## 1bhsea2019005

## Source details

**Canonical URL:** [1bhsea2019005](https://www.imf.org/-/media/files/publications/cr/2019/1bhsea2019005.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2019/1bhsea2019005.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2019/1bhsea2019005.pdf.json)

---

### Executive summary — macrofinancial risks and overall assessment
- Macrofinancial risks arise from vulnerability to external shocks to tourism and real estate, exposure to frequent and severe hurricanes, and a small, illiquid real estate market.
- Stress tests indicate the overall banking system is resilient to a range of adverse scenarios given large aggregate capital and liquidity buffers, but some domestic banks and the two largest credit unions are more vulnerable to asset quality shocks and tail risk conditions.
- Asset quality and profitability are key determinants of financial institutions’ resilience.
- Liquidity, market, sovereign and financial contagion risks are low. The offshore banking sector is not a source of traditional banking risks.

### Stress testing coverage, methodology, and scenarios
- Coverage:
  - All seven commercial banks and the two largest credit unions (representing 97 percent of the assets of domestic credit institutions).
  - Ten large offshore banks (61 percent of international banking assets), including stress tests for offshore banks with more traditional banking models.
- Methodology:
  - Scenario-based solvency stress tests cover credit risk only.
  - A small DSGE model plus satellite panel regressions for bank NPLs, ROA, and credit growth feed bank-level projections.
  - Panel regressions and DSGE use annual data (DSGE: 1976–2018; panel: balanced 2001–17, NPL regression excluding 2016–17).
- Macroeconomic scenarios (3-year horizon, 2019–21):
  - Baseline: IMF WEO October 2018 projections; domestic growth: 2.3 percent in 2018, 2.1 percent in 2019, 1.5 percent by 2020.
  - Natural disaster (major hurricane): losses of 22 percent of GDP at the 99th percentile; cumulative 3-year decline of real GDP relative to baseline equivalent to one standard deviation.
  - U.S. recession: calibrated on GFC history; cumulative 3-year decline relative to baseline equivalent to 1.5 standard deviations.
  - “Perfect storm”: combined U.S. recession and major hurricane (hurricane losses of 10 percent of GDP at the 95th percentile); calibrated to a 2-standard deviation 3-year cumulative GDP decline.
- DSGE and hurricane calibration:
  - Hurricane losses measured as damages to capital stock as percent of nominal GDP (data from Azevedo (2016), EM-DAT, updated 2016–18).
  - Estimated DSGE coefficients (selected): Real GDP growth (t-1): 0.343 (2.70)***; Growth in tourist arrivals (t): 0.302 (7.73)***; Hurricane losses to GDP (t) on growth in tourist arrivals: -0.469 (2.00)**. (T-values in parentheses; significance codes: * p<0.1; ** p<0.05; *** p<0.01.)
  - Hurricane impacts: an increase in hurricane damages by 1 percentage point of GDP → 0.47 percentage point drop in growth rate of tourist arrivals relative to baseline. Average hurricane damage in sample = 4 percent of GDP. Tail-risk shock of 22 percent of GDP → growth in tourist arrivals 10 percentage points lower relative to baseline (cumulative two-year decline).

### Main stress-test results — solvency and credit risk
- Baseline:
  - System-wide CAR increases to 32.7 percent by 2021, supported by projected stronger activity in 2018–19 then moderating.
  - System-wide CAR (as of September 2018 across seven commercial banks): 31.2 percent; all banks above the 17 percent target.
- Hurricane scenario:
  - Hurricane losses: 22 percent of GDP.
  - System-wide NPL ratio rises by 3 percentage points; CAR declines by 2.5–3 percentage points; all banks remain above 17 percent CAR target.
  - Aggregate NPLs increase by about 70 percent.
- U.S. recession scenario:
  - Spillovers push system NPL ratio up by 10 percentage points relative to baseline; aggregate NPLs increase by about 120 percent.
  - One domestic bank with large legacy loans and low profitability becomes undercapitalized (below 14 percent CAR trigger).
  - Domestic banks’ solvency deteriorates by 10 percentage points relative to baseline; foreign banks’ solvency weakens by about 5 percentage points.
- “Perfect storm” scenario:
  - System-wide CAR declines to 23 percent.
  - System-wide NPL ratio rises by 13 percentage points relative to baseline; aggregate NPLs increase by about 150 percent.
  - Two domestic banks representing 15 percent of banking assets fall below the 14 percent CAR threshold and need additional capital.
  - Total banking system capital shortfall estimated at 0.4 percent of 2018 GDP.
- Single-factor sensitivity and static tests:
  - Two banks fall below the 14 percent CAR trigger if existing NPLs increase by 150 percent.
  - Increase in NPLs equivalent to 10 percent of total loans drives two banks representing 32 percent of banking assets somewhat below 17 percent CAR, but above 14 percent.
  - Mortgage portfolio shocks dominate: mortgage NPLs up 150 percent would cause one bank (7 percent of assets) to breach 14 percent CAR.
  - Direct interest rate shocks (±300, ±500 basis points): system resilient to direct repricing, but indirect interest-rate induced credit risk significant due to prevalence of floating-rate loans.
  - Sovereign haircuts: haircuts above 25–30 percent could affect some banks with substantial domestic sovereign exposure.
  - Exchange rate shocks analyzed: 15 percent and 25 percent devaluation; direct FX exposure small given small net open FX positions.

### Liquidity, market, sovereign and contagion risk findings
- Liquidity:
  - Net eligible liquid assets reached 270 percent of required net liquid assets by end-June 2018; liquid assets to total assets = 30 percent at end-June 2018.
  - Liquidity stress-test assumptions: 25 percent of total non-equity funding withdrawn within a year, with 21 percent withdrawn in the first 90 days; roll-off rates on loans: 50 percent up to one month, 30 percent up to six months, 10 percent longer.
  - Under stress parameters, 21 percent of non-equity funding withdrawn within first 3 months; total withdrawal 25 percent by year-end.
  - Identified vulnerability: one bank (8 percent of banking assets) runs out of liquidity after a sustained 6-month deposit run; one large bank reliant on short-term deposits would need to repo significant government securities in first 3 months.
- Market risk:
  - Direct interest rate risk low; indirect interest-rate induced credit risk significant.
  - Small net open FX positions imply limited currency risk.
- Sovereign risk:
  - Government securities represent about 15–20 percent of banking assets (public sector credit share rose from 20 percent to 23.5 percent in Dec 2013–Jun 2018); two thirds of government holdings have maturities greater than one year.
  - Roll-over risk mitigated by long duration of bond portfolios.
- Contagion and network analysis:
  - Onshore-offshore segregation maintained by exchange controls.
  - Aggregate interlinkages among domestic banks: B$115 million (5 percent of aggregate capital).
  - Gross exposures in network: B$146.5 million or 6.5 percent of aggregate capital.
  - Contagion simulations (100 percent LGD on exposures): only one credit union vulnerable to potential default of its counterparties; offshore banks and insurers neither important sources nor recipients of spillover risk.

### Credit unions and offshore banking sector
- Credit unions:
  - Two largest credit unions (65 percent of credit union assets) have pro-forma average CAR = 16.3 percent; undercapitalized under 10 percent unweighted capital-to-asset threshold.
  - Slightly above 14 percent risk-weighted threshold (not applicable currently), but become undercapitalized under moderate credit shocks (e.g., 50 percent increase in NPLs).
  - Portfolios dominated by small retail loans; sensitivity to interest-rate risk low due to short-term assets and liabilities.
- Offshore banks:
  - Total offshore assets: US$256 billion in June 2018 (about 21 times GDP); Bahamas offshore assets (end-June 2018) include offshore banks total assets US$168 billion.
  - Most offshore activity is private banking and treasury/intragroup operations; limited traditional banking risks. For five largest offshore banks, fees = 75 percent of income.
  - Ten large offshore banks stress tested; results focused on two with more traditional banking business showing small risk exposures.
  - Offshore sector trends: total offshore assets down from US$427 billion at end-2013 to US$256 billion in June 2018.

