major sectors, with focus on NBFIs, to develop an aggregate balance sheet approach. — ST, FSTB/HKMA/SFC/IA

### Banking system structure, funding, and market structure metrics
- Banking and financial sector scale and composition (selected exact figures and ratios):
  - Banking system assets ≈ 950 percent of GDP.
  - Total assets of banks (2020Q3): HK$ 26,080bn (multiples of GDP 9.5).
    - Local banks - domestic: 10 banks, Total assets HK$ 7,037bn (multiples of GDP 2.6).
    - Local banks - subsidiaries of foreign banks: 46, Total assets HK$ 8,990bn (multiples of GDP 3.3).
    - Foreign bank branches: 137, Total assets HK$ 10,053bn (multiples of GDP 3.7).
    - Retail banks: 34, Total assets HK$ 17,012bn (multiples of GDP 6.2).
    - Non-retail banks: 159, Total assets HK$ 9,068bn (multiples of GDP 3.3).
  - Asset and wealth management AUM (2019): HK$ 28,769bn (multiples of GDP 10.0).
  - Hong Kong stock exchange market capitalization: US$ 6.1 trillion.
- Funding structure (September 2020):
  - Customer deposits: 57 percent of total funding (72 percent for locally incorporated banks; 33 percent for foreign branches).
  - Funds from banks: 21 percent.
  - Debt securities: 6 percent.
  - Equity and other liabilities: 16 percent.
- Currency denomination of liabilities (September 2020):
  - Local banks: HKD 56, USD 31, RMB 5, other foreign currency 7.
  - Foreign banks: HKD 21, USD 61, RMB 7, other foreign currency 11.
- Recent credit and debt dynamics:
  - Total bank lending: increased by 6.7 percent in 2019 and by 1.2 percent in 2020.
  - Total NFC debt: from 202 percent of GDP in 2014 to 257 percent in 2020Q3.
  - Household debt: from 65 percent of GDP in 2014 to 88 percent in 2020Q3.
  - Average loan-to-deposit ratio: stable around 75 percent since 2018.

### Corporate sector vulnerability analysis and debt-at-risk
- Corporate debt and exposures:
  - Non-financial corporate debt: from 127 percent of GDP in 2009 to 243 percent in 2020Q2.
  - HKSAR banking sector exposures to Mainland China’s non-banks: HK$ 6.5 trillion by end-September 2020 (about 237 percent of GDP).
- Debt-at-risk and default projections (adverse scenario):
  - Adverse scenario includes a 3.77-ppt increase in borrowing interest rate and combined earnings, interest rate, and RMB/USD exchange rate shocks.
  - Aggregate PD (ICR-mapped) results:
    - Aggregate PD would increase to 8.3 percent of the overall bank corporate loan book (cumulative new defaults over two years) from 2.9 percent in 2019 and 3.6 percent in the baseline.
    - Combined shocks increase new defaults in the MC-oriented real estate sector from 1.1 percent in the baseline to 7.9 percent in the adverse scenario.
  - BuDA model:
    - Projects median PD rising in the near term under baseline and rising rapidly in the adverse scenario, approaching GFC levels but remaining below AFC peaks.
    - Sectorally, BuDA implies significantly higher default risk for MC-oriented real estate in the adverse scenario.
- Key caveats:
  - Analyses include only listed firms; small firms excluded may face higher stress.
  - Debt-at-risk is a vulnerability indicator and should not be directly equated with bank losses.
- Policy recommendations:
  - Continue to closely monitor credit risks facing MC borrowers and ensure banks’ internal risk models for high-risk MC borrowers are sufficiently forward-looking. — ST, HKMA
  - Consider incorporating non-financial corporates in the stress testing framework and communicating key findings regularly in HKMA’s Monetary and Financial Stability Report. — MT, HKMA

### LIBOR transition risks (Box 1) — exposures and stress implications
- Timing:
  - USD LIBOR rates will cease publication mid-2023; USD one- and two-week tenors and all non-USD LIBOR rates will cease at end-2021.
- Exposures (as of September 2020):
  - Almost 40 percent of banks' assets are referenced to LIBOR or HIBOR.
  - 5 percent of banks' liabilities (mostly LIBOR) are referenced to LIBOR or HIBOR.
  - Off-balance sheet derivative exposures referenced to LIBOR amount to additional 120 percent of HKSAR banking assets; two-thirds mature after end-2021; 93 percent lack adequate fallback provisions.
- Stress-test implication:
  - Under a scenario of increased interest rate volatility amounting to a 200bps shock to HKD and USD interest rates, local banks would experience a 1.4 percent decline in their total capital ratio, absorbed by existing capital buffers.
- Regulatory stance and recommendation:
  - HKMA advised banks to address legal, operational and market risks early, to cease issuance of LIBOR-linked products after June 2021, and to adopt fallback provisions; authorities should continue to monitor the transition vigilantly.

### Policy recommendations — selected and prioritized
- Integrate all bank liquidity stress tests, streamline reporting, and enhance monitoring of liquidity position of foreign branches that appear more vulnerable. — MT, HKMA
- Analyze early redemption risk for time deposits and banks’ measures to address such early redemption risk in a thematic review. — ST, HKMA
- Incentivize banks to diversify funding sources, lengthen and stagger funding tenors. — MT, HKMA
- Support development of broad and deep repo markets and safe assets (e.g., a covered bond regime) to increase HQLA availability. — MT, HKMA
- Continue close monitoring of credit risks facing MC borrowers and ensure internal risk models are forward-looking. — ST, HKMA
- Consider incorporating non-financial corporates in the stress testing framework and communicate key findings regularly. — MT, HKMA
- Enhance data collection of intersectoral claims across banking, NBFIs, non-financial corporates, households, government, and rest of world, focusing on NBFI claims. — ST, FSTB/HKMA/SFC/IA
- Enhance oversight over banking groups with both foreign branches and local subsidiaries and heighten monitoring of liquidity risk at group and entity level. — ST, HKMA
- Periodically stress test banks’ large exposures separately from total loan books, considering collateral cash flows. — ST, HKMA

### Risk Assessment Matrix — selected high-risk items
- Global risks:
  - Unexpected shift in the Covid-19 pandemic — Likelihood: Medium; Impact: Medium.
  - Accelerating de-globalization — Likelihood: Medium; Impact: High.
  - A sharp rise in global risk premia exposing financial and fiscal vulnerabilities — Likelihood: Medium; Impact: High.
- Domestic risks:
  - A further slowdown in China — Likelihood: Medium; Impact: High.
  - A sharp housing market correction and decline in demand — Likelihood: Medium; Impact: High.

*Source: EXECUTIVE SUMMARY (1hkgea2021002)*

### EXECUTIVE SUMMARY __________________________________________________________________________________ 6

### EXECUTIVE SUMMARY

### Overview
- Hong Kong SAR (HKSAR) is a small and open economy, and a major international financial center with extensive linkages to Mainland China.
- Over the past two years, Hong Kong SAR’s economy and financial sector were adversely impacted by domestic social unrest, US-China tensions, and the global COVID-19 pandemic, resulting in an unprecedented two consecutive years of negative economic growth.
- Sound macroeconomic and prudential policies have provided HKSAR with important buffers to cope with the unprecedented COVID-19 shock and future shocks.
- HKSAR’s exchange rate mechanism, the Linked Exchange Rate System (LERS), remains well-anchored and resilient to shocks, underpinned by large foreign exchange reserves.

### Macroeconomic impact and key indicators
- Following a GDP decline of 1.2 percent in 2019, with the onset of the pandemic in early 2020 the GDP declined by 8.9 percent in 2020 Q1 compared to a year earlier.
- For the full year 2020, GDP declined by 6.1 percent.
- HKSAR’s unemployment rate rose from 3.3 percent at end-2019 to 6.6 percent at end-2020, with retail and hospitality being the most affected sectors.
- Growth started to recover since the second half of 2020 and an economic rebound in 2021 and 2022 is projected as authorities took measures to support the economy and external demand started recovering.

### Main macro-financial vulnerabilities
- Extensive linkages to Mainland China.
- Stretched real estate valuations.
- Exposure to shifts in global market and domestic risk sentiment.
- Risks can manifest from:
  - A prolonged COVID-19 pandemic,
  - A further slowdown in Mainland China economy,
  - A sharp rise in global risk-premia compounded by escalating US-China tensions,
  - A sharp housing market correction.
- In a severe scenario, the four risks above could materialize at the same time and amplify each other’s effects.

### Bank solvency and stress test findings
- The banking sector remained well capitalized, profitable, and nonperforming loan ratios remained low during the COVID-19 shock period.
- Stress tests, carried out with pre-COVID-19 data, indicate that bank solvency is resilient to severe macrofinancial shocks thanks to the initial high capital buffers and high pre-shock profitability.
- Even under a severely adverse stress scenario, banks’ average CET1 ratio decreases by about 4.5 percentage point to 11.5 percent in 2022—remaining well above regulatory requirements for all banks in the sample.
- HKSAR banks have sizeable concentrated credit exposures:
  - Assuming the default of the five largest non-financial corporate (NFC) exposures in each bank, the average CET1 ratio would decline by more than 6 percentage points.
  - Non-financial corporate debt-at-risk and default risk could increase significantly under an adverse scenario, particularly in the Mainland China real estate sector.
- Recommendation: stress testing of banks’ large exposures separately from their total loan books by HKMA.

### Liquidity risk analysis and stress test findings
- The liquidity stress tests revealed overall high-stress resilience and a comprehensive reporting framework in HKSAR.
- The local banking system has sufficient liquidity buffers to cover the outflows in the scenarios, although the contractual maturity of the deposit base is short.
- The Covid-19 pandemic did not lead to elevated liquidity risk of the banking system. This was partly due to a sufficient pre-crisis liquidity risk bearing capacity and partly due to measures taken by the HKMA.
- The HKMA enhanced the provision of systemic liquidity and pro-actively communicated the usability of the LCR as buffer to absorb potential COVID-19 related liquidity shocks.
- The HKMA approach to liquidity stress testing is well advanced and comprehensive, aligned with international standard.
- Some foreign branches are more vulnerable to liquidity stress due to:
  - Higher reliance on unsecured interbank funding,
  - Inability to tap the local deposits base.
- Recommendation: integrate all bank liquidity stress tests, streamline reporting, and enhance monitoring of liquidity position of foreign branches that appear more vulnerable.

### Interconnectedness and contagion
- Cross-border and domestic intra-financial system lending data suggest that foreign branches in HKSAR source funding from local banks and extend significant amounts of lending to other financial entities: NBFIs in HKSAR and other banking systems abroad.
- Interconnectedness patterns suggest that foreign branches could be a source of spillovers to other financial sectors both at home and abroad.
- Recommendation: enhance oversight over banking groups that have both foreign branches and local subsidiaries in HKSAR; heighten monitoring of liquidity risk at the group and entity level for banks that operate with multiple group entities.

### Policy recommendations (selected from Table 1)
- Integrate all bank liquidity stress tests, streamline reporting, and enhance monitoring of liquidity position of foreign branches that appear more vulnerable. — MT, HKMA
- Analyze the early redemption risk for time deposits and banks measures to address such early redemption risk in the form of a thematic review. — ST, HKMA
- Take measures to incentivize banks to diversify funding sources, lengthen and stagger funding tenors. — MT, HKMA
- Support the development of funding sources that are more likely to be available also under stress, such as broad and deep repo markets. — MT, HKMA
- Support the development of safe assets to reduce the shortage of high-quality liquid assets in HK (e.g. a covered bond regime). — MT, HKMA
- Continue to closely monitor the credit risks facing MC borrowers and to ensure that the internal risk models used to determine the capital charge for potential high-risk MC borrowers are sufficiently forward-looking. — ST, HKMA
- Consider incorporating non-financial corporates in the stress testing framework and communicating key findings in the HKMA’s Monetary and Financial Stability Report on a regular basis. — MT, HKMA
- Enhance authorities’ data collection of intersectoral claims across major sectors of the economy (banking, NBFIs, non-financial corporates, households, government, and the rest of the world) with a focus on claims of the NBFI, to better gauge the importance of intersectoral linkages and develop an aggregate balance sheet approach in systemic risk assessment. — ST, FSTB/HKMA/SFC/IA
- Enhance oversight over banking groups that have both foreign branches and local subsidiaries in HKSAR to improve the authorities’ ability to monitor and mitigate risks, given their potential for contagion. — ST, HKMA
- Enhance monitoring of liquidity risk both at the group and entity level for banks that operate with multiple group entities in HKSAR to ensure that banks adhere to internal risk management practices. — ST, HKMA
- Periodically stress tests banks’ large exposures separately from their total loan books, considering potential cash flows from collateral in case of default of large borrowers. — ST, HKMA

*Source: EXECUTIVE SUMMARY (1hkgea2021002)*

### 2.      Despite the economic contraction, total bank lending increased by 6.7 percent in 2019,

### 1hkgea2021002 - 2.      Despite the economic contraction, total bank lending increased by 6.7 percent in 2019,

### Recent bank credit and sectoral debt developments
- Total bank lending increased by 6.7 percent in 2019 and by 1.2 percent in 2020.
- The slowdown in bank credit growth in 2020 was largely driven by the moderation in trade finance credit.
- Recent modest growth followed a peak credit growth of 16 percent in 2017.
- Total NFC debt increased from 202 percent of GDP in 2014 to 257 percent in 2020Q3.
- Household debt increased from 65 percent of GDP in 2014 to 88 percent in 2020Q3.
- The increase in debt led to a widening of the credit-to-GDP gaps in 2020, with much of the recent increase driven by the significant decline in GDP.
- Customer deposits remained the primary source of bank funding and grew at a moderate pace in 2020, with greater contributions from KSD deposits.
- The average loan-to-deposit ratio has been stable around 75 percent since 2018.

