## tarea2024090-print-pdf

## Source details

**Canonical URL:** [tarea2024090-print-pdf](https://www.imf.org/-/media/files/publications/tar/2024/english/tarea2024090-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/tar/2024/english/tarea2024090-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/tar/2024/english/tarea2024090-print-pdf.pdf.json)

---

### Mission objective and activities
- Aim: build capacity to enhance financial stability analyses and assessments in the Bank of Jamaica (BOJ).
- Activities:
  - Review of the latest available Financial Stability Report (FSR) and the analytical toolkit.
  - Workshop covering financial stability and macroprudential policy, financial stability indices, credit risk, stress testing, insurance & pension balance sheets and corresponding financial stability risks, essential elements of climate risk, interconnectedness, and contagion risk.
  - Estimation of sectoral credit risk models to enhance the forward-looking element of the BOJ’s financial stability assessment.
  - Discussion of data sources and data management, including options for a data warehouse, a mirroring database for analytics, and a centralized credit register.

### Main analytical deliverable
- Sectoral credit risk models estimated to project impact of different macroeconomic scenarios on banks’ balance sheets.
- Methodology and tools provided:
  - Bayesian Model Averaging (BMA) explained as the suitable methodology for short time series of sectoral non-performing loans.
  - R script with initial estimates for five sectoral credit risk models:
    - mortgages to households,
    - personal loans,
    - corporate loans for tourism,
    - corporate loans for construction,
    - other corporate loans.
- Models enable NPL ratio projections using the macroeconomic forecasting framework available to the BOJ.

### Key characteristics and main risks
- Key statistics:
  - Total financial assets correspond to 221.4% of GDP at the end of 2023.
  - Financial system composition (2023):
    - commercial banks: 40% of the total financial system assets (TFA) in 2023,
    - securities companies: 15% of TFA in 2023,
    - pension funds: 11% of TFA in 2023,
    - insurance companies: 8% of TFA in 2023.
- Main risks identified:
  - Rising interest rates and consequent credit and market risk increases.
  - Financial groups’ interconnectedness.
  - Climate risk.
  - Cyber risk.
  - Market risk is more substantial in Jamaica relative to many Caribbean peers due to a developed securities market and significant role of securities dealers.
  - Credit risk specifics:
    - personal loans typically have fixed interest rates,
    - mortgage interest rates are somewhat variable.
  - Insurance and pension sectors: market risk analyses need enhancement; liabilities may not be reported as market-consistent values.

### FSR integration, structure, and publication practice
- FSRs should integrate macroeconomic and financial analysis and quantify transmission channels between financial and real sectors.
- Stress tests:
  - Assumptions must be justified: severity, speed and scale of impact on default probabilities, hurdle rates on capital and liquidity, dividend distribution, and other parameters.
  - Preferable: general equilibrium framework with satellite econometric models; partial-equilibrium/ad hoc tests acceptable if limitations are clearly described.
- Data, methodology, and transparency:
  - Indicate data sources and cutoff dates clearly; make data available on the same website as the report.
  - Explain methodologies in text with technical details in boxes/annexes.
  - Include aggregate and distributional indicators; highlight weakest/vulnerable institutions or asset classes.
- Report structure and publication practice:
  - Logical, integrated structure with unifying themes centered on key risks; include executive summary, chapters on external/domestic developments, changes in financial infrastructure, methodological annex, and statistical annex.
  - Publication supported by a communication strategy and a dedicated webpage with data used.
  - Publication timing guidance:
    - Reports should be published within three months of the cutoff date for the data, preferably at least two times a year.
    - Publication date should be announced in advance and should not change frequently.

### BOJ financial stability toolkit: components and operational status
- Toolkit components listed:
  - value at risk (VAR) calculation,
  - macro model for NPL projection based on real GDP development,
  - macroprudential indices,
  - regular stress testing involving interest rates, foreign exchange rates, funding, credit, and contagion risks.
- Macroprudential indices include (not exhaustive): EWS, AFSI, Amihud index, BSI, CISS, CCA, CoVaR, COVAR, RAI, Z-score, financial stability-cobweb, credit-to-GDP gap, TRE spread, network topology indicators, absorption ratio, residential real estate price index.
- Operational status:
  - Some frameworks not currently operational due to staff loss, including VaR, Co-VaR, CCA, BSI, and absorption ratio.
- VAR framework methods: historical simulation, variance-covariance, and Monte-Carlo simulation methods.
- Macro model for NPL projection: projects aggregate NPL development based on real GDP growth to estimate total credit risk under adverse GDP-drops scenarios.

### Selected indicator definitions and uses (high-level)
- Amihud Index: daily price response associated with one dollar of trading volume; covers FX and stock exchange markets.
- BSI: combines soundness, asset quality, profitability, liquidity, IRR, and FX risk; assessed via standard deviations from historical average over a 10-year period or available data.
- CISS: joint impact of activity in money, equity, bond, and FX markets; higher CISS = higher correlation and systemic risk.
- CCA and CoVaR: market-based approaches for probability of default and institution’s contribution to systemic risk.
- RAI: measures changes in risk aversion.
- Z-score: measures insolvency risk; BOJ employs RORAC instead of ROA.
- Cobweb diagram, credit-to-GDP gaps (one-sided Hodrick-Prescott filter; BOJ uses three definitions), TRE spread, exposure-at-default framework, network topology, absorption ratio, residential real estate price index described and used as appropriate.