### Structure of the domestic banking system and key metrics (selected)
- Banking system composition and funding:
  - Seven commercial banks; four foreign-owned banks account for roughly three-quarters of system assets.
  - Banks primarily funded by household and commercial deposits; foreign banks rely more on wholesale bank deposits and intra-group claims.
  - Approximately one fourth of foreign banks’ balance sheets represent operations with nonresidents.
- Credit composition (end-September 2018 / June 2018):
  - Mortgages comprised 60 percent of total credit to the private sector (residential mortgages 43 percent of total credit) at end-September 2018.
  - System-wide NPL ratio: 8.9 percent as of end-June 2018; peaked at 15.4 percent in 2014. Two domestic banks had NPL ratios above 15 percent at end-September 2018.
  - Sectoral shares of private credit (June 2018): Consumer credit 38 percent; Business credit 17 percent; Residential mortgages 43 percent; Commercial mortgages 2 percent.
- Capitalization and liquidity (as of Sep/Jun 2018):
  - Seven commercial banks’ CAR: 31.2 percent (Sep 2018).
  - Eligible liquid assets = 270 percent of required by end-June 2018; liquid assets = 30 percent of total assets at end-June 2018.
  - Domestic financial sector assets estimated at US$18 billion in June 2018.
- External dependence and real economy links:
  - Tourism estimated at 45 percent of GDP; real estate and construction contribute 22 percent of GDP.
  - Preliminary Ministry of Tourism data Jan–Sep 2018: 80 percent of stopover visitors were from the U.S.
  - Exchange rate fixed to the U.S. dollar; strict exchange controls limit cross-border liquidity flight and segregate offshore sector.

### Key recommendations (with timeframes and responsible authorities)
- Stress testing framework:
  - Enhance analytical capacity for solvency and liquidity risk assessment; strengthen stress test focus on systemic and macroeconomic risks. (ST) — Responsibility: CBOB.
  - Adopt solvency tests based on macroeconomic scenarios with multi-factor shocks building on FSAP methodologies. (ST) — Responsibility: CBOB.
  - Implement financial contagion stress tests. (MT) — Responsibility: CBOB.
  - Run periodic bottom-up stress tests with individual banks under consistent scenarios. (MT) — Responsibility: CBOB.
  - Request corrective action from credit unions to strengthen capital adequacy. (ST) — Responsibility: CBOB.
- Addressing data gaps:
  - Develop a real estate price index to improve market monitoring and collateral valuation. (ST) — Responsibility: CBOB/DOS.
  - Initiate collection of loan-level data on loan-to-value and debt-to-income ratios and household leverage. (ST/MT) — Responsibility: CBOB/DOS.
  - Improve data on loan classifications, provisioning, restructured loans, yields and duration of bonds, offshore banking sector and credit unions. (ST) — Responsibility: CBOB.
  - Operationalize the proposed Consumer Credit Bureau. (MT) — Responsibility: CBOB/MOF.
- Staffing and capacity:
  - Establish a CBOB interdepartmental financial stability group focused on quantitative analysis. (MT) — Responsibility: CBOB.
  - Conduct training and build capacity to perform stress tests using cutting-edge methodologies. (ST/MT) — Responsibility: CBOB.

### Selected numeric outcomes and parameters (preserved exactly)
- Hurricane scenario losses: 22 percent of GDP.
- Hurricane scenario NPL increase: about 70 percent (aggregate).
- U.S. recession NPL increase: about 120 percent (aggregate).
- Perfect storm NPL increase: about 150 percent (aggregate).
- System-wide CAR in perfect storm: 23 percent.
- Banks needing additional capital in perfect storm: two banks representing 15 percent of banking assets.
- Banking system total capital shortfall in perfect storm: 0.4 percent of 2018 GDP.
- Trigger CAR ratio used in analysis: 14 percent.
- Initial adjustment: assumes 10 percent of restructured loans in 2018 become nonperforming.
- LGD assumptions: Business loans 80 percent; Consumer loans 100 percent; Mortgage loans 50 percent.
- Sensitivity test LGD assumptions: 70 percent (general), 100 percent (sectoral shocks), 100 percent (unsecured interbank lending in contagion tests).
- Interest rate shocks analyzed: -300, -500, +300, +500 basis points.
- Exchange rate shocks analyzed: 15 percent and 25 percent devaluation.
- Sovereign haircut scenarios analyzed: 10 percent, 20 percent, 30 percent.
- Run-off / withdrawal assumptions in liquidity tests: 25 percent of total non-equity funding withdrawn within a year, with 21 percent withdrawn in the first 90 days; 21 percent withdrawn within first 3 months; total 25 percent by the end of the year.
- Offshore sector size (selected): Total offshore assets US$256 billion (June 2018); Offshore banks total assets US$168 billion; Investment funds total assets ~US$86 billion.
- Onshore-offshore exposures: around 5.8 percent of domestic banking sector assets held in offshore counterparties (US$725 million) as of September 2018.
- Aggregate interlinkages among domestic banks: B$115 million (5 percent of aggregate capital).
- Gross exposures in the network: B$146.5 million (6.5 percent of aggregate capital).

*Source: 1bhsea2019005 — IMF staff analysis (FSAP) — EXECUTIVE SUMMARY and chapters on banking system structure, stress testing, offshore sector, liquidity, and network analysis.*

### EXECUTIVE SUMMARY ___________________________________________________________________________ 5

### EXECUTIVE SUMMARY

### Key findings
- Macrofinancial risks stem from the economy’s vulnerability to external shocks to tourism and real estate investment, exposure to frequent and severe hurricanes, and a small and illiquid real estate market.  
- Stress tests reveal the overall banking system is resilient to a range of adverse scenarios given large aggregate capital and liquidity buffers.  
- Some domestic banks and the two largest credit unions are more vulnerable to asset quality shocks and tail risk conditions.  
- Asset quality and profitability are key determinants of financial institutions’ resilience to adverse shocks.  
- Liquidity, market, sovereign and financial contagion risks are low.  
- The offshore banking sector is not a source of traditional banking risks.

### Stress testing coverage and methodology
- The FSAP stress testing exercise covered:
  - All seven commercial banks and the two largest credit unions (representing 97 percent of the assets of domestic credit institutions).  
  - Ten large offshore banks (61 percent of international banking assets), including stress tests for offshore banks with more traditional banking models.
- Scenario-based solvency stress tests cover credit risk only.  
- Macroeconomic scenarios and transmission:
  - A small dynamic stochastic general equilibrium (DSGE) model and satellite models for bank credit quality, profitability and credit growth were used to estimate impacts on tourist arrivals, real GDP growth, inflation, and the short-term interest rate.  
  - Macroeconomic projections feed into panel regressions forecasting key bank-level drivers of bank solvency under each scenario.

### Macroeconomic scenarios assessed
- Three adverse scenarios assess banking system resilience to key risk factors:
  - Natural disaster (major hurricane): losses of 22 percent of GDP at the 99th percentile of the historical distribution; models negative impacts on tourism, employment, and bank asset quality.  
  - U.S. recession: steep and protracted drop in U.S. demand for tourism and investment services, causing a deep domestic slowdown, higher unemployment, sharp correction in real estate prices, and significant bank losses.  
  - “Perfect storm”: combined U.S. recession and major hurricane; the hurricane shock is less severe than in the first scenario (losses of 10 percent of GDP at the 95th percentile of the historical distribution).  
- The decline in real GDP growth in scenarios was calibrated not to exceed two standard deviations.

### Main results — solvency and credit risk
- Baseline projection:
  - System-wide CAR increases to 32.7 percent by 2021, supported by projected stronger economic activity in 2018–19 but growth subsequently moderates.
- Hurricane scenario:
  - A short period of stress weakens banks’ performance but leaves all banks above the minimum capital requirement.  
  - System-wide NPL ratio rises by 3 percentage points and CAR declines by 2.5 percentage points.
- U.S. recession scenario:
  - One domestic bank with a large portfolio of legacy loans and low profitability becomes undercapitalized.
- “Perfect storm” scenario:
  - Combined shock leaves two domestic banks (15 percent of the system’s assets) below the 14 percent CAR threshold.
- Single factor sensitivity analysis confirms the importance of credit risk for financial stability.