### Policy measures in the context of COVID-19
- Authorities adopted measures beginning in February 2020 to support the economy and maintain financial stability, including increasing market liquidity, supporting businesses and households, and releasing buffers in the banking system.
- HKMA reduced the issuance size of Exchange Fund Bills in early April 2020 to increase overall HKD liquidity in the interbank market.
- Following the Federal Reserve’s 50-basis points rate cut on March 4, HKMA adjusted the Base Rate from 2.0 to 1.5 percent.
- After a further Fed Funds rate reduction by 100 basis points, HKMA adjusted the Base Rate to 0.86 percent; the Base Rate reached 0.5 percent around mid-May.
- To improve USD liquidity, HKMA introduced the temporary “US Dollar Liquidity Facility” related to the Federal Reserve’s FIMA Repo Facility, offering USD liquidity to HKSAR banks via 7-day repos for a total volume of US$10 billion.
  - Take-up of the USD repos by banks has been limited; the facility has not been used since July 2020.
- Pre-approved Principal Payment Holiday Scheme: banks offered a six-month moratorium on principal repayments for qualifying SMEs (those with less than HKD800 million annual turnover).
  - Up to end-March 2021, more than 62,000 requests were made for moratoria and other relief measures involving loans totaling HK$ 770 billion (total domestic non-financial corporate debt HKD 3 trillion).
- Special 100% Loan Guarantee scheme for SMEs: loans distributed via private banks and guaranteed by a subsidiary of the Hong Kong Mortgage Corporation.
  - Against the total commitment of HK$ 50 billion through March 31, 2021, around 30,000 applications were made for loans totaling around HK$ 48 billion.
- Employment Support Scheme: wage subsidies totaling HK$ 80 billion or 2.8 percent of GDP.
- Cash payouts to permanent residents aged 18 or above: 10,000 HKD per person, estimated to cost HKD 71 billion or 2.5 percent of GDP.
- Household loan relief: through September 15, 2020 more than 19,000 customers benefited from mortgage/loan relief involving total credit of HK$ 35 billion (total outstanding residential mortgages HK$ 1.4 trillion, outstanding credit card debt HK$ 130 billion).
- Countercyclical capital buffer (CCyB) reductions:
  - From 2.5 to 2 percent in October 2019.
  - From 2 percent to 1 percent in March 2020.
- In April 2020, HKMA lowered the regulatory reserve by 50%.
- HKMA encouraged banks to use liquidity buffers in excess of LCR and LMR requirements.

### Banking sector recent performance and indicators
- Banking system remained profitable in aggregate in 2020, with ample capital and liquidity buffers.
- Aggregate pre-tax operating profit-to-asset ratio for all locally incorporated banks fell by about 29 percent from its 2019 level to 0.84 percent (annualized) during Q1-Q3 2020.
- 3-month HIBOR declined by about 180 basis points from 2019YE to September-end 2020.
- Net interest margin (NIM) stayed at just under 1.3% at the end of Q3, a 23 percent decline from its 2019YE level.
- Flow of loan loss provisions for locally incorporated banks almost doubled in 2020 compared to the preceding decade’s average, but stood at only 0.1% of total assets as of 2020Q3 (annualized).
- Non-performing loan (NPL) stock ratio for all retail banks increased from just over 0.4 percent of total loans at 2019YE to 0.8 percent at 2020YE.
  - Non-performing loan stock ratio for the larger group of all authorized institutions was 0.9 percent at 2020YE.
- Cost-to-income ratio for locally incorporated banks increased from just under 40 percent at 2019YE to 42 percent as of Q2 2020.
- Banking system achieved a 0.8 percent return on assets (ROA) during Q1-Q3 2020 (annualized), down by about 20 basis points from its 2019 level.
- Tier 1 capital ratio was 18.3 percent as of Q3 2020, slightly lower than 18.5 percent at 2019YE.
- Liquidity indicators as of 2020Q3:
  - Average LCR for Category 1 institutions: just under 157 percent (2019YE: 160 percent).
  - Average Liquidity Maintenance Ratio (LMR) for Category 2 institutions: about 56 percent (regulatory minimum: 25 percent).
  - NSFR ratio: around 134 percent (2019YE: around 132 percent; regulatory minimum: 100 percent).
- Customer deposits comprised 57 percent of total liabilities as of 2020Q3 (72 percent for locally incorporated banks and 33 percent for foreign branches).

### Financial system landscape and structure
- Banking system assets equivalent to around 950 percent of GDP.
- Life insurance penetration and size:
  - Insurance premiums-to-GDP ratio: about 18 percent in 2019.
  - Life premiums per capita: around US$9,000 in 2019.
- Hong Kong stock exchange market capitalization: US$ 6.1 trillion (world’s fifth largest).
- Mainland China-related companies:
  - Account for around half of listed companies in HKSAR and represent 81 percent of total capitalization.
- Total assets under management (AUM) of asset and wealth management companies: 1,000 percent of GDP in 2019.
  - Investors outside HKSAR owned 64 percent of AUM in 2019.
  - 387 licensed corporations and registered institutions established by Mainland China-related groups active in HKSAR.
- Financial sector structure (2020Q3 or latest available data):
  - Banks: 193 institutions, Total assets HK$ 26,080bn, Multiples of GDP 9.5
    - Local banks - domestic: 10, Total assets HK$ 7,037bn, Multiples of GDP 2.6
    - Local banks - subsidiaries of foreign banks: 46, Total assets HK$ 8,990bn, Multiples of GDP 3.3
    - Foreign bank branches: 137, Total assets HK$ 10,053bn, Multiples of GDP 3.7
    - By business nature:
      - Retail banks: 34, Total assets HK$ 17,012bn, Multiples of GDP 6.2
      - Non-retail banks: 159, Total assets HK$ 9,068bn, Multiples of GDP 3.3
  - Authorized Insurance Companies (2019): 163, Total assets HK$ 3,872bn, Multiples of GDP 1.4
    - Life: 49, Total assets HK$ 3,562bn, Multiples of GDP 1.2
    - Non-Life: 84, Total assets HK$ 166bn, Multiples of GDP 0.06
    - Reinsurance: 17, Total assets HK$ 144bn, Multiples of GDP 0.05
  - Asset and wealth management business (2019) AUM: HK$ 28,769bn, Multiples of GDP 10.0
    - Asset management business: HK$ 17,898bn, Multiples of GDP 6.2
    - Fund advisory business: HK$ 1,853bn, Multiples of GDP 0.6
    - Private banking and private wealth management business: HK$ 7,774bn, Multiples of GDP 2.7
    - SFC-authorized REITs: HK$ 289bn, Multiples of GDP 0.1
    - Assets held under trusts attributable to non-licensed corporations and non-registered institutions: HK$ 955bn, Multiples of GDP 0.3

### Banking system segmentation and business models
- Local banks:
  - Strong local deposit funding base.
  - Engage in financial intermediation mostly in HKSAR with a significant portion of Mainland China-related loans (about one-third of their loans).
  - Invest largely in loans to non-banks (46 percent of total assets).
  - Approximately 30 percent of non-bank lending are to households (mortgages and consumer loans); the remainder to corporates.
- Foreign branches:
  - Rely more on interbank borrowing from local banks and on corporate deposits.
  - Interbank lending constitutes a higher share of foreign branches’ assets compared to local banks (34 vs. 16 percent).
  - Non-bank loan books of branches consist almost entirely of loans to corporate borrowers many of which are located outside HKSAR.
  - Branches are allowed limited retail business; if local retail lending grows beyond a certain level, HKMA requires subsidiarization in HKSAR, bringing the bank within local capital regulations.

*Source: IMF staff compilation from BIS, HKMA, IMF Staff Computations, and national authorities as presented in the content unit.*

### 17.      The banking system has significant loan exposures to Mainland China. Although  only  8

### 17.      The banking system has significant loan exposures to Mainland China. Although  only  8 percent of loans are to borrowers located in Mainland China, close to 39 percent of loans are Mainland China-related loans

### Loan exposures and loan-book composition
- 8 percent of loans are to borrowers located in Mainland China.
- Close to 39 percent of loans are Mainland China-related loans.
- Mainland China-related lending by bank type:
  - Close to 51 percent of foreign branches’ loan books are Mainland China-related.
  - About one third for local banks are Mainland China-related.
- Geographic location of borrowers by bank type:
  - Local banks: 84 percent of loans are to borrowers located in HKSAR;  five percent of borrowers in Mainland China.
  - Foreign branches: 62 percent of loans are to borrowers located in HKSAR.
- Definition note: Mainland China-related loans cover a broader group of borrowers which derive a majority of their revenues from Mainland China-related activities.

### Balance sheet structure (selected asset and liability shares, September 2020)
- Local Banks (Asset share / Liability share):
  - Total Loans to Non-banks: 46% asset share; 72% liability share (Customer Deposits).
  - Cash and Due from Banks: 16% asset share; 7% liability share (Due to banks).
  - Securities Holdings: 23% asset share; 1% liability share (Debt Securities Outstanding).
  - Fixed Assets: 1% asset share; 20% liability share (Equity and Other Liabilities).
  - Other Assets: 15% asset share; 8% liability share.
  - Total Assets: 100%; Total Liabilities: 100%.
- Foreign Branches (Asset share / Liability share):
  - Total Loans to Non-banks: 37% asset share; 33% liability share (Customer Deposits).
  - Cash and Due from Banks: 34% asset share; 44% liability share (Due to banks).
  - Securities Holdings: 20% asset share; 13% liability share (Debt Securities Outstanding).
  - Fixed Assets: 0% asset share; 10% liability share (Equity and Other Liabilities).
  - Other Assets: 8% asset share.
  - Total Assets: 100%; Total Liabilities: 100%.
- Aggregate observation: Loans to non-banks dominate local banks’ assets; foreign branches hold more cash and interbank positions.

### Funding structure of the overall HKSAR banking system (September 2020)
- Customer deposits account for 57 percent of total funding.
- Funds from banks account for 21 percent of total funding.
- Debt securities account for 6 percent of total funding.
- Equity and other liabilities (including reserves) account for 16 percent of total funding.
- Asset side composition (All Banks):
  - Total Loans to Non-banks: 42% asset share.
  - Cash and Due from Banks: 23% asset share.
  - Securities Holdings: 22% asset share.
  - Fixed Assets: 1% asset share.
  - Other Assets: 12% asset share.
  - Total Assets: 100%; Total Liabilities: 100%.
- Balance sheet structure remained relatively stable from 2019YE to 2020Q3.

### Funding model differences between local banks and foreign branches
- Local banks:
  - Fund 72 percent of their balance sheet by deposits from non-banks.
  - Interbank funding accounts for 7 percent.
  - Debt securities issuance accounts for 1 percent.
  - Equity and other liabilities amount to 20 percent.
- Foreign branches:
  - Fund just under 33 percent of their balance sheet by deposits from non-banks.
  - Interbank funding accounts for 44 percent.
  - Debt securities issuance close to 13 percent.
  - Equity and other liabilities amount to 10 percent.
- Implication: Local banks are largely locally funded and redistribute excess funding to foreign branches via unsecured interbank loans; foreign branches also receive unsecured funding from overseas.
- Recent change: From 2019YE to 2020Q3, interbank funding declined by about one percentage point for foreign branches and increased by a similar amount for local banks.

### Currency denomination of liabilities (September 2020)
- Local Banks (in % of total liabilities):
  - in HKD: 56
  - in USD: 31
  - in RMB: 5
  - in other foreign currency: 7
  - Total: 100
- Foreign Banks (in % of total liabilities):
  - in HKD: 21
  - in USD: 61
  - in RMB: 7
  - in other foreign currency: 11
  - Total: 100
- Observation: Local banks are largely funded in HKD; foreign branches feature a larger share of USD funding.
- Stability: During 2020, the funding currency mix stayed mostly stable for both groups.

### Cross-border exposures and interconnectedness
- HKSAR’s banking sector has one of the highest levels of cross-border exposures in the world.
- Mainland China’s banking sector receives the largest share of HKSAR’s cross-border lending by far.
- BIS locational banking statistics: total cross-border claims exceeded liabilities by close to 30 percent in 2019.
- Since late 1999, HKSAR banks have been a net lender to other banking sectors abroad, increasing exposure to potential external shocks.
- Domestic financial connectedness: large fund flows from local banks to foreign branches and from foreign branches to nonbank financial institutions (NBFIs).