### BOJ stress testing framework and specific tests
- Framework: simple static methodology covering interest rate, foreign exchange rate, funding, credit, and contagion risks; covers deposit-taking institutions and security dealers.
- Aggregate Stress Test (AST): combines simultaneous increases in interest rates, currency depreciation, credit quality deterioration, and deposit outflow; shocks often calibrated based on early 2000s Jamaican stress period.
- Interest rate stress test:
  - Assesses fair value impact from changes in securities prices and net interest income impact from maturity mismatch.
  - Losses reduce capital buffers; excess losses reduce capital base.
- Foreign exchange stress test:
  - Assesses direct solvency risk from FX asset-liability positions and indirect solvency risk via increased loan loss provisions on foreign currency loans.
- Funding risk stress test:
  - Models liquidity drain/hypothetical runs; BOJ can vary percentage of deposits withdrawn.
- Credit risk stress tests:
  - Evaluate impact of shocks to NPLs and performing loans on CAR, include sectoral loan data and balance sheet items, and consider loss in interest income due to additional NPLs.
  - Aggregate BOJ model linking NPLs to real GDP growth can calibrate shocks.
- Documentation: toolkit elements, VAR calculation, and macroprudential indices are documented in internal manuals; the macro model for NPL projection based on real GDP is described in the FSR.

### Yield curve stress test (paragraph 42 method summary)
- Assesses hypothetical changes to the yield curve on portfolio holdings of securities to determine impact on net earnings and capital adequacy.
- Domestic government Jamaica zero-coupon yield curve constructed using the Svensson functional form estimating a zero-coupon curve from current market yields of existing bond issues.
- The implied forward rate curve exemplifies expected short-term future interest rate at points along the term structure.

### Analytical findings from initial sectoral models (empirical results preserved)
- Modeling approach:
  - Dependent variable: NPL ratio transformed by logit: ln(NPL ratio/(1-NPL ratio)).
  - Potential explanatory variables: real GDP growth, unemployment, inflation, foreign exchange rate, T-bill 180-day yield, BOJ policy rate.
  - Time series and sectoral NPL ratios available quarterly.
- Key empirical findings:
  - Household mortgages: NPL ratio rises with higher T-bill 180-day yield and depreciation of the local currency; inflation had a rather positive role (decreasing NPLs), possibly due to increased house prices and decline of LGDs.
  - Personal loans: higher inflation increased NPLs.
  - Tourism and other corporate loans: inflation increases NPLs together with exchange rate depreciation.
  - Construction sector: impact of inflation similar to mortgages to households, possibly explained by increased value of collaterals with higher inflation.

### Non-bank sectors, insurance, and pension funds
- Context:
  - BOJ has macroprudential mandate; supervision of non-banking sector lies with the Financial Services Commission (FSC); BOJ access limited to aggregate FSC data.
  - BOJ has less experience in insurance and pension sectors than banking; twin-peaks regulatory model should help.
- Insurance sector recommendations:
  - Current FSR limited to descriptive statistics; should discuss impact of changing yields, inflation, climate risk, life insurance guarantees, and role of reinsurance.
- Pension funds recommendations:
  - Distinguish defined benefits (DB) and defined contribution (DC) schemes.
  - For DB schemes, report consistent market value of both assets and liabilities.
  - Sector exposed to interest rate and equity risk.

### Organizational, communication, and FSR production recommendations
- Governance and roles:
  - Bank headed by the Governor, supported by five deputy governors and one senior deputy governor.
  - Financial Stability Department reports to the Deputy Governor for Research & Economic Programming and Financial Stability (also responsible for monetary policy).
  - Department split into two units: (1) Financial Risk Management Oversight and (2) Macroprudential Surveillance.
- Communication and production:
  - Treat financial stability as equally important as monetary policy and supervision; reflect in external communication and internal processes.
  - Establish project group/coordinator for the FSR and prepare a detailed production plan with internal deadlines and approvals.
  - Develop external communication strategy: press conferences, social media, seminars, interviews; use FSR foreword to communicate Governor’s key messages.
  - Provide background figures/statistics published in the FSR on the BOJ website and state the publication date in the report.
  - Streamline the FSR: fewer nominal figures, move additional information/statistics to annexes; integrate box arguments into main text or move boxes after main part.
  - Communication protocols for exceptional stress test results exist; broaden communicators beyond governor and senior deputies.