### Liquidity, market, sovereign and contagion risks
- Liquidity:
  - Despite excess liquidity accumulated during the credit slowdown, one bank (7 percent of the system’s assets) with lower than average liquid assets runs out of liquidity after a sustained six-month deposit run.
- Market risk:
  - Direct sensitivity to interest rate risk is low but indirect interest-rate induced credit risk is significant.
  - Small net open FX positions imply limited currency risk.
- Sovereign risk:
  - Modest government bond holdings by most banks limit sovereign risk, except under extreme sovereign default scenarios, where some banks with more substantial sovereign exposure could be affected.
- Financial contagion:
  - The contagion analysis focused on banking and insurance sectors and assessed resilience to spillovers from a default of a major counterparty.
  - One credit union is vulnerable to defaults of major counterparties but will not generate spillovers to other institutions.
  - Offshore banks and insurers are neither an important source nor a recipient of spillover risk within the financial sector.

### Credit unions
- Solvency tests for the two largest credit unions:
  - Undercapitalized under the 10 percent unweighted capital-to-asset threshold.
  - Slightly above the 14 percent risk-weighted threshold (which currently does not apply to credit unions), but become undercapitalized under moderate credit shocks.
  - Portfolios dominated by small retail loans limit credit concentration risk.
  - Sensitivity to interest rate risk is low given mainly short-term assets and liabilities.

### Offshore banking sector
- Traditional banking risks are small among the ten large offshore banks assessed; only two are engaged in more traditional banking, with stress test results indicating small risk exposures.
- Offshore banks were not covered by the macroeconomic scenarios due to lack of long time series and limited exposure to traditional banking risks.

### Vulnerabilities and structural features
- The economy is highly dependent on cyclical tourism and real estate:
  - Tourism is estimated to account for 45 percent of GDP and real estate and construction activities contribute 22 percent of GDP.
  - Preliminary Ministry of Tourism data for January-September 2018 indicate that 80 percent of total stopover visitors were from the U.S.
- Policy framework constraints:
  - The exchange rate is fixed to the U.S. dollar.
  - Capital controls provide some scope for independent monetary policy, but the transmission mechanism appears weak.
  - Large developmental needs reduce fiscal flexibility.
- Banking system structure:
  - Seven commercial banks; four foreign-owned banks account for roughly three-quarters of the system’s assets.
  - Banks were grouped into two peer groups: three foreign banks operating with residents and nonresidents under a consolidated balance sheet form the “foreign bank” group; four banks operating only with residents form the “domestic banks” group.

### Key recommendations (Table 1)
- Stress testing framework (Short term = 1–3 years; Medium term = 3–5 years):
  - Enhance analytical capacity for assessing solvency and liquidity risks; strengthen the focus of stress tests on key systemic and macroeconomic risks. (ST) — Responsibility: CBOB.  
  - Adopt solvency tests based on macroeconomic scenarios with multi-factor shocks building on FSAP methodologies. (ST) — Responsibility: CBOB.  
  - Implement financial contagion stress tests. (MT) — Responsibility: CBOB.  
  - Run periodic bottom-up stress tests with individual banks under consistent scenarios. (MT) — Responsibility: CBOB.  
  - Request corrective action from credit unions to strengthen capital adequacy. (ST) — Responsibility: CBOB.
- Addressing data gaps:
  - Develop a real estate price index to enhance market monitoring and collateral valuation. (ST) — Responsibility: CBOB/Department of Statistics (DOS).  
  - Initiate collection of loan-level data on loan-to-value and debt-to-income ratios and household leverage. (ST/MT) — Responsibility: CBOB/DOS.  
  - Improve data collection on loan classifications and provisioning, restructured loans, yields and duration of bonds in banks’ portfolios, offshore banking sector and credit unions. (ST) — Responsibility: CBOB.  
  - Operationalize the proposed Consumer Credit Bureau. (MT) — Responsibility: CBOB/MOF.
- Staffing and capacity:
  - Establish a CBOB interdepartmental financial stability group focused on quantitative analysis. (MT) — Responsibility: CBOB.  
  - Conduct training and build capacity to perform stress tests based on cutting-edge methodologies. (ST/MT) — Responsibility: CBOB.

*Source: 1bhsea2019005 - EXECUTIVE SUMMARY.*

### 11.      The domestic banking system is primarily engaged in conventional banking

### 1bhsea2019005 - 11.      The domestic banking system is primarily engaged in conventional banking

### Structure and funding of the banking system
- The domestic banking system is primarily engaged in conventional banking services.
- Banks are funded mainly with household and commercial deposits and lend to the private sector and the government.
- Foreign banks rely more for funding on wholesale bank deposits than domestic banks and invest more in claims on other banks, mainly related entities.
- Approximately one fourth of foreign banks’ balance sheets represent operations with nonresidents.
- Mortgages comprised 60 percent of total credit to the private sector (residential mortgages represented 43 percent of total credit) at end-September 2018.
- Government securities accounted for less than 20 percent of banking assets at end-September 2018.
- Despite large exposure to mortgages, banks’ maturity mismatches are moderate due to the prevalence of variable-rate loans.

### Capitalization, profitability, and income composition
- System-wide position (as of September 2018):
  - Capital adequacy ratio (CAR) of the seven commercial banks: 31.2 percent.
  - All banks were well above the 17 percent target ratio.
- Foreign banks experienced some reduction in CARs due to growth in risk-weighted assets, albeit from high initial levels.
- Domestic banks’ CARs benefited from a decline in risk-weighted assets and a capital injection at one bank.
- System-wide profitability remained stable.
  - Domestic banks rely more on net interest income.
  - Foreign banks rely more on noninterest income.
- Profitability is more uneven among domestic banks.

### Asset quality and provisioning risks
- System-wide nonperforming loan (NPL) ratio:
  - 8.9 percent as of end-June 2018.
  - Peaked at 15.4 percent in 2014.
- Weaknesses are more pronounced among domestic banks; two banks had NPL ratios above 15 percent at end-September 2018.
- Risks in the credit market are compounded by uncertainty regarding loan recovery values:
  - Collateral valuations are not updated by periodic appraisals.
  - The real estate market is small and prone to illiquidity; challenges include lack of comparable sales, limited transaction disclosure requirements, and no published index of local housing prices.
  - The Home Owners Protection Act, 2017 increased protection of homeowners, leading to a prolonged foreclosure process and the chance of significant write-downs.
- Given a soft medium-term outlook, asset quality problems are likely to continue to weigh on some banks’ performance.

### Credit market behavior and interest rate sensitivity
- Commercial credit declined significantly following the GFC.
- From 2012 to June 2018 bank credit to the private sector dropped by 11 percentage points.
- Sluggish credit provision reflects heightened risk aversion and informational asymmetries:
  - Absence of a credit bureau for households.
  - Lack of financial statements for SMEs.
- Interest rate exposure:
  - Direct sensitivity to interest rate risk is low because banks have small maturity mismatches due to prevalence of variable-rate loans and deposits.
  - The system-wide cumulative maturity gap is positive, implying that a rise in interest rates would on average boost net interest margins.
  - However, higher interest rates would increase borrowers’ debt-servicing burden and banks’ credit losses; given small maturity mismatches and strong NPL sensitivity to interest rate shocks, credit-loss effects are likely to dominate.

### Exposure to government debt and liquidity
- Public sector credit share in domestic banking assets increased from 20 percent to 23.5 percent in the December 2013–June 2018 period.
- Government securities represent about 15 percent of total assets, of which two thirds are with maturities greater than one year.
- Rollover risk is mitigated by the relatively long duration of banks’ bond portfolios.
- Liquidity accumulation:
  - Net eligible liquid assets reached 270 percent of required net liquid assets by end-June 2018.
  - Ratio of liquid assets to total assets stood at 30 percent at end-June 2018.
  - Strict exchange controls limit the risk of cross-border liquidity flight.

### Offshore banking sector
- Analysis focused on the business models and balance sheets of the ten largest offshore institutions.
- Of the ten large offshore banks:
  - Two are primarily engaged in wholesale banking (mainly intra-group Treasury operations).
  - Six are active in private banking.
- All 10 institutions were stress tested; reported results focused on two banks engaged in more traditional banking activities.