### Scope and methodology of the FSAP financial stability analyses
- Focus: resilience of the banking system (solvency and liquidity), domestic and cross-border interconnectedness; complemented by corporate sector vulnerability analysis.
- Coverage:
  - 12 local banks (including all six D-SIBs) and 11 foreign branches for various analyses.
  - Solvency stress test (ST) focused on subset of local banks accounting for about 93 percent of total consolidated assets of all local banks in HKSAR subject to HKMA capital regulation.
- Solvency ST:
  - Utilized a baseline scenario and an adverse scenario.
  - Generated three-year estimates for profit and losses, and the evolution of risk-weighted assets and capital.
- Liquidity analysis:
  - Assessed regulatory liquidity ratios: LCR and NSFR, and equivalents for Category 2 banks: LMR and CFR.
  - Conducted cash flow-based liquidity stress tests using supervisory data on contractual cash flows and security flows across multiple scenarios and time horizons.
- Interconnectedness and contagion analysis:
  - Mapped intra-financial linkages using exposure data from the largest 20 banks by HKSAR office basis assets (sample included 9 local banks and 11 foreign branches).
  - Cross-border interbank credit shock simulation (counterparty default) and funding shock simulation (withdrawal of funding counterparty) using BIS locational banking statistics.
  - Fire-sale asset shock used to simulate extreme funding shock.
  - Market-based contagion analysis using Diebold-Yilmaz (2014) methodology with daily equity return data through December 2020; sample included G-SIBs and five publicly traded HKSAR banks.
- Corporate sector vulnerability analysis:
  - Two scenario-based firm-level analyses:
    - debt-at-risk to assess overall corporate sector vulnerabilities.
    - bottom-up default analysis (BuDA) using market-based information and balance sheet indicators under an adverse macroeconomic scenario.

### Key vulnerabilities and risks identified
- Main macro-financial vulnerabilities: extensive linkages to Mainland China, stretched real estate valuations, and exposure to shifts in global market and domestic risk sentiment.
- Interconnectedness with Mainland China (high and rising):
  - Real sector ties (trade and tourism): Mainland China visitors historically account for over three quarters of tourism spending in HKSAR; tourism spending accounts for about 30 percent of total retail sale.
  - Direct credit exposures: close to 40 percent of banks’ total loan books comprise Mainland China-related lending (33 percent for local banks and 51 percent for foreign branches).
  - Financial services hub role: HKSAR is the main fundraising platform for Chinese firms and asset management services; most listed firms and bond issuers in HKSAR are from Mainland China.
    - Annual issuance of offshore non-bank Mainland China bonds increased to US$190billion in 2019 — more than 10 times the total issuance of US$18billion in 2014.
    - Non-bank Mainland China corporate bond outstanding totals US$662billion: most are short-term (2-3 years) and only one-third have investment-grade ratings.
- Potential spillovers from Mainland China shocks:
  - A significant slowdown in China could reduce trade flows and tourism, lower corporate profits and investment, raise unemployment, weaken consumption, and increase bank losses on Mainland China-related and other corporate loans.
  - A slowdown would also increase defaults among corporate bond issuers and raise credit risk for bondholders in HKSAR.
- Real estate risks:
  - Housing valuations appear stretched; real house prices exceeded fundamental-based valuations by about 33 percent in 2020Q2.
  - Nominal wages grew by only about 23 percent between early 2014 and September 2020.
  - Housing production has fallen short of the Long-Term Housing Strategy supply target by around 30 percent on average since 2015.
  - Real house prices stagnated between March and December 2020 (up 0.2 percent in nominal terms).
  - House-price downside risk examples:
    - One-year-ahead HaR 5% = -3%
    - One-year-ahead HaR 10% = -0.8%
    - Three-year-ahead HaR 5% = -9.4%
    - Three-year-ahead HaR 10% = -6.9%
  - Risk channel: continued overvaluation or a sharp increase in interest rates could produce a sharp downside correction, higher mortgage payments, larger declines in prices, negative feedback on households’ and banks’ balance sheets, and adverse effects on financial stability and the HKSAR economy.
- External risk sensitivity:
  - As a global financial center with an open economy, HKSAR is vulnerable to shifts in global risk sentiment, including tighter global liquidity conditions and US-China tensions.
- LIBOR transition risk:
  - Discontinuation of LIBOR by end-2021 implies transition risk, but the risk appears to be well-managed.
  - HKMA advised banks to address legal, operational and market risks early, and to cease further use of LIBOR as a benchmark starting from June 2021.

*Source: IMF staff calculations and HKMA data as presented in the IMF technical note.*

### Box 1. LIBOR Transition Risks for the Hong Kong SAR Financial  Sector

### Box 1. LIBOR Transition Risks for the Hong Kong SAR Financial Sector

### Overview of the transition and timing
- The London Interbank Offered Rate (LIBOR) will be discontinued as a benchmark interest rate for adjustable-rate financial products.
- USD LIBOR rates will cease publication mid-2023, while USD one- and two-week tenors and all non-USD LIBOR rates will cease at end-2021.
- LIBOR is being discontinued because it is now based on too few transactions following past LIBOR market manipulation by several large international banks.

### Exposures in the HKSAR financial sector (as of September 2020)
- Banks, funds, and insurers: HKSAR banks have considerable LIBOR and HIBOR exposures, while funds and insurers have negligible exposures.
- On- and off-balance sheet exposures:
  - Almost 40 percent of banks' assets are referenced to LIBOR or HIBOR.
  - 5 percent of banks' liabilities (mostly LIBOR) are referenced to LIBOR or HIBOR.
  - Off-balance sheet derivative exposures referenced to LIBOR (mostly interest and cross-currency swaps) amount to additional 120 percent of HKSAR banking assets.
  - Two-thirds of those derivative exposures mature after end-2021.
  - 93 percent of those derivative exposures are not yet covered by adequate fallback provisions in underlying contracts.
- Expectation: These exposures are expected to decline rapidly over 2021, but residual positions could entail legal uncertainties and become less liquid or experience volatile value changes as cessation dates approach.
- Note on terminology: Exposure here refers to the notional of instruments that use these benchmarks as reference rates, not to market value sensitivity to fluctuations in these rates.

### Market structure, alternative reference rates, and regulatory stance
- Alternative reference rate identified: In 2019, the Treasury Markets Association identified the HKD Overnight Index Average (HONIA) as the alternative reference rate (ARR) to the HIBOR.
- Coexistence and computation:
  - No panel instabilities or conduct problems observed in the HIBOR benchmark; HIBOR and HONIA are expected to coexist in HKSAR markets under a multiple rate approach with market participants free to choose benchmarks.
  - Currently, no change in the computation of the HKMA’s base rate for monetary policy is required.
- HKMA guidance and measures:
  - The HKMA advised banks to address legal, operational and market risks early and appropriately and is currently reviewing banks’ LIBOR transition plans.
  - The HKMA has encouraged banks to no longer issue LIBOR-linked products after June 2021.
  - Market participants are urged to be prepared to provide products linked to ARR and to start including fallback provisions for new contracts.
  - If needed, the HKMA has tools to discourage continued use of LIBOR, such as larger collateral haircuts on LIBOR-referencing products in its liquidity facility framework or higher haircuts in the LCR framework.
- Derivatives market and overseas compliance:
  - Issues around a lack of derivative markets for ARRs may be of less concern for HKSAR because HIBOR provides a full-term curve.
  - Remaining concern: compliance of HIBOR with relevant EU law to ensure HIBOR can continue to serve as a benchmark for EU banks (the transition period for third-country benchmarks expires by end-2023 under current EU law).

### Risk assessment and stress-test implication
- LIBOR volatility during the transition may feed into HIBOR volatility, but related risks appear manageable for local banks in HKSAR.
- Macroprudential backstops:
  - Under a scenario of increased interest rate volatility amounting to a 200bps shock to HKD and USD interest rates, local banks would experience a 1.4 percent decline in their total capital ratio, which is easily absorbed by existing capital buffers.
  - The HKMA’s standby liquidity facility and its contingent term facility could provide additional protection, if needed.
- International coordination and market infrastructure:
  - Broad international efforts, including the FSB’s coordination role through its OSSG group, and ISDA’s new fallbacks for IBOR derivatives, in effect since early 2021, are supporting a smooth transition.
- Overall assessment and recommendation:
  - While overall benchmark transition risk is assessed to be limited, HKSAR authorities should continue to monitor the situation vigilantly.

*Sources: HKMA and IMF Staff Calculations.*

### 40.      Changes in net income during the scenario are an important driver of changes in the

### 1hkgea2021002 - 40.      Changes in net income during the scenario are an important driver of changes in the

### Net income drivers and assumptions
- Net income components: i) net interest income, ii) net non-interest income, and iii) credit loss provisions.
- Estimation of net interest income accounts for changes in balance sheet size, net interest margins including the effect of funding shocks, and non-performing loans (NPLs). It is assumed that banks would not receive interest income from NPLs.
- Non-interest income assumptions:
  - Under the adverse scenario: net fee and commission income and other non-interest income decline by 30 percent from its initial level in the first year of the scenario, and recover by 15 percent in the third year.
  - Under the baseline scenario: non-interest income items decline by 25 percent in the first year, start recovering in the second year, and reach 90 percent of their initial level in the third year.
- Operational and administrative expenses assumed to grow in line with the overall size of banks’ balance sheets, carrying over the zero-growth floor assumption.
- Loss provisions estimated separately for standardized (STA) and IRB banks using NPL and probability of default (PD) estimates.

### Calculation of loss provisions for standardized banks
- Two main components: projection of new flow of NPLs and projection of loan provisioning rates.
- FSAP estimated satellite credit risk models to project ratio of stock of NPLs by loan category (corporate loans, mortgages, retail loans, and bank loans) and converted NPL stock ratios to flows.
- Provisioning rates:
  - Used banks’ historical realized loss-given-default (LGD) rates as proxies.
  - Calculated 4-quarter moving averages of quarterly realized LGDs and used the peak of the moving average series to calibrate stressed provisioning rates for the adverse scenario.
  - Average provisioning rates (adverse scenario sample):
    - Corporate loans: 58 percent.
    - Retail loans (mostly uncollateralized): 76 percent.
  - Mortgages:
    - Average LTV ratio for solvency ST sample around 39 percent at end of 2019; share of high-LTV mortgages at less than two percent of total mortgage books.
    - With low average LTVs, average provisioning rate for STA banks was 10 percent in the adverse scenario.
- Baseline scenario provisioning rates:
  - More recent LGD rates utilized with elevated LGDs in the first year, reverting to longer-term median values as the economy recovers in second and third years.
  - Average provisioning rates in the sample for baseline scenario:
    - Corporate loans: just over 35 percent.
    - Retail loans: 67 percent.
    - Mortgages: minimal.

### Calculation of loss provisions for IRB banks
- Based on Expected Loss (EL) approach: EL = PD x LGD x EAD, requiring point-in-time (PiT) PDs and LGDs.
- FSAP estimated satellite credit risk models to project PiT PDs for IRB exposure categories: corporate, mortgages, individual retail, and bank.
- LGD treatment:
  - Used banks’ historical realized LGD rates; calculated 4-quarter moving averages and used peak to calibrate stressed LGDs for the adverse scenario.
  - Through-the-cycle (TTC) LGDs used as floors for stressed LGDs in the adverse scenario.
  - Average LGDs (sample):
    - Corporate exposures: 48 percent.
    - Individual retail exposures: close to 82 percent.
  - Mortgages:
    - Ground-up loss rates using portfolio LTV distributions and simulated house prices, and TTC (downturn) LGDs provided by banks.
    - Average LGD for credit loss calculation was 12 percent in the adverse scenario.
- Baseline scenario LGDs: more recent LGD rates with elevated LGDs in first year, reverting to longer-term values as economy recovers.
  - Average LGD rates in sample (baseline):
    - Corporate exposures: around 31 percent.
    - Retail loans: 70 percent.
    - Mortgages: minimal losses due to increasing house prices and low starting LTVs.

### Changes in Risk-Weighted Assets (RWA)
- Stress test accounts for changes in RWA reflecting higher credit risk under adverse conditions.
- IRB banks:
  - RWA formula uses TTC PDs and downturn LGDs.
  - Changes in TTC PDs derived from PiT PDs using Δ TTC PD = λ * Δ PiT PD (with 0<λ<1).
  - Chosen λ values:
    - Corporate exposures: λ = 0.4.
    - Other exposures: λ = 0.2.
  - Downturn LGDs: used as reported by banks with no additional increases.
- STA banks:
  - Risk weights determined by loan risk ratings per HKSAR’s Banking (Capital) Rules (BCR).
  - Many corporate loans unrated and receiving 100 percent risk weight; average risk weights for corporate loan exposures already above 90 percent.
  - Limited room for increases in RWAs due to risk-weight shifting for STA banks compared to IRB banks.