### Data management and surveys
- Current data systems:
  - BOJ has a data warehouse supporting SQL and analytical tools such as Tablo and Power BI; prudential data for deposit-taking institutions stored; nonbank data being included.
  - Other data: remittances, financial market data, national accounts, tourist data, FSC data, Central Database Management System (CDMS), Jamaica Financial Institutions Reporting Management System, Bloomberg.
- Recommended improvements:
  - Organize all data in one data warehouse and fully utilize for analytical purposes.
  - Set up a mirroring database for analytical purposes to support large-data processing tools (e.g., R, Power BI) and server-side processing.
  - Ensure the data management solution provides full flexibility for complex analyses.
  - Consider establishing a centralized credit register; note two private data registers (CRIFF and Creditinfo) cover 98% of the market but BOJ lacks access and obtains only aggregate reports.
  - Interim measure: cover data gaps with targeted surveys (e.g., bank-lending survey) to collect:
    - Average PD/default rate for corporate loans, consumer loans, and mortgages;
    - Average loss given default for corporate loans, consumer loans, and mortgages;
    - Average lending rates for corporate loans, consumer loans, and mortgages;
    - Potential breakdown for corporates: SME and large corporates;
    - Type of collateral – real estate, other assets, etc.;
    - Average deposit rates.
- Proposed cyber-risk survey items:
  - Number of cyber incidents with the impact exceeding the defined threshold,
  - Total losses related to cyber incidents,
  - Breakdown by type of cyber incidents (e.g., malware attack, phishing attack, insider threat, etc.).

### Enhancements to analytical toolkit and stress testing
- Recommendations:
  - Make the FSR and toolkit more forward-looking; strengthen credit risk and market risk analysis.
  - Explore indicators to complement NPL ratios, such as probability of default (PD).
  - Employ econometric modeling to project other financial institutions’ balance sheet items under adverse scenarios.
  - Develop a multi-factor, multiperiod solvency stress testing tool for deposit-taking institutions and move toward a dynamic balance sheet approach; coordinate with macroeconomic modeling team to generate adverse macroeconomic scenarios.
  - BOJ should become familiar with the provided R script and further improve initial sectoral credit risk model estimates.
  - Publish a comprehensive methodological document covering the whole financial stability analytical toolkit.
  - Improve non-bank analysis under twin-peaks: consider reinsurance for insurance, distinguish DB and DC schemes for pensions.
  - Regularly report on emerging risks (especially climate and cyber risks) within the Financial Stability Department’s scope.

### Priority recommendations (priorities and timeframes preserved)
- Strengthen forward-looking financial stability analyses and toolkit.
  - Priority: High; Timeframe: Medium-term.
  - a. Streamline the FSR to follow the central risk story and move additional information to annexes, including a statistical annex.
    - Priority: High; Timeframe: Near-term.
  - b. Improve credit risk and market risk analysis.
    - Priority: High; Timeframe: Medium-term.
  - c. Employ econometric modeling to project key financial stability variables; further improve initial sectoral credit risk model estimates.
    - Priority: High; Timeframe: Medium-term.
  - d. Improve non-bank analysis under the new regulatory regime (twin-peaks).
    - Priority: High; Timeframe: Medium-term.
- Communicate financial stability indicators together with key aspects of the existing regulatory framework for different segments.
  - Priority: High; Timeframe: Near-term.
- Ensure surveillance of emerging risks such as cyber and climate risks falls within the Financial Stability Department and is consistently covered in the FSR.
  - Priority: High; Timeframe: Near-term.
- Consider publishing a comprehensive methodological document covering the whole financial stability analytical toolkit.
  - Priority: Medium; Timeframe: Medium-term.
- Prepare a detailed FSR production plan and develop the external communication strategy related to financial stability.
  - Priorities: High; Timeframe: Near-term.
- Provide background figures/statistics together with the FSR on the BOJ website; use foreword to communicate Governor’s key messages; state FSR publication date.
  - Priorities: Medium; Timeframe: Near-term.
- Organize data in one data warehouse and fully utilize for analytics; consider a centralized credit register.
  - Priority: High; Timeframe: Medium-term.
- Cover existing data gaps with surveys where necessary.
  - Priority: Medium; Timeframe: Medium-term.

*Source: Executive Summary and selected sections, tarea2024090-print-pdf*

### Executive Summary

### Executive Summary

### Mission objective and activities
- The mission aimed to build up capacity to enhance financial stability analyses and assessments in the Bank of Jamaica (BOJ).
- Activities included:
  - Review of the latest available Financial Stability Report (FSR) and the analytical toolkit.
  - A workshop covering financial stability and macroprudential policy, financial stability indices, credit risk, stress testing, insurance & pension balance sheets and corresponding financial stability risks, essential elements of climate risk, interconnectedness, and contagion risk.
  - Estimation of sectoral credit risk models to enhance the forward-looking element of the BOJ’s financial stability assessment.
  - Discussion of data sources and data management, including options for a data warehouse, a mirroring database for analytics, and a centralized credit register.

### Main analytical deliverable
- Sectoral credit risk models were estimated to allow projection of the impact of different macroeconomic scenarios on banks’ balance sheets.
- Methodology and tools provided:
  - Bayesian Model Averaging (BMA) explained as the suitable methodology for short time series of sectoral non-performing loans.
  - R script with initial estimates for five sectoral credit risk models:
    - mortgages to households,
    - personal loans,
    - corporate loans for tourism,
    - corporate loans for construction,
    - other corporate loans.
- The estimated models enable NPL ratio projections using the macroeconomic forecasting framework available to the BOJ.