### Stress-testing framework: scenarios, horizon, and modeling
- Time horizon and data:
  - Analysis conducted on data as of September 2018 over a 3-year horizon (2019–21).
- Scenarios:
  - Baseline scenario:
    - Based on IMF WEO staff projections as of October 2018.
    - Assumes domestic growth accelerates to 2.3 percent in 2018 and 2.1 percent in 2019, then moderates to 1.5 percent by 2020.
    - Improved U.S. consumer and business confidence support travel and real estate investment; near-term inflationary pressures contained; real interest rates remain steady.
  - Natural disaster (hurricane) scenario:
    - Gauges effects of major hurricanes on tourist arrivals, GDP growth and bank performance.
    - Replicates two major hurricanes, Francis and Jeanne, which hit The Bahamas in September 2004, inflicting combined damage estimated at 22 percent of GDP.
    - Results in a cumulative 3-year decline of real GDP relative to the baseline equivalent to one standard deviation.
  - U.S. recession scenario:
    - Models contagion from a severe U.S. recession via tourism and real estate channels.
    - Shocks calibrated based on historical data for the GFC.
    - Results in a cumulative 3-year decline of real GDP relative to the baseline equivalent to 1.5 standard deviations.
  - “Perfect storm” scenario:
    - Simulates simultaneous effects of a U.S. recession and a major hurricane.
    - Hurricane shock defined as losses of 10 percent of GDP (95th percentile of the historical loss distribution).
    - Calibrated to lead to a 2-standard deviation 3-year cumulative GDP decline relative to the baseline.
- Modeling approach:
  - A small DSGE model estimated to calibrate the impact of exogenous shocks (U.S. growth, U.S. property prices, hurricane losses) on domestic tourist arrivals, real GDP growth, consumer price inflation, and the short-term interest rate.
  - Projections feed into panel regressions that project key bank-level drivers of solvency and ultimately bank CARs; scenarios consider second-round contagion and feedback effects from credit growth to real activity.
  - Novelty: use of a variable that measures the severity of hurricane losses to GDP to calibrate effects on tourist arrivals, real GDP growth, bank NPLs and real credit growth.
  - DSGE model specification (parsimonious, annual data 1976–18) includes four equations for real GDP growth (Y), percent change in stopover visitor arrivals (TTA), consumer price inflation (INF), and the nominal T-bill rate (INT), with explanatory variables including U.S. real GDP growth and a hurricane-losss-to-GDP ratio.

*Source: 1bhsea2019005 - 11.      The domestic banking system is primarily engaged in conventional banking (PDF chapter/section).*

### 27.       The impact of hurricanes on growth and tourist arrivals is calibrated using a

### 1bhsea2019005 - 27.       The impact of hurricanes on growth and tourist arrivals is calibrated using a

### Hurricane shock calibration and data sources
- The hurricane shock is represented by a novel variable measuring actual damages to the capital stock as a percentage of nominal GDP.
- Data sources: Azevedo (2016) and the International Disaster Database EM-DAT (Guha-Sapir, Hoyois and Below, 2015); series updated from news reports and U.S. National Hurricane Center for 2016–18.
- Rationale: This approach modulates the shock according to scenario severity and focuses analysis on tail risk (losses to GDP variable better suited to model extreme shocks than dummy variables).

### DSGE model estimation and structural parameters
- Structural parameters estimated by maximum likelihood using a Kalman filter to form the log-likelihood function.
- Shocks assumed i.i.d. (but not normally distributed); estimation conducted with robust standard errors to ensure consistency under the weaker distributional assumption.
- Sample period for DSGE estimates: 1976-2018.
- Key estimated relationships (consistent with theory and Bahamian features):
  - GDP growth driven by visitor arrivals and foreign real estate investment (proxied by buoyancy of U.S. housing prices, Florida index more closely associated than broader U.S. index).
  - Visitor arrivals strongly related to U.S. GDP growth and momentum in U.S. housing market.
  - Hurricanes significantly impact visitor arrivals and GDP growth.
  - Inflation process persistent and relatively backward looking; expectations play a limited role.
  - Muted response of interest rates to inflation (weak policy transmission).

- Selected DSGE coefficient examples (as reported):
  - Real GDP growth (t-1): 0.343 (2.70)***
  - Growth in tourist arrivals (t): 0.302 (7.73)***
  - Change in U.S. housing prices (t) on Growth in tourist arrivals: 0.061 (1.82)*
  - Hurricane losses to GDP (t) on Growth in tourist arrivals: -0.469 (2.00)**
  - Inflation (t-1): 0.746 (9.91)***
  - Inflation (t): 0.582 (5.65)***
- Note: T-values in parentheses; significance codes: * p<0.1; ** p<0.05; *** p<0.01. All variables expressed as percentages.

### Hurricane impact estimates on tourist arrivals and GDP
- Tourist arrivals:
  - An increase in hurricane damages by 1 percentage point of GDP leads to a 0.47 percentage point drop in the growth rate of tourist arrivals relative to the baseline.
  - The average hurricane damage in the sample is 4 percent of GDP, implying a 2-percentage point difference with the growth in tourist arrivals in the baseline.
  - Under a tail risk shock of 22 percent of GDP (more than 4 standard deviations above the mean), growth in tourist arrivals would be 10 percentage points lower relative to the baseline.
  - Note: This is a cumulative two-year decline in arrivals because hurricane losses are spread over two years.

- Real GDP growth:
  - An average hurricane would reduce real GDP growth by about 0.6 percentage points relative to the baseline.
  - An extreme shock at the far end of the distribution would cause real GDP growth to deviate from the baseline path by 3.3 percentage points.
  - Comparable literature: Strobl (2011) finds declines of 0.45 and 3 percentage points (per capita basis) for average and extreme hurricanes, respectively.

### Shock persistence and scenario calibration
- Impulse response functions show strong shock persistence: effects of one-standard deviation shocks on domestic GDP growth unwind over a 3-year period; more severe shocks take longer to peter out.
- Shocks carried over through lags of domestic growth and inflation; in-sample forecasts closely replicate key domestic variables during the GFC conditional on actual U.S. data.
- Scenarios:
  - Model calibrated to generate the October 2018 IMF WEO baseline; deviations in adverse scenarios driven by exogenous shocks to hurricane losses, U.S. growth, and U.S. housing prices.
  - The model has three external exogenous variables: ratio of hurricane losses to GDP, U.S. GDP growth, and percent change in U.S. housing prices.
  - Adverse hurricane losses correspond to the tail of the historical distribution; U.S. variable shocks calibrated on historical data for the GFC.

- “Perfect storm” scenario contributions (2019–21):
  - Among external factors, both hurricanes and U.S. housing prices explain about 25 percent of the decline in GDP each; U.S. real GDP growth explains 17.5 percent.
  - Two thirds of the decline in GDP is due to external factors and one third to domestic factors in the “perfect storm” scenario.
  - Domestic amplification: lagged GDP growth explains about 25 percent of the total cumulative GDP slowdown.

### Satellite models, panel methodology, and dataset
- Panel models model bank-level NPLs, ROA, and credit growth as functions of lagged dependent variables and macro controls with bank fixed effects.
- NPL dependent variable: logit transformation of the NPL ratio into a credit quality index (Virolainen, 2004); higher index values imply better credit quality.
- Dynamic specification addressed with Arellano-Bond GMM, but final specifications based on fixed effect estimation because lagged dependent variables proved insignificant.
- Dataset: balanced panel with annual data for 2001-17. NPL regression excluded period 2016–17 due to direct NPL sales to asset management companies unrelated to macro fundamentals.
- Credit quality index was nonstationary in levels and included in regressions in first difference.

### Panel regression findings on bank performance and macro links
- NPLs:
  - A one percentage point drop in GDP growth leads to a 0.75 percentage point increase in the NPL ratio.
  - A one percentage point increase in the real rate is associated with a 1.2-percentage point increase in the NPL ratio.
  - Interaction: Hurricane losses with GDP growth statistically significant at the 10 percent level—suggests reinforcing effects of hurricanes and recessions on credit quality (caution: based on small number of observations).
- ROA and credit growth:
  - GDP growth has significant and economically important effects on ROA and real credit growth (the latter with a one-year lag).
  - NPLs are an important driver of bank ROA and real credit growth.
  - Bank income and credit growth show low sensitivity to interest rate shocks once GDP growth and NPLs are controlled for—reflecting close matching of asset and liability maturities and positive overall asset-liability gap.
- Reinforcing feedbacks:
  - Estimates point to reinforcing effects between GDP, real credit and NPLs; feedback appears stronger from real GDP to credit than vice versa.
- Hurricanes in panel regressions:
  - In NPL regression, hurricanes appear to have a negative impact on NPLs during recessions (short-term tourism interruptions that do not trigger recession less likely to increase mortgage/consumer losses).
  - In credit growth regression, hurricanes have some positive effect on credit growth, possibly due to reconstruction-related loan demand.
- Reverse causality checks:
  - Panel Granger causality tests (one and two lags) did not detect reverse causality from dependent variables to regressors at conventional significance levels.