### Long-term view of credit risk in HKSAR
- Aggregate NPL stock ratio increased substantially during the AFC but has been consistently low and stable since mid-2000s.
- Historical episodes:
  - GFC: HKSAR GDP growth declined from 6.5 percent in 2007 to 2.1 percent and -2.5 percent in 2008 and 2009, respectively. Aggregate NPL stock ratio of retail banks peaked at 1.5 percent in 2009Q2, increasing by about 0.6 percentage points from 2007 levels.
  - Since 2010: NPL ratio consistently below 1 percent.
  - Growth slowdown late 2015-2016: NPL ratio increased by less than 0.3 percentage points and peaked at 0.7 percent.
  - 2019: NPL ratios around 0.5 percent while GDP contracted by 1.2 percent.
- Mortgage underwriting and macroprudential measures:
  - Average LTV ratio of new mortgages declined from peak of 69 percent in 2002 to 46 percent in 2019Q3; increased to 57 percent in 2020Q3 partly due to expansion of Mortgage Insurance Programme starting 2019Q4.
  - Average LTV of banks’ stock of mortgage loans around 42 percent at 2020 year-end.
  - HKMA introduced a debt servicing ratio (DSR) cap in 2010 and applies “stressed DSR” limits assuming a 300-basis point increase in interest rates; stressed DSR must not increase beyond 10 percentage points above the base DSR limit.
  - As of 2020 year-end: base DSR limit stands at 50 percent and stressed DSR limit stands at 60 percent for first time borrowers whose income is mainly derived in HKSAR (40 percent and 50, respectively, for non-first time borrowers).
- Household balance sheet indicators:
  - Average household net worth-to-liabilities-ratio: 11.2 in 2019.
  - Safe assets-to-liabilities ratio: 2.88.
  - Wealth distribution skewed toward higher-income households; lower and medium-income households have higher DSRs and are more vulnerable to higher rates and income shocks.
- Corporate sector:
  - Corporate leverage upward trend since early 2000s; accelerated after GFC.
  - Interest coverage ratio (ICR) of all listed NFCs dipped below 2.50 at trough in 1998; remained around 4 at end-2019.
  - Higher leverage increases vulnerability to interest rate shocks despite average ICR buffer.

### PD and NPL projections for credit loss simulations
- Bank-level PD and NPL data available starting 2013; recent years show muted delinquencies due to absence of significant downturns.
- FSAP used longer-term aggregate NPL series to benchmark PD/NPL predictions.
- PD projections for IRB banks:
  - Models estimated using bank-level PiT PD history for each exposure category and benchmarked against aggregate NPL series.
  - GDP growth identified as most significant driver of corporate delinquencies.
  - Given high exposure to Mainland China-related loans, FSAP utilized a weighted average of HKSAR GDP growth and Mainland China GDP growth; weights chosen based on each bank’s relative share of Mainland China-related loans.
  - Predicted PDs under the adverse scenario:
    - Corporate PD predictions approach five percent in the second year and decline slowly in the third year as economy starts recovering.
    - Mortgage PDs peak at just over 2 percent in the second year and decline in the third year.
    - Retail PDs peak at around 3.5 percent in the second year.
  - Banks with relatively higher exposures to corporates and lower exposures to mortgages would experience higher credit losses and larger capital declines under an adverse environment.

*Source: 1hkgea2021002 - 40.      Changes in net income during the scenario are an important driver of changes in the*

### 53.      The FSAP took  a ratings-based approach  to simulating PDs for bank exposures. The

### 1hkgea2021002 - 53.      The FSAP took  a ratings-based approach  to simulating PDs for bank exposures. The

### Ratings-based PD simulation for bank exposures
- Methodology:
  - Used a ratings-based approach because IRB banks in the sample experienced zero PDs in their bank exposures during the limited granular data sample, preventing econometric modeling.
  - Utilized PDs based on simulated ratings of average banking sector exposures using Moody’s banks rating methodology linking initial rating, financial performance, and the macro profile of the country.
- Adverse scenario rating path and PD mapping:
  - Simulated downgrades: Aa in 2019 → Baa in 2020 → Ba in 2021 and 2022.
  - These letter ratings were mapped to the annual default rate observed in 2008 during the GFC.
  - Resulting adverse scenario PDs: 0.5 percent in 2020 and 2021; just over 1 percent in 2022.
- Baseline scenario rating path and PD mapping:
  - Simulated path: Aa in 2019 → A in 2020 and 2021 → upgraded to Aa in 2022.
  - For 2020, the letter rating was mapped to the annual default rate observed in 2008; 2021 and 2022 reverted to the longer-term average default rate.
  - Resulting baseline scenario PDs: 0.5 percent in 2020 and 0.1 percent in 2021 and 2022.

### NPL projections for STA banks
- Aggregate NPL model and sectoral behavior:
  - Estimated an NPL model using aggregate NPL stock ratio series to project STA banks’ NPL ratios.
  - Observed NPL stock ratios of STA banks were muted in recent years.
  - Corporate loan NPL ratio increase is higher than other loan categories.
- Predicted NPL stock ratio peaks (adverse scenario):
  - Corporate NPL stock ratio: peaks at around 5.5 percent.
  - Mortgage NPL stock ratio: peaks at around 3 (percent).
  - Retail NPL stock ratio: peaks at around 4.5 percent in the second year.
- Conversion from NPL stocks to NPL flows:
  - NPL flows are not directly observed; conversion requires assumptions on speed of removal of NPLs from balance sheets.
  - Assumption: banks remove only two thirds of the stock of NPLs from their balance sheets each year (to reflect reduced speed of dealing with NPLs during stress).
  - Once converted, average annual inflow of projected delinquencies are at roughly similar levels between IRB and STA banks.
- Loans to other banks (STA banks):
  - Applied same ratings-based approach as IRB banks; projected NPL flows matched to flows implied by IRB PDs.
- General provisions applied:
  - Average adverse scenario general provisions: 0.4 percent of loans in the first year and 0.3 percent of loans in the second year.
  - Reduced general provisions by 0.1 percent in the third year as economy recovers.

### Simulation of funding cost shocks and interest rate risk
- Funding sources and dataset:
  - Input dataset: bank-level quarterly funding composition and average funding cost data from early 2010s.
  - Funding categories: i) overnight (repricing) deposits, ii) term deposits repricing in one year or less, iii) non-deposit funding repricing in one year or less, iv) all funding repricing with maturities in more than one years.
  - Focused on HKD, USD, and RMB (jointly account for about 90 percent of all interest-bearing liabilities).
- Denomination structure (in percent of total interest-bearing assets/liabilities):
  - Interest-bearing Assets: in HKD 44, in USD 31, in RMB 12, in other significant foreign currencies 14, Total 100.
  - Interest-bearing Liabilities: in HKD 45, in USD 33, in RMB 12, in other significant foreign currencies 10, Total 100.
- Regression calibration:
  - Regressed quarterly average funding cost data on benchmark interest rates and included bank capital ratios.
  - Benchmark variables: 3-month HIBOR, 12-month HIBOR, 3-month USD LIBOR, 3-month HIBOR/LIBOR spread, and 12-month USD LIBOR.
  - 3-month HIBOR was the most stable explanatory variable across funding categories.
- Funding cost behavior:
  - Overnight HKD deposits roughly follow 3-month HIBOR but display low sensitivity; historically lowest average funding costs.
  - Term deposits display larger sensitivity: increase by approximately 25 basis points in the first year, stay elevated, and start declining towards the second half of the second year.
- Pass-through assumptions to asset yields:
  - Assumed 80 percent pass-through from changes in 3-month HIBOR and LIBOR rates to interest-bearing assets for the adverse scenario.
  - Under the baseline scenario assumed 60 percent pass-through reflecting improved economic environment and greater pricing power for banks.
- Impact on net interest margins:
  - Funding cost increases compress net interest margins; initial asset-side yields rise as benchmark rates increase but decline later as benchmark rates fall, producing profitability pressures.
  - Asset and liability side shocks and interest rate sensitive off-balance sheet positions captured via banks’ interest rate risk exposures.

### Stress test results and drivers
- Baseline scenario outcomes:
  - Average decline about one percentage point reduction in CET1 ratio during Year1.
  - Capital levels stabilize with quick return to positive growth in Year2; banks build up capital buffers in Year3.
  - Noted realized average capital levels in 2020 were mostly unchanged from 2019, better than simulated decline.
  - Realized 2020 GDP growth: -6.1 percent (better than the baseline scenario GDP of -7.5 percent used in simulations).
  - FSAP took conservative approach by not incorporating temporary borrower-support policies.
- Adverse scenario outcomes and resilience:
  - Even under the adverse scenario, system broadly resilient due to high initial capital buffers and pre-shock profitability.
  - All banks remain above regulatory minimum of 4.5 percent CET1 ratio plus capital conservation buffer of 2.5 percent.
  - Average CET1 ratio in sample: 15.9 percent in 2019 → 11.5 percent in 2022.
  - IRB banks: CET1 ratio decreases from 16.4 percent in 2019 to 11.7 percent in 2022 (a decline of 4.6 percentage points).
  - Standardized banks: CET1 ratio decreases from 14.4 percent in 2019 to 10.8 percent in 2022.
  - Average leverage ratio (Tier 1 capital/total assets as defined here): declines from 7.8 percent in 2019 to 6.7 percent in 2022.
- Main drivers of capital declines (adverse scenario):
  - Loan loss provisions (credit losses) are a primary driver due to increased defaults.
  - Increase in RWA reflecting higher credit risk, especially for IRB banks.
  - Declines in net interest margins (interest rate risk) contribute to a lesser extent.
  - Secured nature of exposures moderates losses (average mortgage LTV just under 40 percent in 2019).
  - Standardized banks have higher initial risk weights (e.g., average risk weights for corporate loan exposures already above 90 percent), so RWA increases are smaller for them.
- Market risk and interest rate risk impacts:
  - Market risk (MTM declines in securities holdings) causes modest CET1 declines: about 0.8 percentage points in the first year from spread widening, especially corporate bonds.
  - Securities portfolios’ higher quality composition (significant EFBNs and investment grade securities) moderates market losses.
  - Market risk losses partially reverse in years 2 and 3 as base interest rates remain low and spread widening reverses.
  - Interest rate risk: higher funding costs compress margins, but balanced interest rate sensitivities limit overall losses.

*PEOPLE’S REPUBLIC OF CHINA—HONG KONG SAR, INTERNATIONAL MONETARY FUND*

### 67.      The FSAP team also assessed the impact of an increase in interest spreads of banks’

### 67. The FSAP team also assessed the impact of an increase in interest spreads of banks’ securities portfolios

### Interest spread shock on securities portfolios
- Scenario: Stand-alone shock of 300 basis points increase in the spread on banks’ all non-sovereign securities portfolios.
- Scope: All non-sovereign securities in banks’ portfolios designated as held-for-trading (HFT) or fair value through other comprehensive income (FVTOCI) are marked to market (MTM) under the assumed spread widening.
- Dynamics under concurrent adverse shocks:
  - Spreads on securities increase significantly.
  - USD benchmark interest start declining in the first year of the simulation.
  - HKD benchmark rates start to decline towards the end of the second year of the simulation.
  - Declining benchmark rates reduce the negative price effect of spread widening, especially for banks with large USD securities portfolios.
- Impact:
  - The MTM effect of the widening for the full sample of banks is a decline of 1.9 percentage points in the average CET 1 ratio.
  - Without the buffering effect of declining benchmark interest rates, estimated MTM losses increase compared to those estimated under the solvency scenario, especially for banks with larger USD securities portfolios.
  - The total losses are modest compared to banks existing capital buffers.

### Concentration risk (credit exposures)
- Single-name concentration:
  - Method: Hypothetical default of the five largest non-financial corporate (NFC) exposures per bank, assuming zero recovery on the unsecured part of the exposure and full recovery of the value of any available collateral.
  - Impact: Assuming default of the five largest NFC exposures in each bank, the average CET1 ratio decline would be close to 6.5 percentage points of CET 1 capital ratio.
  - More severe collateral assumption: Zero recovery on unsecured part and a 50 percent haircut on collateral increases losses modestly by 0.5 percentage points.
  - Note: Low sensitivity to collateral recovery assumption due to many large exposures being backed by minimal collateral.
- Sectoral concentration:
  - Largest exposures of the banking system are to the real estate sector followed by the IT sector.
  - Impact: Default of all real estate-related borrowers that are among the top ten large exposures of each bank would reduce the average CET1 ratio by about 2.5 percentage points.
  - Exposures to sectors likely to suffer more from the COVID-19 pandemic (transportation, hotels, recreational activities) correspond to a much smaller amount and their defaults would pose minimal losses for banks in the sample.
  - Risk drivers: Deterioration in household balance sheets driven by the pandemic and recent softening in house prices may pose risks for some borrowers operating in the real estate sector.
- Regulatory and supervisory context:
  - HKSAR adopted Basel Committee’s large exposure framework in 2019, effectively into effect on July 1, 2019 with a six months’ implementation grace period for compliance with single counterparty and connected party exposure limits.
  - The new framework strengthened requirements on banks’ management of large exposures and enhanced HKMA’s monitoring.
- Recommendation:
  - HKMA should periodically stress test banks’ large exposures separately from their total loan books, considering potential cash flows from collateral in case of default of large borrowers.
  - Emphasis on close monitoring of large exposures given potential for large losses in rare default events.