### Broader analytical topics covered
- Measurement and communication of:
  - financial stability,
  - credit risk,
  - stress testing,
  - insurance & pension economic balance sheets and corresponding financial stability risks,
  - basic elements of climate risk,
  - interconnectedness and contagion risk.
- Macroprudential policy methods and calibration:
  - systemic risk buffer (near term),
  - borrower-based tools (medium term),
  - methodological approaches calibrating loan-to-value ratio (LTV) and countercyclical capital buffer (CCyB).

### Key characteristics of Jamaica’s financial system (preserved figures)
- Total financial assets correspond to 221.4% of GDP at the end of 2023.
- Financial system composition (2023):
  - commercial banks: 40% of the total financial system assets (TFA) in 2023,
  - securities companies: 15% of TFA in 2023,
  - pension funds: 11% of TFA in 2023,
  - insurance companies: 8% of TFA in 2023.

### Main risks identified
- Rising interest rates and consequent credit and market risk increases.
- Financial groups’ interconnectedness.
- Climate risk.
- Cyber risk.
- Market risk is more substantial in Jamaica relative to many Caribbean peers due to a developed securities market and significant role of securities dealers.
- Credit risk specifics:
  - personal loans typically have fixed interest rates,
  - mortgage interest rates are somewhat variable.
- For insurance and pension sectors, market risk needs enhancement through economic balance sheet analysis; liabilities may not be reported as market-consistent values.

### Organizational context
- BOJ governance and financial stability responsibilities:
  - Bank headed by the Governor, supported by five deputy governors and one senior deputy governor.
  - Financial Stability Department reports to the Deputy Governor for Research & Economic Programming and Financial Stability (who is also responsible for monetary policy).
  - The department is split into two units: (1) Financial Risk Management Oversight and (2) Macroprudential Surveillance.

### Data management recommendations
- Organize all data in one data warehouse and fully utilize them for analytical purposes to support the financial stability analytical toolkit.
- Set up a copy of the database — mirroring database — for analytical purposes, supporting large-data processing tools (e.g., R) and server-side processing.
- Ensure the data management solution provides full flexibility for complex analyses.
- Consider setting up a centralized credit register at the bank to calculate indicators such as default rates or average LGDs.
- In the interim, cover data gaps with targeted surveys of financial institutions/market participants.

### Communication and FSR production recommendations
- Treat financial stability as equally important as monetary policy and supervision; reflect this in external communication and internal processes.
- Establish a project group/coordinator for the FSR and prepare a detailed production plan providing sufficient time for internal discussion and analytical work.
- Develop a detailed external communication strategy: press conferences, social media, seminars with market participants, interviews, etc.
- Use the FSR foreword to communicate the Governor’s key messages.
- Provide background figures/statistics published in the FSR on the BOJ website and state the publication date in the report.
- Streamline the FSR: fewer nominal figures to focus on the central risk story; move additional information and statistical material to annexes.
- Integrate key arguments/messages from thematic boxes into the main text; move boxes after the main part of the FSR.

### Enhancements to financial stability analysis
- Make the FSR and toolkit more forward-looking, with strengthened credit risk and market risk analysis.
- Explore additional indicators to complement NPL ratios, such as probability of default (PD).
- Employ econometric modeling to project other key financial institutions’ balance sheet items under adverse scenarios.
- Publish a comprehensive methodological document covering the whole financial stability analytical toolkit to serve as a single reference for FSR readers.
- Improve non-bank analysis in the context of the twin-peaks regulatory regime:
  - consider the role of reinsurance for the insurance sector,
  - distinguish defined benefits (DB) and defined contribution (DC) schemes for pension sector assessments.
- Regularly report on emerging risks (especially climate and cyber risks) within the Financial Stability Department’s scope.

### Priority recommendations (selected from the mission’s recommendation table; priorities and timeframes preserved)
- Financial stability analyses and corresponding toolkit should be strengthened to be more forward-looking.
  - Priority: High; Timeframe: Medium-term.
  - a. Streamline the FSR to follow the central risk story and move additional information to annexes, including a statistical annex.
    - Priority: High; Timeframe: Near-term.
  - b. Improve credit risk and market risk analysis.
    - Priority: High; Timeframe: Medium-term.
  - c. Employ econometric modeling to project key financial stability variables; further improve initial sectoral credit risk model estimates.
    - Priority: High; Timeframe: Medium-term.
  - d. Improve non-bank analysis under the new regulatory regime (twin-peaks).
    - Priority: High; Timeframe: Medium-term.
- Communicate financial stability indicators together with key aspects of the existing regulatory framework for different segments of the financial system.
  - Priority: High; Timeframe: Near-term.
- Ensure surveillance of emerging risks such as cyber and climate risks falls within the Financial Stability Department and is consistently covered in the FSR.
  - Priority: High; Timeframe: Near-term.
- Consider publishing a comprehensive methodological document covering the whole financial stability analytical toolkit.
  - Priority: Medium; Timeframe: Medium-term.
- Prepare a detailed FSR production plan and develop the external communication strategy related to financial stability.
  - Priorities: High; Timeframe: Near-term.
- Provide background figures/statistics together with the FSR on the BOJ website; use foreword to communicate Governor’s key messages; state FSR publication date.
  - Priorities: Medium; Timeframe: Near-term.
- Organize data in one data warehouse and fully utilize for analytics; consider a centralized credit register.
  - Priority: High; Timeframe: Medium-term.
- Cover existing data gaps with surveys where necessary.
  - Priority: Medium; Timeframe: Medium-term.