### Bank solvency stress test assumptions and mechanics
- Loss given default (LGD) assumptions by loan type:
  - Business loans: 80 percent.
  - Consumer loans: 100 percent.
  - Mortgage loans: 50 percent.
  - Average LGD varies from the low 60 percent for banks with large mortgage portfolios to 90–95 percent for banks specialized in consumer lending.
- Credit losses:
  - Calculated as incremental increase in projected NPL levels (from regression forecasts) multiplied by assumed LGD.
  - NPLs equal projected NPL ratio multiplied by the credit projection; adjusted for credit losses from restructured loans.
- Restructured loans:
  - Actual restructured loans as of September 2018 added to initial positions.
  - Assumed that 10 percent of these loans would fall back into NPLs by end-2018; banks’ 2018 capital adjusted for additional provisions.
  - Subsequently, new NPLs from restructured loans modeled as a fixed fraction of projected NPLs each period, varying from 5 to 20 percent depending on scenario; converted into credit losses using same LGD assumptions.
- Profits and dividends:
  - Profits derived from regression projections for ROA and projected average assets.
  - Total assets assumed constant (constant balance sheet assumption).
  - Dividends: banks pay dividends if profitable, amounting to 50 percent of bottom-line profits (determined from discussions with banks).
- Risk-weighted assets:
  - Change in risk-weighted assets calculated as incremental increase in credit multiplied by average weight of credit in risk-weighted assets, about 70 percent.

### Macro scenario and stress-test results for the banking system
- Initial capital positions:
  - Assessment constrained by data limitations; provisioning coverage difficult to analyze due to gaps in loan classifications, provisioning and collateral valuations.
  - Composition of regulatory capital indicates high share of common equity for most banks.
- Adjustment for restructured loans:
  - Banks’ initial CARs adjusted for potential reversals of restructured loans into NPLs.
- System resilience and vulnerabilities:
  - Overall banking system resilient to a range of adverse shocks, but pockets of vulnerabilities remain and widen under severe stress.
  - Banks’ initial NPL levels and income-generating capacity are key drivers of resilience.
  - Despite high current capital ratios, banks with high NPLs can experience capital degradation under adverse scenarios, resulting in some capital breaches.

*Source: IMF chapter text (1bhsea2019005 - 27).*

### 49.      The estimates indicate that U.S. recessions impact growth and bank credit

### 49.      The estimates indicate that U.S. recessions impact growth and bank credit

### Key findings
- U.S. recessions affect growth and bank credit quality through tourism and real estate channels; nonlinearities in the effects of hurricanes on NPLs appear to amplify bank losses during downturns.
- Panel regression analysis suggests that hurricanes’ impact on NPLs could increase in severe recessions.
- Banks show resilience to small-to-medium shocks but vulnerability in tail-risk conditions, with specific institutions and credit unions notably exposed.

### Baseline scenario
- System-wide CAR increases on the back of stronger economic activity and slow accumulation of risk-weighted assets.
- Banks’ NPL ratios initially decline then increase modestly in outer years when economic activity slows.
- One bank with poor asset quality and low profitability sees its capital ratio decline over time but remains well above capital requirements.

### Major Hurricane scenario
- A short period of stress from a major hurricane weakens bank performance but does not push any banks below the 17 percent CAR target.
- The hurricane scenario incurs losses equivalent to 22 percent of GDP.
- The system-wide NPL ratio rises by 3 percentage points relative to the baseline.
- This shock implies an increase in aggregate NPLs of about 70 percent.
- System-wide CAR declines by about 3 percentage points, with weaker banks’ capital ratios declining but remaining above the 17 percent target.

### U.S. Recession scenario
- A steeper growth slowdown driven by a U.S. recession drives one bank with a large portfolio of legacy loans and low profitability below the 14 percent CAR trigger.
- Spillovers push up the system’s NPL ratio by 10 percentage points relative to the baseline.
- This shock implies an increase in aggregate NPLs of about 120 percent.
- Banks’ capital buffers are reduced by increased provisions, write-offs and falling profitability.
- Domestic banks with higher initial NPL ratios: solvency position deteriorates by 10 percentage points relative to the baseline.
- Foreign banks: solvency position weakens by about 5 percentage points.

### Perfect Storm (tail risk) scenario
- Combined effects of a deep and protracted recession, higher real interest rates and large hurricane losses significantly weaken bank performance.
- System-wide CAR declines to 23 percent.
- Two banks (15 percent of banking assets) need additional capital.
- System-wide NPL ratio rises by 13 percentage points relative to the baseline.
- This shock leads to an increase in aggregate NPLs of about 150 percent.
- Bank interest income declines sharply as rising NPLs squeeze profit margins; increased provisions and write-offs weigh on profits.
- Hurricane shock further amplifies falls in tourist arrivals, growth, employment and the rise in NPLs.
- Given high initial capital buffers, the banking system’s total capital shortfall is estimated at 0.4 percent of 2018 GDP.

### Second-round interbank contagion
- Tail risk scenario generates limited interbank contagion.
- One bank is exposed to a vulnerable institution but can absorb the potential loss without becoming undercapitalized.
- One credit union is potentially at risk of facing large deposit losses in a vulnerable bank, which could erode its capital.
- Tests assume LGD of 100 percent for unsecured interbank lending.

### Solvency sensitivity tests — Methodology
- Sensitivity tests assess resilience to instantaneous single-factor shocks; static projections hold other factors constant.
- Tests assume no profit buffer and hold risk-weighted assets constant.
- Credit risk tests: increases in NPLs (proportional to existing NPLs or to total loans), sectoral NPL shocks, and defaults of largest borrowers.
- Market risk tests: interest rate shocks on net interest income using a repricing gap framework and FX devaluation shocks on net open FX positions.
- Uniform LGD of 70 percent for domestic and international banks and credit unions in stress tests based on total credit portfolio and large exposures; uniform LGD of 100 percent for sectoral shocks.

### Solvency sensitivity tests — Results for commercial banks
- Two banks fall below the 14 percent CAR trigger if existing NPLs increase by 150 percent.
- If shocks are proportional to total loans, an increase in NPLs equivalent to 10 percent of total loans drives two banks representing 32 percent of banking assets somewhat below the 17 percent CAR target but all banks remain above the 14 percent trigger.
- Mortgage portfolio shocks dominate: an increase in mortgage NPLs of 150 percent would cause one bank (7 percent of banking assets) to breach the 14 percent capital trigger.
- Exposure to large-borrower concentration: foreign banks are more exposed, but simultaneous default of their five largest borrowers would not leave any banks undercapitalized.
- Direct interest rate risk: banking system overall resilient to direct interest rate shocks; CARs remain well above domestic trigger and target ratios after 500 basis point shocks.
- Indirect interest-rate induced credit risk is significant given prevalence of floating rate loans; interest rates are an important determinant of NPLs.
- Sovereign risk: modest sensitivity for most banks; extreme sovereign haircuts above 25-30 percent could affect some banks with substantial domestic sovereign bond exposure.
- Direct exchange rate risk is low given small net open FX positions; regulatory limits on net open FX positions are the minimum of 5 percent of bank Tier 1 capital and B$5 million.

### Solvency sensitivity tests — Results for credit unions
- Two largest credit unions have weak capital buffers and NPL ratios above 10 percent of total loans; they represent 65 percent of credit union sector’s assets.
- Estimated average CAR of the two credit unions is 16.3 percent (pro-forma).
- The two credit unions become undercapitalized under a 50 percent increase in NPLs.
- Credit unions are more resilient to credit concentration and interest rate risks than domestic banks due to predominant investment in short-term consumer loans.