### Bank liquidity risk analysis and stress tests — Overview
- Objectives:
  - Assess current banking system liquidity risks via analysis of short-term and structural liquidity ratios and cash flow–based liquidity stress tests.
  - Short-term measures: Basel III liquidity coverage ratio (LCR) and Liquidity Maintenance Ratio (LMR) for category 2 banks.
  - Structural measures: Net stable funding ratio (NSFR) and Core Funding Ratio (CFR) for category 2 banks.
- Cash-flow stress tests:
  - Use supervisory data on contractual cash flows and security flows across maturity buckets.
  - Multiple scenarios of increasing severity covering horizons of 7 days, 1 month, and 3 months with varying assumptions on shocks to inflows/outflows and haircuts on liquidity buffers.
  - Calibration based on historical evidence.
- Importance:
  - Ensure liquidity support provision aligns with the currency board arrangements.
  - Clear understanding of liquidity risk needed to mitigate risks or explore credible backstop options.

### Liquidity stress-test methodology and sample
- Approach to uncertainty: Conduct tests over a wide range of scenarios featuring different degrees of severity and central bank support to focus on liquidity risk tolerance rather than single parameter calibration.
- Sample:
  - 23 banks with total assets of 24,000 bn HKD, corresponding to 96 percent of total assets of banks in HK (about 24,900 bn HKD under a simplified consolidation approach as of June 2020).
  - Composition: 13 category 1 banks (including the 6 D-SIBs) and 10 category 2 banks; 11 locally incorporated banks among category 1; 11 foreign branches in the sample.
- Quantitative analysis components:
  - Analysis of composition and evolution of regulatory liquidity ratios.
  - Comprehensive cash-flow based liquidity stress tests.

### Regulatory liquidity ratios — Key statistics and drivers
- Implementation and minima:
  - LCR implemented for category 1 banks in 2015 with phase-in until 2019 when minimum reached 100 percent.
  - Category 2 banks subject to LMR with minimum requirement of 25 percent.
- Aggregate positions (as of June 30, 2020):
  - Average LCR: 157 percent.
  - Average LMR: 57 percent.
  - Since December 2018: LCR decreased by 10 percentage points; LMR increased by about 3 percentage points.
  - Since the outbreak of Covid-19 in Q1 2020 both short-term regulatory liquidity ratios remained quite stable.
- Distributional statistics (category 1 banks, after removing outliers, June 2020):
  - Median LCR: about 163 percent.
  - 90th percentile LCR: about 190 percent.
  - 10th percentile LCR: about 130 percent.
  - For local banks (after removing outliers): median LCR about 169 percent; 90th percentile about 200 percent; 10th percentile about 130 percent.
- Category 2 banks (LMR distribution, after removing outliers, June 2020):
  - Median LMR: around 46 percent.
  - 90th percentile LMR: 92 percent.
  - 10th percentile LMR: 36 percent.
- LCR in individual currencies:
  - USD LCR (category 1 banks) sensitive to the 75 percent inflow cap:
    - If cap not applied: median USD LCR about 311 percent; 10th percentile about 78 percent.
    - If cap applied: median USD LCR about 81 percent; 10th percentile about 52 percent.
  - RMB LCR (local banks) sensitive to 75 percent inflow cap:
    - If cap not applied: median RMB LCR about 154 percent; 10th percentile about 101 percent.
    - If cap applied: median RMB LCR about 115 percent; 10th percentile about 59 percent.
  - Note: Disruption of USD inflows (e.g., via the FX swap market) might result in USD liquidity risk exposure.
- LCR drivers:
  - Outflows of unsecured wholesale funding are the main driver of net cash outflows (NCOF).
  - Largest cash inflows are from secured and unsecured loans, including unsecured interbank lending.
  - Aggregate category 1 banks feature high-quality liquid assets (HQLA) of about 36 percent after net cash outflows.
- LMR drivers (category 2 banks, June 2020):
  - Liquefiable assets main drivers: net due from banks and debt issued/guaranteed by banks (64 percent), debt issued by governments (43 percent).
  - Qualifying liabilities main component: other (one-month) liabilities (-188 percent).
  - After accounting for qualifying liabilities, amount of liquefiable assets is negative at 90 percent of the starting value of 64 percent (consistent with a LMR of about 50 percent).
- NSFR/CFR positions:
  - All category 1 banks fulfill NSFR minimum requirement of 100 percent.
    - Median NSFR: about 120 percent.
    - 90th percentile NSFR: about 150 percent.
    - 10th percentile NSFR: about 115 percent.
    - Since December 2018 median decreased from 130 percent.
  - Local banks NSFR distribution: median 127 percent; 90th percentile about 148 percent; 10th percentile about 115 percent.
  - All category 2 banks fulfill CFR minimum requirement of 75 percent.
    - Median CFR: about 109 percent.
    - 90th percentile CFR: about 169 percent.
    - 10th percentile CFR: about 96 percent.
    - Compared to December 2018, CFR increased somewhat across the distribution.
- NSFR composition (category 1 banks, June 2020):
  - Available Stable Funding (ASF): Retail deposits account for 50 percent of ASF.
  - Required Stable Funding (RSF): Loans to non-banks are 50 percent of RSF.
- CFR composition (category 2 banks, June 2020):
  - Available Core Funding (ACF) drivers: deposits ~50 percent; other funding ~30 percent.
  - Required Core Funding (RCF) main component: loans to non-banks ~60 percent.
  - After accounting for RCF the amount of ASF is still positive at 20 percent of the starting value.
- Summary conclusion:
  - Quantitative analysis indicates banks hold sufficient liquidity to cover short-term outflows.
  - Structural liquidity ratios indicate the funding structure of HK banks is structurally sound.

### Cash-flow based liquidity stress tests (CFLST) — Method and objectives
- Purpose:
  - Estimate magnitude of potential liquidity needs of individual banks and the banking system (sample of 23 banks) under baseline and multiple stress scenarios.
  - Identify banks’ liquidity risk tolerance and circumstances requiring additional liquidity support due to insufficient counterbalancing capacity (CBC).
  - Assess common liquidity risk exposures (e.g., reliance on unsecured short-term funding, common exposures to less liquid assets).
- Method:
  - Transform reported cash-flow and security-flow data into stressed cash-flows and security-flows based on a matrix of scenarios defined by stress factors.
  - Two key indicators:
    - Cumulated net-funding gap (CNFG): sum of net-funding gaps (NFG) across time buckets (difference between cash-inflows and cash-outflows in each bucket).
    - Cumulated counterbalancing capacity (CCBC): sum of counterbalancing capacities (CBC) across time buckets and current one after taking into account CNFG; represents liquidity left at disposal of the bank at end of horizon after stress.
- Data and assurance:
  - Analysis builds on reporting data collected by the HKMA.
  - HKMA confirmed a thorough data quality assurance process had been applied.

*Source: https://www.imf.org/-/media/files/publications/cr/2021/english/1hkgea2021002.pdf*

### 91.      Short-term  contractual outflows of banks in the sample are high. Within  the first 4

### Short-term contractual outflows of banks in the sample are high

### Key findings on contractual outflows and inflows
- Short-term contractual outflows within the first 4 weeks amount to about 82 percent of total assets (weighted average; including current account deposits (38 percent of TA) and derivatives outflows (22 percent of TA)).
- Excluding derivatives outflows would result in short-term contractual outflows of 60 percent of total assets.
- Derivatives outflows are matched by derivatives inflows at the system level; a disruption of the derivatives markets (e.g., FX swap markets) might would have substantial effects on liquidity risk exposure.
- Compared to December 2018, contractual outflows arising from derivatives contracts in the first 6 months declined from 64 to 47 percent of TA.
- Contractual outflows from deposit funding remained basically unchanged, but their composition shifted further towards overnight (from 35 to 38 percent of TA).

### Contractual inflows (first month)
- Contractual inflows in the first month amount to about 62 percent of total assets.
- Main drivers:
  - Inflows from derivatives: 22 percent of total assets (identical to the corresponding outflows).
  - Readily monetizable debt securities: 17 percent of total assets.
- Compared to December 2018:
  - Contractual inflows from derivative contracts declined from 29 to 22 percent of TA.
  - Inflows from the Exchange Fund increased from 0.5 to 1 percent of TA.

### Contractual maturity mismatch and liquidity exposure
- Cumulated contractual maturity mismatch over the first month: 25 percent of total asset (mostly unchanged from December 2018 level of 24 percent).
- Over the first 6 months the maturity mismatch increases to 31 percent.
- Liquidity risk exposure of the system increased somewhat since December, while liquidity risk bearing capacity remained unchanged.

### On-balance sheet timing breakdown (selected exact values)
- On-balance sheet liabilities total by bucket: Next day 51%, 2 to 7 days 13%, 8 days to 1 month 17%, >1 month up to 3 months 29%, >3 months up to 6 months 13%.
- Major liability components (percent of TA in Next day bucket where shown):
  - Demand, savings and current account deposits: 38%
  - Amount payable arising from derivative contracts: 4% (Next day), 9% (2 to 7 days), 9% (8 days to 1 month), 17% (>1 month up to 3 months), 9% (>3 months up to 6 months)
  - Term, call and notice deposits: 3% (Next day), 3% (2 to 7 days), 6% (8 days to 1 month), 8% (>1 month up to 3 months), 2% (>3 months up to 6 months)
  - Due from banks / Due to banks and other granular items reported in the source tables.

- On-balance sheet assets timing breakdown (selected exact values):
  - Total inflows by bucket: Next day 32%, 2 to 7 days 13%, 8 days to 1 month 16%, >1 month up to 3 months 24%, >3 months up to 6 months 14%.
  - Readily monetizable: 17% (Next day)
  - Amount receivable arising from derivative contracts: 4% (Next day), 9% (2 to 7 days), 9% (8 days to 1 month), 17% (>1 month up to 3 months), 9% (>3 months up to 6 months)
  - Contractual Maturity Mismatch by bucket: -23% (Next day), -1% (2 to 7 days), -1% (8 days to 1 month), -6% (>1 month up to 3 months), 0% (>3 months up to 6 months)
  - Cumulative Contractual Maturity Mismatch: -23%, -24%, -25%, -31%, -31% (by cumulative bucket as shown).

### HKMA actions during March–April 2020 market disruption
- Measures taken to reduce liquidity risk during tightness in the Hong Kong dollar money market:
  - Clarifications on access to HKMA Standby Liquidity Facilities (SLF), including:
    - possibility to allow automatic rollover of liquidity support to provide longer-term funding to banks;
    - pricing the funding at levels to help reduce market volatility;
    - scope of eligible collateral not confined to HQLA as defined under the Banking (Liquidity) Rules.
  - Reduced issuance size of Exchange Fund Bills by a total of HK$20 billion to make more Hong Kong dollar liquidity available.
  - Launched a temporary US dollar liquidity facility (funds to be obtained through the FIMA Repo Facility established by the US Federal Reserve).
  - Active communication that the LCR was a usable buffer to be drawn under stress.
- HKMA reviewed banks’ internal policies and processes and conducted drills; found banks were ready to access SLF and draw on LCR under stress.
- Outcome: banks weathered the episode without any liquidity strains.

### Calibration and parameter uncertainty in the CFLST
- Scenario dimensions: (i) severity (level of run-off rates of liabilities), (ii) approach to counterbalancing capacity (level of haircuts), (iii) time horizon.
- Ceteris paribus, scenario impact increases with higher run-off rates, higher haircuts on non-marketable CB-eligible assets, and longer stress horizons.
- Deposit run-off calibration for CFLST (applies to mild, medium, and severe market scenario):
  - Stable retail, SME, and operational deposits: 1 percent
  - Less stable retail, SME, and operational deposits: 4 percent
  - Operational deposits not covered by deposit insurance: 8 percent
  - Non-operational deposits: 15 percent
- Severe/combined scenario (based on LCR calibration) deposit run-offs: 3 percent, 10 percent, 25 percent, and 40 percent for the corresponding deposit categories.

### Scenario parameters (selected exact ranges from Table 7)
- Unsecured LT/ST Issuances and financial deposits: 50–100
- Stable retail deposits: 1-3
- Unstable retail deposits: 4-10
- Operational corporate deposits (NFCs): 8–25
- Non-operational corporate deposits (NFCs) & other deposit outflows: 15–40
- Repo across all collateral classes*: 0
- Deposits FI: 50–100
- FX-Swaps in-/outflows: 0
- Derivative in-/outflows: 100
- Retail / corporate inflows: 0
- Central bank inflows: 100
- FI inflows: 50–100
- Other inflows: 100
- Committed lines provided by the bank: 10
- N.B.: * Stressed outflow rates for repos are captured by increasing haircuts for the underlying collateral class.