*Source: Executive Summary, tarea2024090-print-pdf*

### 11. FSRs  should  integrate  macroeconomic  and  financial  analysis. This  analysis  should  flow  in  both

### tarea2024090-print-pdf - 11. FSRs  should  integrate  macroeconomic  and  financial  analysis. This  analysis  should  flow  in  both

### Integration of macroeconomic and financial analysis
- FSRs should assess key macroeconomic trends that can impact the stability of the financial sector and key financial sector developments that can, in turn, impact the real economy.
- The report should identify main transmission channels linking the financial and real economy and assess in quantitative terms how shocks in one area could reverberate in other areas.

### Stress tests: assumptions, justification, and modeling
- FSRs should properly justify the assumptions used in stress tests, since interpretation depends critically on:
  - severity of the shocks,
  - speed and scale of the impact of shocks on default probabilities,
  - hurdle rates on capital and liquidity,
  - dividend distribution,
  - other parameters of the test.
- Ideally, stress tests should be computed within a general equilibrium framework with the support of satellite econometric models that link macroeconomic and financial conditions.
- Simpler, partial-equilibrium or ad hoc tests can be appropriate due to capacity or data constraints, but limitations of such choices should be described and explained.

### Data, methodology, and transparency
- FSRs should indicate the data and methodology used and make them readily available.
- The sources and cutoff date of the data should be indicated clearly and in an easily identifiable place.
- Data displayed in the reports should be made available to the public on the internet on the same website where the reports are published.
- Methodology for projections, sensitivity analyses, and stress tests should be explained in clear and simple terms in the text, with more technical details in boxes or annexes.
- Analyses should consider aggregate and average measures, distributional indicators, and, where appropriate, extreme or individual values (with proper confidentiality safeguards), highlighting positions of the weakest or most vulnerable institutions or asset classes.

### Report structure and publication practices
- Reports should follow a logical and integrated structure with unifying themes centered on key risks.
- Structure should:
  - enable readers to identify parts with specific information,
  - facilitate discussion of cross-cutting topics,
  - be consistent over time to allow comparisons.
- Reports should contain boxes or appendices for slowly evolving issues or passing concerns.
- Reports should have:
  - an executive summary,
  - chapters on external and domestic developments affecting the financial sector,
  - chapters on changes in financial infrastructure, including regulatory and supervisory framework,
  - preferably a table of acronyms, a glossary, a methodological annex, and a statistical annex.
- Publication should be supported by a well-designed communication strategy and a dedicated, easy-to-navigate webpage that includes a database of the data used and links to other relevant publications and agencies.
- Launches should be supported by outreach (audiovisual broadcasts, press releases, public presentations, press conferences) and tailored formats by audience.
- Publication timing:
  - Reports should be published within three months of the cutoff date for the data, preferably at least two times a year.
  - Publication date should be announced in advance and should not change frequently.
  - Past publication dates should be clearly indicated on the website.

### FSR practice and scope (Jamaica example)
- The FSR is regularly published to communicate key financial stability risks based on analyses and assessments.
- The report has been published on an annual basis since 2015.
- Scope covers:
  - macroeconomic environment,
  - deposit-taking institutions,
  - pension funds,
  - collective investment schemes,
  - securities dealers,
  - life and general (non-life) insurance companies.
- The report includes a Foreword with financial stability aims and objectives and uses mission/theme statements to navigate readers.

### BOJ financial stability toolkit: components and status
- Toolkit components:
  - value at risk (VAR) calculation,
  - macro model for NPL projection based on real GDP development,
  - macroprudential indices,
  - regular stress testing involving interest rates, foreign exchange rates, funding, credit, and contagion risks.
- Macroprudential indices cover:
  - early warning system (EWS) for the banking crisis,
  - aggregate financial stability index (AFSI),
  - Amihud index of market debt,
  - banking stability index (BSI),
  - composite indicator of systemic stress (CISS),
  - contingent claims approach (CCA),
  - conditional value-at-risk (COVAR),
  - risk appetite index (RAI),
  - Z-score index,
  - financial stability-cobweb,
  - credit-to-GDP gap,
  - TRE spread,
  - network topology indicators,
  - absorption ratio,
  - residential real estate price index.
- Some frameworks are not currently operational due to staff loss, including VaR, Co-VaR, CCA, BSI, and absorption ratio.
- VAR framework:
  - Uses historical simulation, variance-covariance, and Monte-Carlo simulation methods.
  - Jamaican investment portfolios are more material compared to other Caribbean countries, making market risk monitoring important.
- Macro model for NPL projection:
  - Projects aggregate NPL development based on real GDP growth to estimate total credit risk under adverse GDP-drops scenarios and anticipate loan migration to NPLs.
- EWS framework:
  - Examines indicator values relative to a ‘tranquil period’ mean.
  - Composed of micro-prudential index (MiPi) and macro-financial index (MaFi).
  - Signals based on EWS scores computed from number of standard deviations from tranquil-period mean.
- AFSI:
  - Single quantitative measure comprising microeconomic, macroeconomic, and international factors.
  - Sub-indexes: financial soundness index (FSI), financial vulnerability index (FVI), world economic climate index (WECI), financial development index (FDI).