### Solvency sensitivity tests — Results for offshore banks
- Offshore banks show low risks from traditional banking activities: low NPL ratios and low levels of credit to the private sector.
- An increase in NPLs equivalent to 10 percent of total loans would not have a large impact on their solvency.
- Offshore banks can withstand a 15 percent deterioration in quality of corporate loans.
- One offshore bank exposed to credit concentration would see its capital-to-asset ratio approach the 5 percent hurdle if its five largest borrowers default.
- Small maturity mismatches lead to low sensitivity to interest rate risk.

### Selected numeric outcomes (preserve source figures)
- Hurricane scenario losses: 22 percent of GDP.
- Hurricane shock NPL increase: about 70 percent (aggregate).
- U.S. recession shock NPL increase: about 120 percent (aggregate).
- Perfect storm shock NPL increase: about 150 percent (aggregate).
- System-wide CAR in perfect storm: 23 percent.
- Banks needing additional capital in perfect storm: two banks representing 15 percent of banking assets.
- Banking system total capital shortfall in perfect storm: 0.4 percent of 2018 GDP.
- Trigger CAR ratio used in analysis: 14 percent.
- Recapitalization shortfall measure: estimated capital injection needed to restore CAR to 14 percent.
- Initial adjustment: assumes 10 percent of restructured loans in 2018 become nonperforming.
- Sensitivity test LGD assumptions: 70 percent (general), 100 percent (sectoral shocks), 100 percent (unsecured interbank lending in contagion tests).
- Interest rate shock magnitudes analyzed: -300, -500, +300, +500 basis points.
- Exchange rate shocks analyzed: 15 percent and 25 percent devaluation.
- Sovereign haircut scenarios analyzed: 10 percent, 20 percent, 30 percent.

*Source: The Bahamas — IMF staff analysis (FSAP) extracted from the chapter on macroeconomic and solvency stress tests.*

### 69.      The FSAP team conducted liquidity stress tests based on the maturity ladder

### 69. The FSAP team conducted liquidity stress tests based on the maturity ladder of bank assets and liabilities

### Liquidity stress tests — methodology
- Tests based on maturity ladder contractual cash flow analysis with assumptions for run-off rates on funding sources and roll-off rates on assets (Appendix V Table 1).
- Run-off rates: percentage of liability exposure maturing in a given period that is withdrawn (and not rolled over).
- Roll-off rates: percentage of asset exposure maturing that is converted into a cash inflow (and not rolled over) by the bank.
- Banks finance negative funding gaps by drawing down liquid assets; a bank fails the test if it runs out of liquidity and needs emergency liquidity assistance from CBOB.
- Data limitations: analysis could not distinguish between different currencies and resident and nonresident deposits.

### Key stress-test assumptions and parameters
- Shock duration and profile:
  - Funding shock assumed persistent, lasting for up to 1 year.
  - Intensity of run-off rates higher in shorter maturity buckets (up to 6 months) and declines afterwards.
  - Actual outflows peak within 1 month due to concentration of deposits in that maturity.
- Run-off rates by liability type:
  - Deposit liabilities have lower run-off rates than unsecured and wholesale funding.
  - Lack of deposit breakdowns by maturity prevented varying run-off rates by deposit types.
- Roll-off rates on performing loans (standard stress-test assumption):
  - 50 percent of loan amounts of shorter maturities (up to one month).
  - 30 percent for maturities up to six months.
  - 10 percent for longer maturities.
- Liquid assets definition and treatment:
  - Banks’ liquid assets comprise cash, excess reserves at the CBOB, deposits with financial institutions and government securities.
  - Government securities are assumed to be rediscounted (repoed) through the CBOB facility at a minimum haircut of 15 percent as required by CBOB regulations (Appendix V Table 1).
- Test parameter on total withdrawals:
  - Assumes that 25 percent of total non-equity funding is withdrawn within a year, with 21 of it withdrawn in the first 90 days (Note in Table 8).

### Liquidity stress-test results and implications
- Systemwide projected withdrawals under test parameters:
  - 21 percent of non-equity funding withdrawn within the first 3-months.
  - Total withdrawal reaches 25 percent by the end of the year.
- Identified vulnerabilities:
  - A sustained run would cause a shortfall in one bank representing 8 percent of banking assets, occurring after a prolonged 6-month period.
  - Most banks have large liquidity buffers; the affected bank has lower liquidity buffers and a large portfolio of illiquid long-term loans.
  - One large bank reliant on short-term deposit liabilities would need to repo a significant fraction of its government securities during the first 3 months to meet projected outflows.
- Mechanics of banks’ responses:
  - Banks offset withdrawals with cash inflows from asset portfolios and cover shortfalls with cash, excess reserves and repos of government securities, subject to valuation haircuts.

### Financial interconnectedness — onshore vs offshore (key statistics)
- Segregation and cross-border exposures:
  - Onshore and offshore financial sectors effectively segregated by strict exchange controls.
  - In September 2018, around 5.8 percent of domestic banking sector assets were held in offshore counterparties (US$725 million), most as foreign banks’ intragroup transactions with related offshore entities.
- Onshore interconnectedness and exposures:
  - Aggregate interlinkages among domestic banks: B$115 million, equivalent to 5 percent of aggregate capital.
  - Total onshore banks’ gross claims on NBFIs: B$62 million, roughly 0.5 percent of aggregate assets in the onshore sector.
  - Banks’ liabilities to insurers: B$83 million and mostly consisted of deposits.
- Network analysis calibration parameters (Table 9):
  - λ = 1
  - ρ = 1
  - δ = 0.3
  - Default assumptions: 100 percent loss given default on exposures; 100 percent share of lost funding that is non-replaceable; 30 percent discount on asset sales.
  - Capital definition: For banks, total regulatory capital; for credit unions, book value of total capital; capital falls to zero implies default.
- Network analysis scope and sample:
  - Data from September 2018.
  - Model includes seven onshore banks, the two largest credit unions, eight large offshore banks, and the ten largest domestic insurance companies.
- Network analysis results:
  - Gross exposures in the network: B$146.5 million or 6.5 percent of aggregate capital.
  - In the bank domain, most domestic banks are connected to only one other domestic bank with interlinkage volumes usually below 1 percent of a bank’s total regulatory capital.
  - Contagion simulation: only one credit union is vulnerable to potential default of its counterparties because deposits it holds with two onshore commercial banks each exceeds its capital.
  - Total exposure between domestic banks and insurers in the network analyzed: around B$340 million, equivalent to 12.7 percent of aggregate capital in the network.
- Conclusion on interconnectedness:
  - Offshore banks and insurers are neither important sources nor recipients of spillover risk within the domestic financial sector; large offshore banks show no interconnections among themselves and very low interlinkages to onshore banks.
  - Domestic insurers’ interconnectedness mainly via deposits with domestic banks; insurers’ deposits are spread across multiple banks, limiting concentration risk.

### Offshore financial sector — size, structure, and business model (key statistics)
- Size and trend:
  - Total offshore assets: US$427 billion at end-2013 down to US$256 billion in June 2018.
  - Offshore assets in June 2018 equal approximately 21 times GDP.
  - The Bahamas ranks fourth in banking assets among 19 countries classified by the BIS as “offshore centers”; Bahamian assets total US$119 billion or 2.5 percent of the aggregate US$4.6 trillion in this peer group.
- Composition of offshore sector:
  - Offshore banks’ total assets: US$168 billion.
  - Investment funds total assets: around US$86 billion.
  - Insurance sector total assets: US$1 billion.
- Business model and activities:
  - Offshore banks primarily provide private banking and trust services, or conduct treasury operations for affiliated institutions; limited traditional banking activities such as term lending or liquidity transformation.
  - Private banking institutions hold substantial off-balance sheet fiduciary assets and relatively smaller amounts of mostly liquid on-balance sheet assets.
  - For the five largest offshore banks, fees represent 75 percent of income.
  - Lending in private banking primarily margin lending against investment portfolios; some small mortgage lending for high-end Bahamian real estate.
  - Treasury-focused offshore entities perform consolidated treasury functions and invest aggregated surplus cash; branches may book wholesale funding in the Bahamian entity and pass it on to clients in other jurisdictions.
- Cross-border linkages:
  - BIS locational bank statistics (end-June 2018): about two thirds of total claims and liabilities of banks to nonresidents were vis-à-vis banks, of which 70 percent were intragroup.
  - Claims mainly on advanced economies, offshore centers, and Latin America; advanced economies predominantly net suppliers of funds.
- Investment funds — NAV and entity counts (selected figures)
  - SMART Funds: Net Asset Value B$20.5 (2013) and B$21.3 (2017); Number of Entities 433 (2013) and 494 (2017).
  - Professional Funds: Net Asset Value B$43.8 (2013) and B$16.2 (2017); Number of Entities 226 (2013) and 219 (2017).
  - Standard Funds: Net Asset Value B$4.5 (2013) and B$3.4 (2017); Number of Entities 41 (2013) and 39 (2017).
  - Recognized Foreign Funds: Net Asset Value B$59.2 (2013) and B$45.4 (2017); Number of Entities 53 (2013) and 31 (2017).
- Legislative and regulatory notes:
  - The Investment Funds Act (2003) recognizes four classes: Professional Funds, SMART Funds, Standard Funds, Recognized Foreign Funds, with descriptions of investor eligibility and licensing/registration features.
  - Non-financial International Business Corporations (IBCs) are excluded from the investment fund analysis.
  - Under regulatory changes announced in December 2018, there may be opportunities for international banks to offer services to domestic residents, likely only for non-BSD accounts.
  - Around 5.8 percent of domestic banking sector assets are with offshore counterparties (US$725 million), most intragroup transactions.