### Counterbalancing capacity (CBC) approaches and haircuts
- Three CBC approaches applied:
  i. Full CBC: full counterbalancing capacity taken into account and haircuts apply to all items.
  ii. Marketable CBC: focus on high credit quality and high market liquidity; haircuts of lower quality assets increase.
  iii. Cross-border CBC: focuses on operational risk and time delays of cross-border transfers; haircuts increase further for assets located outside the jurisdiction where they are central bank eligible.
- Example haircuts (Full CBC):
  - Exchange Fund Notes, HK government bonds, US treasuries: 3 percent
  - RMB denominated assets for the PBoC liquidity facility: 5 percent
  - AAA-rated securities not in former categories and assets eligible at overseas central banks: 10 percent
  - Other investment-grade securities: 15 percent
  - Assets potentially usable for repos: haircuts increase with risk weights (10 percent for 0 risk weight, 15 percent for 20 percent risk weight, and 20 percent for the rest).
- Marketable CBC increases haircuts for “Other investment grade securities” and for “Other assets which are reasonable expected to have potential use as collateral in secured borrowing operations” to 25 and 30 percent respectively.
- Cross-border CBC increases haircuts by 10 percentage points for assets located overseas but to be used predominantly in HK and vice versa.

### Time horizons, scenario matrix, and testing scope
- Horizons tested: 7 days, 1 month, and 3 months.
- For each horizon: mild market scenario, medium market scenario, severe scenario, and severe/combined scenario.
- For each scenario the three CBC approaches applied (Full CBC, Marketable CBC, Cross-border CBC).
- Total embedded stress scenarios: 36 (12 per horizon).
- Presentation of results per scenario matrix includes:
  - Cumulated counterbalancing capacity (CCBC) of the entire sample after stressed outflows in terms of total assets of the sample.
  - Number of banks with a CCBC below zero.
  - Liquidity shortage of the latter banks in terms of their average total assets.
  - Subsamples for local banks and foreign branches presented separately where relevant.

### Composition of unencumbered assets (June 2020)
- Local banks:
  - Mortgages: 19 percent of unencumbered asset portfolio.
  - “Other investment-grade securities”: 28 percent.
  - Smaller categories noted (examples from footnote): HK Government Bonds 0.7%, RMB-denominated assets usable under RMB Liquidity Facility 1.6%, other categories with small shares.
- Foreign banks (foreign branches):
  - “Other investment-grade securities”: 54 percent of unencumbered asset portfolio.
  - Mortgages: 0 percent.
  - Shift since December 2018: composition of unencumbered assets shifted further towards Other investment grade securities (+8 percentage points).

*Source: HKMA and IMF staff, as presented in the FSAP chapter content.*

### 106.      The HKSAR banking sector is highly liquid and highly resilient to all 7-day stress

### 1hkgea2021002 - 106.      The HKSAR banking sector is highly liquid and highly resilient to all 7-day stress

### 7-Day Horizon: resilience and key figures
- Table 9 presents the CCBC under different stress scenarios, at a 7-day horizon.
- Even under the most severe stress scenario, the HKSAR banking system remains very liquid with a CCBC of 15.1 percent of the total banking assets of the sample.
- Compared to December 2018, this constitutes an improvement of 0.7 percentage points.
  - Local banks: +1.3 ppt improvement.
  - Foreign branches: -1.4 ppt deterioration.
- No bank experiences a negative CCBC under any of the 7-day stress scenarios.
- Overall, the average CCBC as a percentage of total assets of the entire banking sector is above 15 percent under most scenarios.

### 1-Month Horizon: resilience, vulnerabilities, and drivers
- System-wide resilience:
  - Overall cumulative counterbalancing capacity decreases mildly to an average of 12 percent under most scenarios and never dips below 10 percent.
  - Under the most severe scenario, for the entire sample the CCBC at the end of the stress horizon stands at 10.4 percent.
  - Local banks: CCBC = 10.1 percent of total assets.
  - Foreign branches: CCBC = 11.8 percent of total assets.
- Change since December 2018:
  - CCBC at the end of the stress test horizon increases by 0.8 percentage points overall.
    - Local banks: +0.8 ppt.
    - Foreign branches: +0.7 ppt.
- Incidence of negative CCBC:
  - At the one-month horizon, all local banks remain liquid while one foreign branch faces liquidity pressure.
  - Only one foreign bank experiences negative CCBC after one-month, and only under the most extreme scenario, with a shortfall of 1 percent of total assets.
  - Comparative December 2018: up to 3 foreign branches with negative CCBC and shortfalls of up to 2.2 percent of their total assets.
- Main drivers of foreign-branch vulnerabilities:
  - Reliance on unsecured interbank funding and unsecured wholesale funding from non-banks.
  - Derivatives transactions generate large outflows (-5% of total assets) compensated by corresponding inflows (+5% of TA), exposing banks to short-term roll-over risk in derivative markets.
  - Other important flows: outflows due to bank loans (-5% of TA), haircuts on liquefiable assets (-2% of TA), outflows of non-bank deposits (-2% of TA).
  - Due from banks is a major source of inflows under stress: +5% of TA.
- Resulting CCBC under the simulated 1-month severe combined scenario: cumulated counterbalancing capacity stands at 13% of TA.
- Notable changes since December 2018:
  - Derivative related in- and outflows decreased from 10% to 5%, reducing roll-over risk.
  - Outflows from non-bank deposits and from haircuts doubled from 1% to 2%.

### 3-Month Horizon: depletion of buffers and differential impact
- System-wide outcomes:
  - Overall CCBC decreases substantially after 3 months of cumulated run-offs and haircuts, to an average of around 12 percent for most scenarios.
  - Under the most extreme severe market run-offs scenario, CCBC dips below 6 percent.
  - For the entire sample, CCBC at the end of the stress horizon stands at 4.9 percent under the most severe scenario.
    - Local banks: CCBC = 6.2 percent of total assets.
    - Foreign branches: CCBC = -0.7 percent of total assets.
- Change since December 2018:
  - Local banks: improvement of 0.3 ppt.
  - Foreign branches: deterioration of -2.1 ppt.
- Incidence of negative CCBC:
  - The number of banks with negative CCBC at the end of the 3-month horizon amounts to up to 9.
  - Local banks: only in the most severe scenarios do 1 to 2 local banks face negative CCBC.
  - Foreign branches: heterogeneous—1 foreign branch features negative CCBC even in the mildest scenario; increases to 7 under the most severe scenario (out of eleven foreign branches).
  - Compared to December 2018, results unchanged for local banks; number of stressed foreign branches increased.
- Magnitude of shortfalls:
  - Foreign branches with negative CCBC feature shortfalls of about 9 percent of their total assets under most scenarios.
  - Two illiquid local banks under the harshest scenarios have negative CCBC representing about 4 to 6 percent of their total assets.
  - Since December 2018, the shortfall increases for the entire system by +1.1 ppt, with both subgroups contributing roughly equally.
- Drivers and structural notes:
  - Local banks: haircuts on CBC assets are the largest outflows reflecting high share of "Other investment-grade securities" in unencumbered assets; high gross derivative flows increase roll-over risk.
  - Foreign branches: main funding source unsecured interbank loans dries up.
  - CBC at the beginning of the period decreased from 24 to 19 percent of total assets and outflows due to non-bank deposits doubled to 4 percent.

### D. Liquidity Risk Assessment and policy recommendations
- HKMA stress-testing framework:
  - HKMA runs four liquidity stress tests with scenarios and frequencies: (i) market-wide liquidity freeze and (ii) short-term confidence-driven bank run run quarterly; (iii) USD and capital outflow and (iv) combined market and idiosyncratic shock run monthly.
  - Stress-test horizon: 1 month (except short-term bank run covers seven-day period).
  - Calibration based on historical event studies.
  - Tests are integrated into supervisory process and can lead to thematic reviews.
  - Data availability is high; improvements suggested on securities flows and behavioral flows reporting and a split by overall currency position and major foreign currencies.
- FSAP conclusions:
  - HKSAR banking system is resilient to liquidity stress; local banking system has sufficient liquidity buffers despite short contractual maturity of deposit base.
  - Availability of HKD-denominated HQLA is limited; HKMA applies an Alternative Liquid Asset (ALA) approach under the LCR.
  - After haircuts, about 20 percent of the counterbalancing capacity (or about 4 percent of bank assets) consists of direct or indirect claims on the Exchange Fund.
  - Exchange Fund amounts to about 9 percent of total bank assets and could absorb systemic bank liquidity shocks.
  - HKMA runs a large excess coverage of its currency board requirements and has significant FX buffers; as of December 2020, HKMA’s foreign reserves stood at USD 491 billion, corresponding to almost fifty percent of HKD M3.
- Policy recommendations (FSAP):
  - Consider incentivizing banks to lengthen and stagger the funding tenor of the deposit base to strengthen depositor and investor confidence.
    - Possible instruments: supervisory expectations for lengthening and staggering deposit funding, ex-ante pricing of lender of last resort facility, differentiated required reserve ratios based on deposit maturity, liquidity regulation, or pricing of deposit insurance premia.
  - For foreign branches: lengthen and stagger tenors of unsecured interbank funding from HQs and other banks to improve resilience; intragroup lending limits by HKMA are welcome.
  - Introduce an HK covered bond regime to:
    - Reduce indirect exposure of HKMA via HKMC.
    - Provide a substitute for less stable unsecured funding under stress.
    - Increase volume and diversification of HQLA in HKSAR.
    - Reduce liquidity risk exposure of HKMC and indirect pressure on the Exchange Fund.
  - Encourage development of a liquid, deep, and broad repo market to provide alternative secured funding under stress and reduce reliance on HKMA backstop; matches banks’ resolution plans that anticipate switching from unsecured funding to repos.

### COVID-19 comparison and recent developments (December 2018 vs June 2020)
- Balance sheet evolution:
  - Marginal shift from interbank assets towards loans to non-banks and other assets.
  - Interbank liabilities decreased and other liabilities (incl. capital) increased.
  - Lower reliance on unsecured interbank funding reduced liquidity risk exposure.
- Regulatory liquidity ratios:
  - In June 2020, average LCR decreased by 10 percentage points from its peak in December 2018 to about 157 percent, which is just above its average level of the past five years (2015-2019).
  - LMR increased by about 3 percentage points from December 2018 to 57.2 percent in June 2020.
- Contractual flows and asset composition:
  - Contractual inflows and outflows from derivatives contracts in the first 6 months declined significantly.
  - Contractual outflows from deposit funding remained basically unchanged, but composition shifted further towards overnight, increasing roll-over risk somewhat.
  - Composition of unencumbered assets shifted to less liquid assets: mortgages (local banks) and other investment-grade securities (foreign branches).
- Net assessment of changes since December 2018:
  - Results improve somewhat for local banks in 1-week and 1-month horizons.
  - Results are somewhat weaker for foreign branches.
  - In the 3-month horizon, results are somewhat weaker for both local banks and foreign branches, driven by lower CBC at the beginning of the period and higher outflows due to non-bank deposits.

*Source: IMF staff.*

### 123.      The HKMA  supported the resilience of the HK banking system with respect  to liquidity

### The HKMA supported the resilience of the HK banking system with respect to liquidity

### Liquidity resilience support and summary recommendation
- The HKMA supported the resilience of the HK banking system with respect to liquidity risk by the pro-active and the clear communication that the LCR is a buffer usable under stress and a series of systemic liquidity measures.
- Recommendations: The authorities should consider the following recommendations, to increase further the liquidity resilience of the banking system.

### Streamline liquidity monitoring and stress testing (Recommendation)
- Streamline the requirements for liquidity monitoring and liquidity stress testing.
- Current situation:
  - HKMA collects substantial data across various HKMA returns (esp. MA(BS)1, MA(BS)1E, MA(BS)18, MA(BS)23).
  - Definitions of items, the currency split, the granularity of sub-items and availability of remaining contractual maturities differ across returns.
- Suggested actions:
  - Consider streamlining the cash-flow/securities-flow reporting in MA(BS)23 and MA(BS)18.
  - Integrate all existing liquidity stress-tests into a single framework that is in line with LCR.
  - New reporting template should allow a clear distinction between HQLA Level 1, 2A and 2B and other CB eligible and other non-CB eligible assets.
- Rationale:
  - Alignment with LCR would reduce reporting costs for banks, especially large internationally active banks, as the LCR is a global standard.
  - Because contractual maturity mismatch is intrinsically large for banks, supervisors need information on how banks plan to fund it; behavioral flows provide additional information about banks' reactions to market changes.
  - Stress tests based on contractual data require a two-step approach: (1) derive the “business as usual” liquidity situation based on supervisors’ assumptions; (2) stress this unstressed liquidity profile. If behavioral flows are available, supervisors can stress these directly.