### Selected indicator definitions and uses
- Amihud Index:
  - Interpreted as daily price response associated with one dollar of trading volume.
  - Covers FX and stock exchange markets; measured by absolute daily change in asset prices divided by absolute daily turnover.
- BSI:
  - Aggregate indicator combining soundness, asset quality, profitability, liquidity, IRR, and FX risk.
  - Assesses partial indicators in terms of standard deviations from historical average computed over a 10-year period or available data.
- CISS:
  - Measures joint impact of activity in money, equity, bond, and FX markets.
  - Sub-indexes for each market; higher CISS indicates higher correlation and systemic risk.
  - Money market: interest rate spread and money market rate changes.
  - Bond market: changes in yields on short-term bonds.
  - Equity market: maximum cumulated loss over a one-year moving window; equity returns computed using the Main JSE Index.
  - FX market: bid-ask spread and FX rate changes.
- CCA:
  - Uses Black-Scholes-Merton option pricing to compute probability of default based on market value of assets versus liabilities.
- CoVaR:
  - Captures an institution’s contribution to systemic risk using market data and VaR methodology.
  - Uses weekly institutional data (assets and liabilities book value, market value of equity, leverage ratio, market value of assets) and weekly macro data (interest rate spread, main stock index returns and volatility, nominal GDP, inflation, 180-day T-Bill).
- RAI:
  - Measures changes in risk aversion using rank correlation between assets' riskiness and excess returns for Jamaica's Stock Market, Money Market, and Foreign Exchange Market.
- Z-score:
  - Measures insolvency risk via leverage, profitability, and potential return shock magnitude; higher Z-score implies lower probability of insolvency.
  - BOJ employs return on risk-adjusted capital (RORAC) instead of ROA for a more robust insolvency probability measure.
- Cobweb diagram:
  - Assesses broad risks across credit markets, the economy, market/funding liquidity, and leverage via five dimensions (three for systemic shock identification and two for shock absorption capacity measured by capital/profitability and funding/liquidity).
- Credit-to-GDP gaps:
  - Measured by deviations from long-term trends using the one-sided Hodrick-Prescott filter.
  - BOJ uses three definitions: (1) private sector credit, (2) total credit (private and public), (3) total credit plus investments in public sector bonds.
  - Gaps could be used to calibrate countercyclical buffer (CCyB).
- TRE spread:
  - Measures premium in the repo rate for default risk; calculated as difference between the 30-day repo rate and the 30-day T-bill rate.
- Exposure-at-default framework:
  - Captures vulnerability of financial institutions to large but plausible macroeconomic/global changes impacting sovereign debt-deficit dynamics.
- Network topology:
  - Examines inter-linkages and bilateral exposures, identifies central institutions and potential shock transmission channels (nodes = financial institutions; links = credit exposures/funding relationships).
- Absorption ratio:
  - Measures co-movement of bank returns (ROA and NIM) using covariance matrix and principal components analysis; significant upward shifts indicate possible market fragility.
- Residential real estate price index:
  - Constructed using market transaction and valuation data on housing units from the National Housing Trust; has yet to signal an excessive build-up of risk.

### BOJ stress testing framework (overview and components)
- Framework is based on a simple static methodology covering interest rate, foreign exchange rate, funding, credit, and contagion risks.
- Exercise covers deposit-taking institutions and security dealers.
- Calculates stresses for risks separately and combines several stresses into the Aggregate Stress Test (AST).
- AST assumes simultaneous impact of increases in interest rates, currency depreciation, credit quality deterioration, and deposit outflow.
- In most instances, shocks are calibrated based on the known period of stress in the Jamaican financial system during the early 2000s.

- Interest rate stress test:
  - Assesses impact of a range of changes in domestic and foreign interest rates on capital adequacy ratio (CAR).
  - Considers (i) fair value impact from changes in securities prices and (ii) net interest income impact from maturity mismatch between rate-sensitive assets and liabilities.
  - Losses reduce capital buffers and any losses beyond the buffer reduce the capital base.

- Foreign exchange stress test:
  - Assesses impact of changes in FX rates on CAR.
  - Evaluates (i) direct solvency risk from FX asset-liability positions and (ii) indirect solvency risk from assumed increase in loan loss provisions on foreign currency loans.
  - Losses reduce capital buffers and any losses beyond the buffer reduce the capital base.

- Funding risk stress test:
  - Calculates change in liquid assets and CAR from hypothetical runs on banks.
  - Models a liquidity drain that simultaneously affects all financial institutions.
  - BOJ can vary assumptions on the percentage of deposits withdrawn.

- Credit risk stress tests:
  - Evaluate impact of shocks to NPLs and performing loans (PLs) on CAR.
  - Increased NPLs and past-due loans (PDLs) lead to increased provisioning, reducing statutory capital while improving portfolio risk profile.
  - Exercise employs sectoral loan data, sectoral profile of past due loans (for sectoral stress test), balance sheet data on total loans, PLs, total assets, NPLs, and provisions of all institutions.
  - Aggregate BOJ model explaining NPLs by real GDP growth can be used to calibrate shocks to NPLs.
  - Exercise considers loss in interest income due to additional NPLs.