*Source: IMF staff summary of chapter content from the FSAP report for The Bahamas.*

### 91.      The offshore insurance sector in The Bahamas is small and populated by

### 1bhsea2019005 - 91.      The offshore insurance sector in The Bahamas is small and populated by

### Offshore insurance sector: size and composition
- Total assets in the offshore insurance sector grew by around 37 percent between 2013 and 2016 to a figure of roughly B$1 billion.
- One third of the sector’s assets are composed of captive insurers.
- External Insurers, End-2016 (Source: Insurance Commission of The Bahamas)
  - Captive Insurers: B$358 (million); Number of Entities: 17
  - Non-captive Insurers: B$689 (million); Number of Entities: 4

### Vulnerabilities and risks from offshore institutions
- Business model characteristics and risk profile
  - Offshore institutions’ business models limit risks associated with traditional banking activities (limited credit creation, maturity and liquidity transformation).
  - Main risks to the domestic financial system are reputational, including adverse impacts on correspondent banking relationships (CBRs).
  - Home-regulated institutions (those without a foreign parent financial group) are:
    - Not subject to group-level risk management frameworks.
    - Dependent on CBRs.
    - Likely to deserve more scrutiny from a risk-based supervisory approach because they are only regulated in The Bahamas.
- Macrofinancial significance
  - The offshore financial sector contributes less than 3 percent to GDP.
  - The offshore financial sector contributes less than 1 percent to government revenue.
  - Authorities emphasize importance of the sector as a source of middle-class employment.
- Exchange controls
  - Exchange controls effectively separate the operations of the onshore and the offshore sectors.

### Data issues and recommended enhancements
- Areas where data could be improved for financial stability analysis and stress testing:
  - Expand granularity of individual bank-level data by loan classifications, provisions, collateral and restructured loans to monitor adequacy of provisions and classifications.
  - Develop a real estate price index to enhance monitoring of housing price trends and collateral valuations.
  - Collect data on household leverage and loan-level loan-to-value and debt-to-income ratios to strengthen monitoring of risks in mortgage portfolios.
  - Expand data collection for credit unions and offshore financial institutions to include them in stress testing exercises conducted by CBOB.

### Risk Assessment Matrix (RAM): key risks, likelihood and impact
- Severe price correction in the real estate market
  - Relative likelihood: Medium
  - Impact and transmission channels: High. Could lead to significant losses on banks' residential mortgage portfolios and potential liquidity issues due to difficulty disposing of non-performing assets.
- Contagion from a major regional economic downturn or banking crisis
  - Relative likelihood: Medium
  - Impact and transmission channels: Medium. Could impair the banking system through losses in cross-border holdings, lack of resolution planning for multi-jurisdiction banks, or strategic retreat by key offshore banks.
- Contagion to the real economy from a severe U.S. recession
  - Relative likelihood: Medium
  - Impact and transmission channels: High. Could trigger a severe local slowdown and lead to significant bank losses in mortgages and consumer loans.
- Tighter global financial conditions
  - Relative likelihood: High
  - Impact and transmission channels: High. An abrupt change in global risk appetite could lead to sudden, sharp increases in interest rates and reduced FDI inflows, impacting tourism, employment, and debt service capacity of borrowers.
- Natural disasters
  - Relative likelihood: High
  - Impact and transmission channels: High. A hurricane may cause severe degradation of domestic infrastructure and impact tourism, employment, and debt service capacity of borrowers.
- Further pressure on CBRs
  - Relative likelihood: Medium
  - Impact and transmission channels: Low/Medium. Could affect profitability and liquidity of banks that are not part of reputable international bank groups, and could trigger business migration toward the latter group.

### Stress Testing Matrix (STEM) — banking sector: top-level assumptions
- Institutional perimeter (Solvency risk)
  - Institutions included:
    - All 7 domestic banks and 2 largest credit unions
    - 10 large offshore banks (results reported for 2 banks with more traditional banking business)
  - Market share coverage:
    - 97 percent of total banking assets
    - 65 percent of credit union assets
    - 61 percent of offshore banking assets
  - Data and baseline date: Supervisory data as of September 2018
- Methodology and horizon
  - IMF stress testing framework with satellite models for macro-financial linkages:
    - Small dynamic stochastic general equilibrium macro-financial model
    - Satellite panel regressions for NPLs, ROA and credit growth
  - Stress test horizon: 3 years (annual frequency)
- Macro scenarios and shocks
  - Baseline: Variables follow the IMF WEO October 2018 projections.
  - Adverse (hurricane): A major hurricane leads to degradation of hotels and key infrastructure; scenario leads to a cumulative decline of real GDP relative to the baseline equivalent to 1 std.
  - Adverse (U.S. recession): A severe recession in the U.S. reduces demand for tourism and real estate investment; scenario results in a cumulative decline of real GDP relative to the baseline equivalent to 1.5 std.
  - Adverse (“perfect storm”): Simultaneous U.S. recession and a major hurricane; scenario leads to a cumulative real GDP decline relative to the baseline equivalent to 2 std.
- Sensitivity analysis (selected shocks and magnitudes)
  - Credit Risk:
    - Increase in NPLs proportional to existing NPLs by 50, 100, 150 and 200 percent.
    - Increase in NPLs proportional to total loans by 3, 5 and 10 percent of total loans.
    - Sectoral NPL increases: Mortgages (150 percent); Commercial loans (150 percent); Consumer loans (150 percent).
    - Defaults of largest 1, 3, 5 borrowers.
  - Interest Rate Risk:
    - Income Effect: Parallel shifts in interest rates in domestic/foreign currency of 300 and 500 bps.
  - Currency Risk:
    - Currency devaluation of 15 and 25 percent.
    - Currency revaluation of 15 and 25 percent.
  - Sovereign Risk:
    - Sovereign bond haircuts of up to 30 percent.
- Risks and buffers assessed
  - Total credit losses, credit growth, profits, repricing gap, shocks to credit quality of sectoral and large exposures, losses from maturity and currency mismatches, indirect FX risk, counterparty risk.
  - Behavioral adjustments: Assumptions for credit growth in scenarios as well as dividend payout ratios.
- Reporting and regulatory parameters
  - Country-specific minimum CAR (trigger ratio).
  - Capital to asset ratio for offshore banks and credit unions.

### Banking sector: liquidity and contagion assumptions
- Liquidity risk
  - Institutional perimeter: All 7 domestic banks.
  - Market share: 97 percent of total banking assets.
  - Data and baseline date: Supervisory data as of September 2018.
  - Methodology: Cash-flow-based (contractual maturity ladder).
  - Risks and buffers: Funding liquidity shock; Market liquidity shock; Buffers include Counterbalancing capacity.
  - Tail shocks: Assumptions for run-off rates on funding sources and roll-off rates on assets to estimate the funding gap.
  - Reporting format: Survival period in days by bank; Number of banks that are liquid/illiquid.
- Contagion risk
  - Institutional perimeter:
    - All 7 domestic banks and 2 largest credit unions
    - 8 large offshore financial institutions
    - 10 largest domestic insurance companies
  - Market share coverage:
    - 7 domestic banks and 2 largest credit unions constitute approximately 97 percent of domestic credit institutions’ assets.
    - 8 large offshore banks constitute around 35 percent of offshore banking assets excluding foreign branches.
    - 10 largest insurance companies constitute approximately 93 percent of aggregate assets in the domestic insurance sector.
  - Methodology: Espinosa-Sole (2010) interbank network model.
  - Tail shocks: Default of institutions with assumption of LGD of 100 percent and that funding from a failed bank is not rolled over and borrower banks must replace it by selling assets with a haircut of 30 percent.
  - Reporting format: Failed capital as a percentage of aggregate capital in the network.