### Integrate HKMA liquidity stress tests with LCR and extend horizons (Recommendation)
- Integration:
  - Integrate all four HKMA liquidity stress tests in one framework and with the LCR once comprehensive cash-flows and securities flows reporting data is available in different currencies.
- Complementary approaches:
  - Complement the implied cash-flow cum liquidity coverage approach with the cash-flow/securities-flow based stress test cum counterbalancing capacity approach.
- Extended horizons:
  - Better position to add stress horizons beyond 1 month (i.e. 3 and 12 months).
  - Integration with LCR would allow monitoring of build-up of cliff-effects after 30 days.

### Funding diversification and tenor management (Recommendation)
- The HKMA should continue to incentivize banks to diversify funding sources, lengthen and stagger funding tenors of the deposit base.
- Observations:
  - Hong Kong banking system heavily relies on short-term deposit funding.
  - Retail and non-financial wholesale deposits offer relative stickiness advantages but are often short-term and term deposits are subject to early withdrawal risk.
  - Contractual maturity of derivative flows is also short; gross in- and out-flows are substantial even if they largely cancel at system level.
- Suggested supervisory measures:
  - Apart from existing measures to monitor and supervise banks’ funding profile, consider supervisory expectations and internal limits on short-term contractual cash-outflows.

### Develop more stress-resilient funding sources (Recommendation)
- Authorities should support development of funding sources likely to be available under stress:
  - Broad and deep repo markets.
  - Development of safe assets to reduce shortage of high-quality liquid assets in HK (e.g. a covered bond regime).
- Rationale:
  - Under stress, secured funding sources (repo or covered bonds) can be important and less expensive substitutes for unsecured funding if infrastructure exists and funding sources are tapped regularly under unstressed conditions.
  - Consider introduction of a Hong Kong covered bonds regime to alleviate the HQLA shortage in HKSAR.

### Operational and early redemption deposit risks (Recommendations)
- Operational deposits:
  - FSAP appreciates HKMA efforts to investigate banks’ practices regarding classification of operational deposits (see section 5.8.20 LM-1 and section 7(1) of the Code of Bank Practices).
  - Materiality tests demonstrate reclassification risk for operational deposits is material; banks’ liquidity position might be less stable under stress than LCR suggests.
  - Supervisory review is adequate instrument to ensure criteria for operational deposits are applied rigorously, banks monitor behavior of these deposits, and react swiftly to reclassify deposits behaving like non-operational deposits.
- Early redemption risk:
  - HKMA should analyze early redemption risk for time deposits and banks’ measures to address it (e.g., the material penalty for early withdrawal of term deposits based on section 5.8.13-14 of the Regulatory Framework for Supervision of Liquidity Risk (LM-1)).
  - Materiality tests show early withdrawal risk is material; retail deposits might be less stable than expected and banks’ liquidity position might be less stable under stress than LCR suggests.
  - Supervisory review should ensure banks’ policies and practices (including significant penalty for early withdrawal) are effective and draw on banks’ historical data relating to early withdrawal risk in normal or stressed situations.

### Address moral hazard from HKMA liquidity backstop (Recommendation)
- Given the important role of HKMA in banks’ liquidity risk management, consider measures to address moral hazard stemming from market perception that HKMA would provide support under stress.
- Observations:
  - Recent reorganization of liquidity facilities at HKMA has not clarified maximum capacity to support financial institutions in Hong Kong, possibly leading market participants to believe HKMA could support them in all circumstances.
  - HKMA policy statement specifies liquidity support to banks would be provided on a collateralized basis and terms and conditions considered case by case.
- Risk:
  - Unpriced liquidity insurance might lead to inefficient excessive liquidity risk exposure of HKSAR banks.

### Interconnectedness and contagion analysis — Background
- HKSAR banking sector has one of the highest levels of cross-border exposures in the world.
- BIS locational banking statistics indicate HKSAR banks’ cross-border exposures on the asset side are significantly higher, with total cross-border claims exceeding liabilities by close to 30 percent in 2019.
- Historical pattern:
  - Prior to AFC, HKSAR banks were net borrowers; with AFC onset they sharply reduced cross-border liabilities and have been a net lender to other banking sectors abroad since late 1999.
- Mainland China exposure:
  - Mainland China’s banking sector receives the largest share of HKSAR’s cross-border lending by far.
  - HKSAR lending to Mainland China increased substantially after the GFC.
  - Mainland China’s share within total HKSAR cross-border lending rose from less than 10 percent in 2009 to close to 25 percent in recent years.
  - Large interbank claims to Mainland China add to HKSAR’s significant non-financial sector loan exposures there (close to 40 percent of all (non-bank) customer loans of the HKSAR’s total banking system are Mainland China-related loans).

### Cross-border and branch structure findings
- A significant portion of cross-border interbank lending from HKSAR is intermediated by foreign branches; foreign branches account for a majority of total cross-border interbank claims despite being substantially smaller than the local banking sector.
- Majority of foreign branches’ cross-border interbank claims are on “related offices” (entities part of the same banking group).
- Together with related office lending of local banks, related office claims account for majority of HKSAR’s total cross-border interbank claims — indicating international banks in HKSAR, especially foreign branches, use HKSAR as a source of funding for global operations.

### Domestic financial connectedness
- Domestic fund flows: strong flows from local banks to foreign branches and from foreign branches to nonbank financial institutions (NBFIs).
- Analysis basis:
  - Domestic connectedness analysis based on bank-level data from a sample of the 20 largest banks (including foreign branches), ranked by assets held by their HKSAR offices.
- Key statistics:
  - The largest portion of local banks’ intra-financial system lending goes to foreign branches, which account for more than 53 percent.
  - Interbank lending to local banks and lending to nonbank financial institutions account for just over 20 percent each.
  - Foreign branches’ intra-financial system lending: 64 percent goes to NBFIs.
- Implications:
  - Foreign branches could be a source of significant spillovers to other financial sectors at home and abroad, sourcing funding from local banks and extending significant lending to NBFIs in HKSAR and other banking systems abroad.

### Data gaps and NBFI build-up risk
- Data gaps:
  - No available data on certain sector-by-sector exposures between non-bank sectors limits mapping of key financial linkages across the whole system.
  - Ideal dataset would be who-to-whom exposures across banking system, NBFIs, non-financial corporates, households, government sector, and rest of the world.
- NBFI build-up:
  - NBFI loans-to-GDP ratio increased by close to 20 percentage points since 2012 and reaching 34 percent recently.
  - A build-up of indebtedness in corporate or household sectors beyond bank-provided loans may not be detected timely, reducing effectiveness of macroprudential measures applied to banking sector and increasing indirect exposures of the banking sector to other sectors.

### Domestic contagion analysis — findings
- Objective: Investigate potential for contagion between local banks and other financial institutions in HKSAR using bank-level data.
- Rationale: Size of local banking system relative to GDP and importance of bank funding for non-financial sectors means adverse shocks to banks could significantly affect the economy.
- Local banks’ exposures (combined bank sample):
  - Local banks’ direct exposures to NBFIs are modest, with exposures to each NBFI sector amounting to around 5 percent, or less, of the banks’ CET1 capital.
  - Exposure to group of foreign branches is just over 42 percent of total CET1 capital.
  - Interbank lending exposures to other local banks amount to less than 18 percent of total CET1 capital of local banks.
- Interbank lending maturities:
  - Most interbank lending is short-term: among sample, more than half of total domestic interbank lending has a maturity of one month or less, and more than seventy percent has a maturity of three months or less.
- Bank-by-bank impact:
  - Distribution of interbank exposures is highly skewed with large outliers.
  - At the median of the sample, the maximum impact of the failure of other individual banks (with an LGD assumption of 100 percent) is 7 percent of CET1 capital for the lender bank.
  - For a few banks the maximum impact is much higher; the highest impact exceeds 100 percent of CET1 capital for the bank with the largest exposures.
- Intragroup concentrations:
  - Most outsized positions are between related offices (intragroup exposures).
  - Such intragroup exposures are exempt from statutory intra-group exposure limits under the Banking (Exposure Limits) Rules (BELR); HKMA asks banks to establish internal limits as appropriate.
  - Large intragroup positions may expose local banks to potential liquidity problems faced by other group entities and to reputational risk, highlighting importance of heightened monitoring by HKMA of liquidity risk at group and entity level for banks operating multiple group entities in HKSAR.

*Source: 1hkgea2021002 - The HKMA supported the resilience of the HK banking system with respect to liquidity (excerpt).*

### 146.      The significant interbank  flows from local banks to foreign branches, whether between

### The significant interbank flows from local banks to foreign branches

### Interbank linkages and oversight
- Significant interbank flows from local banks to foreign branches, whether between same group entities or unrelated ones, create indirect linkages between local banks and other sectors of the economy.
- Local banks have access to a very strong base of local deposits in HKSAR.
- The interbank market provides a venue for liquidity management and enables local banks to deploy short-term excess funding.
- Foreign branches source funding from local banks and lend to large corporates and NBFIs, some of which provide household loans.
- These indirect linkages have the potential to create leakages for regulations aimed at mitigating build-up of risks (for example, capital-based regulation targeting bank lending to nonfinancial corporates may be circumvented if lending shifts from local banks to foreign branches).
- Highlights the importance of oversight over banking groups that have both foreign branches and local subsidiaries in HKSAR.

### Cross-border Contagion Analysis — Balance Sheet-based Approach
- Purpose: assess HKSAR banking system’s resilience to shocks from other banking systems and its potential to act as a source of shock to other banking systems.
- Methodology:
  - Network contagion framework of Espinosa-Vega and Sole (2010).
  - Assumes an initial shock to a jurisdiction’s banking system can propagate through direct claims and indirect linkages via other jurisdictions’ banking systems.
  - If any banking system incurs losses larger than its capital base, it is deemed to “fail”; such failure can cause other systems to fail via credit or funding shocks.
  - Numerical analysis used cross-border interbank claims and liabilities of 33 banking systems as reported in the BIS locational banking statistics.

### Contagion scenarios and assumptions
- Scenario 1 (credit shock):
  - Simulated failure of a banking system leads to credit losses for any system that has claims against the failing system.
  - Assumes a 100 percent loss rate on claims against a failing counterparty.
- Scenario 2 (credit and funding shocks):
  - Credit shock as in Scenario 1.
  - A shock to a banking system leads to the loss of funding for all other banking systems borrowing from the initial shocked system.
  - Fraction of lost funding that is not replaceable assumed to be 35 percent (i.e., 65 percent rollover).
  - Affected systems are assumed to fire-sale assets with a haircut of 25 percent.

### Balance-sheet results and interpretation
- HKSAR’s banking sector is more vulnerable to credit shocks from abroad than to funding shocks.
- Index of Vulnerability (inward spillover): average share of capital lost for the reported bank due to failures of other banking systems.
  - Under Scenario 1, HKSAR is among systems with highest vulnerabilities to credit shocks from abroad.
  - Average impact of other banking systems’ failure on HKSAR is a loss of about 13 percent of its initial capital.
- Index of Contagion (outward spillover): average share of capital lost for other banking systems due to failure of the reported system.
  - HKSAR is among systems with high estimated impacts on other systems, but ranking is lower than its Index of Vulnerability.
  - HKSAR’s hypothetical failure would lead to an average loss of about 6 percent of capital for other systems that have asset exposures to HKSAR.
- The lower outward spillover compared to inward vulnerability is consistent with HKSAR’s cross-border exposures being significantly lower on the liability side compared to the asset side.
- Scenario 2 results:
  - Additional impact of loss of funding on top of credit shock is relatively mild for HKSAR.
  - Patterns similar to Scenario 1: HKSAR among systems with highest vulnerabilities to credit shocks from abroad and lower outward spillover potential.
  - Average losses of HKSAR’s banking system go up only by a modest amount (less than six percent of the base losses) when the additional funding shock is layered onto Scenario 1.
  - The small increase reflects HKSAR’s cross-border liabilities being significantly lower than cross-border assets.

### Cross-border Contagion Analysis — Market-based Approach
- Rationale: capture indirect linkages via similar business models or common exposures; market participants’ revisions to expectations can transmit stress even absent direct linkages.
- Methodology:
  - Spillover approach of Diebold and Yilmaz (2014) using directional interconnectedness measures estimated from a VAR model.
  - Decomposes variation in an asset’s volatility into contributions from shocks to other assets; measures inward and outward spillovers and net contribution.
  - Applied to daily volatility of equity returns for a sample of G-SIBs (as identified by the FSB as of November 2020) and five publicly traded HKSAR banks.
- Findings:
  - D-SIBs in HKSAR appear to be key recipients of spillovers from G-SIBs as measured by bilateral linkages.
  - HKSAR banks are generally net recipients of shocks from G-SIBs in the US and the Euro Area.
  - This market-based result is consistent with the balance-sheet contagion analysis that showed higher inward vulnerability than outward spillover potential for HKSAR.
  - Caveat: equity prices used are those of global parents (many HKSAR banks are subsidiaries), so reported spillover estimates are not fully attributable to local HKSAR banking system conditions.
  - Note: stock return co-movement can be influenced by crisis risk-off sentiment or trading dynamics and may not solely reflect fundamental linkages; market-based results should be evaluated alongside balance-sheet approaches.