### Documentation
- All mentioned elements of the BOJ financial stability toolkit are well documented.
- VAR calculation and macroprudential indices are described in the internal manual with data, variable definitions, and methodologies.
- Description of the macro model for NPL projection based on real GDP is provided in the FSR.

*IMF | CARTAC Jamaica*

### 42. The yield curve stress test assesses hypothetical changes to the yield curve to portfolio holdings of

### tarea2024090-print-pdf - 42. The yield curve stress test assesses hypothetical changes to the yield curve to portfolio holdings of

### Yield curve stress test methodology (paragraph 42)
- The yield curve stress test assesses hypothetical changes to the yield curve to portfolio holdings of securities by financial institutions to determine the potential impact on net earnings and capital adequacy.
- The methodology for constructing the domestic government Jamaica zero-coupon yield curve relies on the Svensson functional form estimating a zero-coupon curve given information on the current market yields of existing bond issues.
- The implied forward rate curve exemplifies the expected short-term future interest rate for a specific point along the term structure.
- Citation included: "4 Cihak, M. (2007): Introduction to Applied Stress Testing, IMF WP 07/59."

### Documentation and current framework (paragraph 43)
- The BOJ stress testing framework is well documented.
- The internal manual describes all essential elements of the methodology, covering data, definition of all variables, and applied methodologies, and describes all steps required to proceed with the calculations.

### Potential improvements and recommendations for the Financial Stability Report (FSR) (paragraphs 44–46, 51–54)
- Streamline and make the FSR more forward-looking:
  - Current FSR uses descriptive statistics rather than a forward-looking modeling framework.
  - Recommend developing a modeling framework to enhance the forward-looking dimension.
  - Replace or complement many nominal figures with relative indicators or move nominal figures to annexes.
  - Foreword could communicate the Governor's key messages; financial stability objectives could be moved to the Preface.
  - Add explanation on the Financial Policy vs. Financial System Stability Committee and clearly state cut-off dates for data sources and publication date.
  - Consider converting extensive boxes into specific thematic articles moved to the end of the report.
- Communicate financial stability indicators together with key aspects of the existing regulatory framework for all segments (banks, insurance companies, pension funds), including whether assets and liabilities are reported as fully market-consistent.
- Discuss upcoming move to IFRS17 reporting for the insurance sector.
- Strengthen credit risk and market risk analysis in the FSR:
  - Capture average PD or LGD trends for different segments/sectors.
  - If PD data are unavailable, calculate implied default rate using write-offs and cured loans.
  - Use granular banking data to monitor credit risk models for mortgages, consumer loans, and industry-specific sectors.
  - Monitor lending interest rates, especially for mortgages, where fixation could mitigate effects of increased interest rates on credit risk.
  - For market risk, elaborate interest rate risk analyses, crucial for insurance companies and pension funds.
- Include key information on duration in regular reporting for assets and liabilities; note that:
  - Banks typically have positive duration mismatches negatively affecting market risk with increased yields.
  - Pension funds and life insurers with usual negative duration mismatch are negatively affected by market risk with decreased yields.
- Include climate and cyber risks in regular financial stability work; build comprehensive climate risk analysis, collect appropriate climate data, cooperate with regional climate scientists, and implement multi-factor, multi-period stress testing models for deposit-taking institutions.

### Analytical toolkit, econometric models, and sectoral credit risk modeling (paragraphs 48–49, 53)
- Strengthen the financial stability analytical toolkit:
  - Develop econometric models to project expected nonperforming loans, credit growth, and net interest rate income for banks.
  - Bayesian Model Averaging (BMA) recommended for short time series and limited observations due to its flexibility in considering multiple models and handling uncertainty in predictions and parameter estimates.
- CARTAC mission provided R script with an initial estimate for five sectoral credit risk models:
  - Mortgages to households
  - Personal loans to households
  - Corporate loans – tourism
  - Corporate loans – construction
  - Other corporate loans
- Modeling approach and data:
  - NPL ratio used as a proxy for credit risk for all five sectors and transformed with logit transformation for regressions. (Footnote: Instead of NPL ratio, ln(NPL ratio/(1-NPL ratio)) was used as a dependent variable for the regressions.)
  - Potential explanatory variables considered: real GDP growth, unemployment, inflation, foreign exchange rate, T-bill 180-day yield, and BOJ policy rate.
  - All time series and sectoral NPL ratios were available quarterly.
- Key empirical findings from initial estimates:
  - Household mortgages: NPL ratio rises with higher T-bill 180-day yield and depreciation of the local currency; inflation had a rather positive role (decreasing NPLs), possibly related to increased house prices and corresponding decline of LGDs.
  - Personal loans: higher inflation increased NPLs (opposite role to mortgages).
  - Tourism and other corporate loans: inflation increases NPLs together with exchange rate depreciation.
  - Construction sector: impact of inflation similar to mortgages to households, possibly explained by increased value of collaterals with higher inflation.
- Recommend BOJ become familiar with the provided R script and further improve initial estimates.
- Recommend development of a multi-factor, multiperiod solvency stress testing tool for deposit-taking institutions and move beyond static methodologies toward a dynamic balance sheet approach; coordinate closely with macroeconomic modeling team to generate adverse macroeconomic scenarios.