### Financial system structure and soundness (selected metrics and observations)
- Domestic financial sector (June 2018)
  - Assets estimated at US$18 billion in June 2018 (domestic financial sector).
  - Two domestic banks out of the 7 have NPLs above 10 percent.
  - NPLs are concentrated in mortgages: Total Mortgages: 61 percent; Consumer credit: 27 percent; Commercial credit: 12 percent.
  - Domestic sector’s asset composition (percent of GDP):
    - Domestic-majority owned banks: 25.5
    - Foreign-subsidiaries banks: 69.3
    - Credit Unions: 3.4
    - Other Local Financial Institutions: 25.9
    - Domestic Insurers: 16.9
- Offshore financial sector
  - Assets estimated at US$256 billion.
  - Number of entities (selected): Banks &/or Trusts: 1,301; Insurers: 8; Investment Funds: 668.
- Bank profitability and capitalization
  - Bank profitability recovered following large loan-loss provisions and write-offs, supported by steady interest income at banks with resident operations and increased noninterest income at banks with both resident and nonresident operations.
  - The banking sector's CAR has held steady overall:
    - CARs of banks with resident operations increased after receiving fresh capital and reducing risk exposure.
    - CARs of banks with resident and nonresident operations decreased slightly due to stronger growth in risk assets.
- Credit quality
  - NPLs tend to spike in slowdowns and persist for extended periods.
  - Domestic Banking Sector NPLs and arrears show concentration in residential mortgages.
  - Credit unions:
    - Have higher NPLs and lower provisions.
    - Credit portfolios concentrated in consumer loans.
- Asset/liability liquidity profile (selected tenor buckets shown in source figures)
  - Holdings include long-term and short-term government securities and cumulative assets/liabilities across tenor buckets (up to 1 month; 1-3 months; 3-6 months; 6-12 months; 1-5 years; over 5 years) as presented in the source figures.

*Source: IMF staff report content as provided in the supplied PDF excerpt.*

### Appendix III Figure 5. The Bahamas: Maturity Profile of Bank Assets and Liabilities, June 2018

### Appendix III Figure 5. The Bahamas: Maturity Profile of Bank Assets and Liabilities, June 2018

### Maturity profile — overview
- Domestic Government Securities (Percent of banking sector assets): chart axis ticks shown at 0, 2, 4, 6, 8, 10 for maturity buckets: up to 1 month; 1-3 months; 3-6 months; 6-12 months; 1-5 years; over 5 years.
- Maturity Profile: Domestic Banking Sector depicted across the same maturity buckets (axis ticks 0, 2, 4, 6, 8, 10).
- Narrative findings from figure captions:
  - "Banks' exposure to government securities, albeit moderate, is increasing..."
  - "...but bank maturity mismatches are not large due to predominantly adjustable rate loans."

### Resident vs nonresident operations
- Maturity Profile: Banks with Resident and Nonresident Operations — axis ticks shown at 0, 2, 4, 6, 8, 10 for maturity buckets: up to 1 month; 1-3 months; 3-6 months; 6-12 months; 1-5 years; over 5 years.
- Maturity Profile: Banks with Resident Operations — same maturity buckets and axis ticks.
- Narrative findings from figure captions:
  - "Banks with nonresident operations are more exposed to interest rate risk..."
  - "...while maturity mismatches are smaller in banks with resident operations..."

### Related composition and liquidity highlights (from adjacent figures)
- Composition of Bank Funding (In percent of total banking sector assets) — deposit composition axis ticks: 0, 10, 20, 30, 40, 50, 60, 70, 80; deposit categories: Demand deposits; Savings deposits; Fixed deposits. Caption: "The share of credit to the public sector in domestic credit increased... while bank deposits remained stable but fixed deposits declined."
- Sectoral Composition of Credit, June 2018 (Percent of credit to the private sector):
  - Consumer credit: 38%
  - Business credit: 17%
  - Residential mortgages: 43%
  - Commercial mortgages: 2%
- Banking Sector Liquidity indicators (axis values shown on charts):
  - Eligible liquid assets (percent of required liquid assets): axis ticks 0, 50, 100, 150, 200, 250, 300
  - Free cash reserves (percent of required reserves): same axis ticks
  - Weighted average deposit rate (RHS): axis ticks 0.7, 0.8, 0.9, 1, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6
- Captioned observation: "Ample liquidity and limited lending opportunities boosted banks' liquid assets and free reserves..."

### Data sources
- Sources cited in figures: Central Bank of The Bahamas, and IMF staff calculations.

*Source: Appendix III Figure 5. The Bahamas: Maturity Profile of Bank Assets and Liabilities, June 2018 (Central Bank of The Bahamas; IMF staff calculations).*

### Appendix VI. Figure 2. The Bahamas: Offshore Banking Sector,

### Appendix VI. Figure 2. The Bahamas: Offshore Banking Sector

### Composition of Assets and Liabilities (September 2018)
- Most private banking assets are held in fiduciary trusts for clients, whereas on-balance sheet activities are mainly cash management funded by retail deposits.
- Treasury management assets are mostly on balance sheet, with most assets held as balances with other banks and in debt securities.
- Sources: Central Bank of The Bahamas, Bank for International Settlements, and IMF staff calculations.

### Private Banking Assets (Billions of U.S. dollars)
- Chart axis tick values shown: 0, 20, 40, 60, 80, 100, 120, 140, 160, 180, 200.
- Chart legend entries preserved exactly as presented: Total AssetsFiduciary Assets.
- Descriptor: Largest Private Banking Offshore Institutions, 2018 (Percent of Total Assets/Liabilities).

### Treasury Management Assets (Billions of U.S. dollars)
- Chart axis tick values shown: 0, 20, 40, 60, 80, 100, 120, 140, 160.
- Chart legend entries preserved exactly as presented: Total Assets Fiduciary Assets.
- Descriptor: Largest Treasury Management Offshore Institutions, 2018 (Percent of Total Assets/Liablities).

### Bank Claims and Liabilities vis-à-vis Nonresidents (June 2018) — Appendix VI. Figure 3
- Exposure to nonresidents is primarily vis-à-vis banks, with net claims concentrated on the U.S. and the U.K. in developed countries, and on Brazil in emerging markets.
- Chart groupings and labels preserved exactly as presented:
  - Vis-à-vis All Countries; Vis-à-vis Developed Countries (June 2018).
  - Vis-à-vis International Financial Centers (June 2018); Vis-à-vis Latin America (June 2018).
- Chart axis tick values and labels preserved exactly as presented:
  - For larger-scale claims/liabilities: 0, 20, 40, 60, 80, 100, 120, 140, 160, 180, 200 with labels: Total claims vis-à-vis non-residents; Total liabilities vis-à-vis non-residents.
  - Time series labels shown: Dec-15, Jun-18.
  - Component labels shown exactly: Claims vis-à-vis Banks; Liabilities vis-à-vis Banks; Claims vis-à-vis Nonbanks; Liabilities vis-à-vis Nonbanks.
  - Country/regional axis tick values shown: 0, 5, 10, 15, 20, 25, 30, 35, 40, 45 for country/regional comparisons.
  - Country/region labels preserved exactly as presented in charts: US; UK; Switzerland; Other; New Zealand; Ireland; Cayman Islands; Barbados; Hong Kong SAR; Other; Panama; Brazil; Other; Chile; Peru; Guatemala; Argentina.
- Descriptor: Appendix VI. Figure 3. The Bahamas: Bank Claims and Liabilities vis-à-vis Nonresidents, June 2018 (In billions of U.S. dollars).

*Sources: Central Bank of The Bahamas, Bank for International Settlements, and IMF staff calculations.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2019/1bhsea2019005.pdf_