### FSAP recommendations to enhance banking system stability
- Enhance HKMA’s data collection of intersectoral claims across major sectors (banking, NBFIs, non-financial corporates, households, government, and the rest of the world), with a focus on claims of the NBFI, to better gauge intersectoral linkages and develop an aggregate balance sheet approach in systemic risk assessment.
- Enhance oversight over banking groups that have both foreign branches and local subsidiaries in HKSAR to improve monitoring and mitigation of contagion risks.
- Heightened monitoring of liquidity risk both at the group and entity level for banks that operate with multiple group entities in HKSAR to ensure adherence to internal risk management practices.

### Corporate sector analysis — Recent developments
- Non-financial corporate debt increased rapidly in HKSAR since the Global Financial Crisis:
  - From 127 percent of GDP in 2009 to 243 percent in 2020Q2.
  - This level remains significantly higher than the average of advanced economies, with the increase since the GFC exceeding most advanced economies and peer financial centers.
- Less than half of that borrowing was intended for use in HKSAR; the remainder was cross-border credit, mostly to Mainland China firms.

### Corporate sector analysis — Key vulnerabilities and risks
- Given large presence of Mainland China firms in HKSAR, vulnerabilities in both local and Mainland China firms in HKSAR are important for financial stability.
- HKSAR provides fundraising platform and asset management services for Mainland China firms.
- HKSAR banking sector became increasingly active in Mainland-related lending, which accounted for nearly 40 percent of total bank loans as of September 2020.
- HKSAR banking sector exposures to Mainland China’s non-banks increased to HK$6.5 trillion by end-September 2020 (about 237 percent of GDP) and accounted for the majority of total banking sector exposures to Mainland China.
- In 2020:
  - Mainland China firms raised a total of US$84 billion equity through HKSAR stock exchanges, equivalent to 88 percent of total equity fund raised in HKSAR.
  - Mainland China firms issued a total of US$128 billion USD-denominated bonds in HKSAR, accounting for more than 60 percent of total offshore issuances by Mainland China firms.
- Deterioration in corporate balance sheets (for example due to weaker Mainland China growth) could reduce asset prices and increase credit risk for financial institutions in HKSAR.

### Firm-level analyses — scope and methods
- Two types of listed firms analyzed:
  - HKSAR-oriented firms: domiciled in HKSAR, listed in HKSAR stock exchanges, mainly exposed to HKSAR.
  - Mainland China-oriented (MC-oriented) firms: listed in HKSAR, domiciled in either HKSAR or Mainland China, but exposed to Mainland China.
- Data and coverage:
  - Only listed firms included because data for non-listed firms are limited; listed firms’ data available at least until 2019 with about 20 percent having financial statement data available by 2020H1.
  - Total debt of listed firms domiciled in HKSAR accounted for nearly 90 percent of HKSAR’s total non-financial corporate debt in 2019.
  - Data source: Bloomberg, L.P.; sample of about 2,000 firms. WIND used as alternative for robustness; main results qualitatively unchanged.
- Scenario-based firm-level analyses conducted:
  - Debt-at-risk analysis:
    - Uses balance sheet and cashflow data.
    - Assesses change over time in overall corporate sector vulnerabilities (debt repayment capacity).
    - Examines which sectors are most vulnerable to shocks in the adverse macroeconomic scenario of the bank stress test.
    - Estimates impact of adverse shocks on the ICR-implied default risk by sector.
    - Debt-at-risk defined as the share of debt owed by firms with interest coverage ratio (ICR) below 1 in total debt.
  - Bottom-up default analysis (BuDA):
    - Uses market-based information and balance sheet indicators to estimate corporate default risk.
    - Assesses sensitivity of corporate defaults to shocks in the adverse macroeconomic scenario and identifies most vulnerable sectors.

*Source: Excerpt from IMF staff analysis in the FSAP chapter on HKSAR banking system and corporate sector (text provided).*

### 158.      Debt-at-risk would increase

### 158. Debt-at-risk would increase

### Debt-at-risk under the adverse scenario
- The adverse scenario applies an interest rate shock of a 3.77-ppt increase in borrowing interest rate (aligned with the peak of the corporate yield increase in the bank solvency stress test adverse scenario); this shock has the largest impact on debt-at-risk.
- Shocks to earnings have a larger impact on HKSAR-oriented firms, reflecting the larger decline in HKSAR’s real GDP growth rate in the adverse scenario.
- A RMB depreciation (against the USD) would significantly reduce firms’ earnings in USD, particularly MC-oriented real estate firms whose business revenues are mostly from onshore Mainland China property development and denominated in RMB; however, the impact on debt-at-risk is mitigated by their RMB-denominated debt.
- All shocks combined would raise the debt-at-risk of both HKSAR- and MC-oriented firms beyond their levels in the GFC, with a more prominent increase in MC-oriented firms.
- MC-oriented firms would have a higher debt-at-risk than HKSAR-oriented firms in the adverse scenario despite having a lower debt-at-risk in the baseline, mainly due to the higher sensitivity of “marginal” MC-oriented firms to combined earnings and interest rate shocks.
- From a sectoral perspective, MC-oriented real estate experiences the largest increase in debt-at-risk under the adverse shocks and contributes the most to total debt-at-risk.
- Firms’ holdings of cash and other liquid assets could lower post-shock debt-at-risk but only marginally because cash use is conservatively limited so that quick ratios (cash + other liquid assets + accounts receivable divided by current liabilities) are kept above or equal to 1.

### Default risk: approaches and aggregate results
- Two approaches estimate firm-level probability of default (PD):
  - Mapping firm-level ICRs from the debt-at-risk analysis to PDs (ICR-mapped PDs).
  - The BuDA model, which uses market-based information (equity prices) and balance sheet indicators to estimate PDs.
- Mapping method details:
  - ICR levels are mapped into cumulative default rates over the next two years using historical default rates by rating agencies.
  - Aggregate PD on corporate debt owed to HKSAR banks is constructed by weighting firm-level PDs by each firm’s share of debt owed to HKSAR banks in total debt; the share of firms’ debt owed to HKSAR banks in total debt is approximated as about 46 percent.
- Key numeric results (ICR-mapped PDs):
  - Aggregate PD would increase to 8.3 percent of the overall bank corporate loan book from 2.9 percent in 2019 and 3.6 percent in the baseline (cumulative new defaults over two years).
  - Combined shocks increase new defaults in the MC-oriented real estate sector from 1.1 percent in the baseline scenario to 7.9 percent in the adverse scenario.
  - The increase in aggregate PDs in the adverse scenario relative to the baseline is slightly below the 5 percentage point increase in aggregate PD during the AFC.
- BuDA model projections:
  - BuDA projects an increase in the median PD in the near term under the baseline due to the continued economic downturn in HKSAR in 2020, followed by a gradual decline as the economy recovers.
  - In the adverse scenario, default risk rises rapidly in the first two years, approaching GFC levels but remaining below AFC peaks.
  - Sectorally, BuDA implies a significantly higher default risk priced for the MC-oriented real estate sector in the adverse scenario.
- Comparison of approaches:
  - BuDA-estimated PD levels are generally lower than ICR-mapped PDs, but both approaches imply significantly higher PDs in the adverse scenario than the baseline, especially for the MC-oriented real estate sector.
  - The peak of the BuDA-estimated median PD in the adverse scenario is about 1½ times that in the baseline for both HKSAR- and MC-oriented firms.
  - The ICR-mapped aggregate PD in the adverse scenario is over 2 times the baseline level and is relatively closer to the 90th-percentile of the BuDA-estimated PDs.

### Key scenario and model assumptions (summary of Box 2)
- Debt-at-risk analysis:
  - Debt-at-risk assessed via interest coverage ratios (ICR = EBIT / interest expense) after applying shocks to earnings, interest rates, and RMB/USD exchange rate derived from bank stress test baseline and adverse scenarios.
  - Sector-specific earnings shocks calibrated using estimated sensitivities of sectoral earnings to real GDP growth during the GFC and the declines in real GDP growth in the baseline and adverse scenarios.
  - Interest rate shocks follow corporate yield assumptions: a reduction of 68 bps in the baseline and an increase of 377 bps in the adverse scenario; interest rate shock assumed to apply to 90 percent of total debt (10 percent fixed-rate).
  - RMB depreciation of 7.4 percent against the USD is assumed to test MC-oriented firms’ sensitivity; regression-based approach used to estimate impact on firms’ debt and earnings.
- BuDA scenarios:
  - Residential property price in Mainland China: baseline follows October 2020 WEO baseline; adverse assumes a 20 percent peak decline during 2020-21 relative to the baseline, recovering to 5 percent below baseline by end-2022.
  - Stock prices in HKSAR and Mainland China: baseline holds end-Q3 2020 levels over 2020Q4-2022; adverse assumes a 40 percent peak decline during 2020-21, recovering to 10 percent below baseline by end-2022.
  - RMB/USD exchange rate: baseline uses historical values for 2020Q1-Q3 (an appreciation of 3.1 percent relative to end-2019) and then unchanged; adverse assumes a 7.4 percent depreciation to 7.55 RMB per USD at peak during 2020-21, recovering to 7.3 RMB per USD by end-2022.

### Caveats and limitations
- Debt-at-risk and default risk results refer to the aggregate non-financial corporate sector; implications for individual banks may differ because of heterogeneity in banks’ corporate exposure composition.
- Debt-at-risk should not be directly interpreted as an indicator of bank losses but as a vulnerability indicator for the aggregate non-financial corporate sector.
- Analyses include only listed firms; most small firms are excluded and could be subject to higher stress in the adverse scenario, so impacts on the overall non-financial corporate sector may be underestimated.
- BuDA is market-based and regression-driven and may be subject to modeling and estimation errors, particularly at the individual firm level.
- Aggregate PD estimates are sensitive to assumptions on the representativeness of the firm sample and the share of firms’ borrowing from HKSAR banks; the corporate loan book is assumed to remain constant during the shock period.

### Policy recommendations
- Authorities should continue to closely monitor credit risks of MC-oriented real estate firms.
- Ensure banks’ internal risk models used to determine capital charges for potential high-risk Mainland China borrowers are sufficiently forward-looking, particularly for MC real estate borrowers with low credit ratings.
- Consider incorporating non-financial corporates into the stress testing framework and communicate key findings in the HKMA’s Monetary and Financial Stability Report on a regular basis.

*Source: Excerpt titled "158.      Debt-at-risk would increase" from the IMF staff analysis in the provided content.*

### Appendix I. Risk Assessment  Matrix

### Appendix I. Risk Assessment  Matrix

### Global Risks
- Unexpected shift in the Covid-19 pandemic  
  - Transmission channels: Prolonged pandemic: The disease proves harder to eradicate (e.g., due to new virus strains, short effectiveness of vaccines, or widespread unwillingness to take them), requiring costly containment efforts and prompting persistent behavioral changes rendering many activities unviable.  
  - Likelihood: Medium  
  - Impact: Medium

- Accelerating de-globalization  
  - Transmission channels: Despite renewed efforts to reach multilateral solutions to existing tensions, geopolitical competition leads to further fragmentation. Reshoring and less trade reduce potential growth.  
  - Likelihood: Medium  
  - Impact: High

- A sharp rise in global risk premia exposes financial and fiscal vulnerabilities  
  - Transmission channels: A reassessment of market fundamentals (e.g., in response to adverse Covid-19 developments) triggers a widespread risk-off event. Risk asset prices fall sharply and volatility spikes, leading to significant losses in major non-bank financial institutions. Higher risk premia generate financing difficulties for leveraged firms (including those operating in unviable activities) and households, and a wave of bankruptcies erode banks’ capital buffers.  
  - Likelihood: Medium  
  - Impact: High

### Domestic Risks
- A further slowdown in China  
  - Transmission channels: A further slowdown in China would reduce trade flows and lead to a sharp decline in economic activity in HKSAR. Lower trade volumes would reduce corporate profits and investment, raise unemployment, and weaken consumption. Trade, logistics, tourism, and retail sectors would be affected most adversely. Banks would likely face higher losses on their loans to Mainland China-related and other corporate borrowers, as well as on their broader credit portfolio due to the overall decline in economic activity.  
  - Likelihood: Medium  
  - Impact: High

- A sharp housing market correction and decline in demand  
  - Transmission channels: A sharp decline in house prices would lower confidence, weaken housing demand, lower residential investment, and constrain new bank lending as many loans are secured by property, leading to a larger downturn.  
  - Likelihood: Medium  
  - Impact: High

### Notes
- This table shows the shocks that will be used for the analysis of resilience of financial institutions.  
- The Risk Assessment Matrix (RAM) shows events that could materially alter the baseline path (the scenario most likely to materialize in the view of IMF staff).  
- The RAM reflects staff views on the source of risks and overall level of concern as of the time of discussions with the authorities.  
- Non-mutually exclusive risks may interact and materialize jointly and amplify each other’s effects.

*People’s Republic of China—Hong Kong SAR: Appendix I. Risk Assessment Matrix*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2021/english/1hkgea2021002.pdf_