### Non-bank sectors, insurance, and pension funds (paragraphs 50–52)
- Non-bank analyses should be improved under twin-peaks regulation context:
  - BOJ has macroprudential mandate, but supervision of non-banking sector lies with the Financial Services Commission (FSC), restricting BOJ’s access to only aggregate FSC data.
  - BOJ has less experience in insurance and pension sectors compared to banking; twin-peak model should help improve work for insurance and pension sectors.
- Insurance sector recommendations:
  - Current FSR covers only a few descriptive statistics (assets, distribution, premiums, profit, solvency, retention ratios).
  - Needs discussion of impact of changing yields, inflation, climate risk, life insurance guarantees in context of interest rate risk, and role of reinsurance to assess potential implications of instability.
  - FSR should discuss key risks rather than present descriptive indicators without clear risk links and assessments.
- Pension funds recommendations:
  - FSR currently provides only total assets and investment split.
  - Distinguish defined benefits (DB) and defined contribution (DC) schemes when assessing financial stability.
  - For assessing solvency of DB schemes, report consistent market value of both assets and liabilities.
  - Sector exposed to interest rate risk and other market risks, especially equity risk.

### Organizational, communication, and production processes (paragraphs 55–60)
- Financial stability should be seen as equally important as monetary policy and supervision and reflected in BOJ external communication and internal processes.
  - Financial stability is primarily responsibility of the Financial Stability Department but requires contributions from supervision division, research and economic programming division, market operations department, and payment system department.
  - Financial stability function must remain independent from monetary policy and supervision despite necessary interaction and coordination.
- Enhance internal communication and processes for FSR production:
  - Establish a project group/coordinator for the FSR.
  - Prepare a detailed production plan with internal deadlines for chapters, boxes, thematic articles, data submissions, responsible persons, and approvals by senior management/Governor.
  - Publication date fixed to the end of May every year; production plan should enable adherence to deadlines.
- Improve cross-departmental interaction during report drafting, especially with supervision.
- Develop a detailed communication strategy complementing internal production plan and promote FSR as a flagship publication:
  - Note: report is published at the end of March and currently followed by a press release; there is no press conference.
  - Existing monetary policy and communication strategy prepared in November 2018; recommend further development focused on financial stability.
  - Plan concrete outreach dates, active promotion via press conferences, social media, seminars, interviews, YouTube videos with management and experts, etc., in two phases: workshop with market participants followed by interviews and broader social media outreach.
  - Schedule promotional activities between April and May, planned before publication.
  - Foreword could communicate Governor's key messages; provide background figures/statistics in chart/table form on the BOJ website and state the publication date in the report.
- Communication protocols for exceptional stress test results exist to outline process and action plan without publishing institutional details; a press release can be issued if warranted.
- Broaden group of communicators beyond governor and senior deputies to include BOJ spokesperson, Head of Department, and experts tailored to audiences (e.g., younger audiences on YouTube).

### Data sources, data management, and surveys (paragraphs 61–65)
- Existing data systems:
  - BOJ has a data warehouse supporting SQL and containing analytical tools such as Tablo and Power BI.
  - Prudential data for deposit-taking institutions stored in the data warehouse; nonbanking data are in the process of being included.
  - Data covered: remittances, financial market data, national accounts, tourist data, and data from the FSC.
  - Other data sources: Central Database Management System (CDMS) with monetary statistics; Jamaica Financial Institutions Reporting Management System with supervisory data; Bloomberg.
- Recommended data management improvements:
  - Organize all data in one data warehouse and fully utilize them for analytical purposes to support the financial stability analytical toolkit.
  - Set up a mirroring database for analytical purposes that supports analytical tools to process large data (e.g., Power BI, R), enabling server-side processing.
  - Ensure data management solution provides complete flexibility for complex analyses.
- Consider establishing a centralized credit register:
  - Currently two private data registers (CRIFF and Creditinfo) cover 98% of the market, but BOJ does not have access and obtains only aggregate reports.
  - BOJ could consider setting up its register without increasing reporting requirements for credit institutions (since they already report to private registers).
  - A fully flexible analytical mirrored database should allow calculation of default rates and average LGDs for different credit portfolio segments and combine credit register data with the data warehouse.
- In the interim, cover data gaps through industry surveys:
  - Initiate surveys (e.g., bank-lending survey) to cover credit risk and cyber risk information not yet monitored.
  - Suggested survey items to collect missing information related to key financial stability risks:
    - Average PD/default rate for corporate loans, consumer loans, and mortgages;
    - Average loss given default for corporate loans, consumer loans, and mortgages;
    - Average lending rates for corporate loans, consumer loans, and mortgages;
    - Information for corporates could potentially be further break down to SME, and large corporates;
    - Information on type of collateral – real estate, other assets, etc.;
    - Average deposit rates.

*IMF | CARTAC Jamaica*

### 66. Another potential survey could address the lack of information on cyber risk. It could cover, for example

### tarea2024090-print-pdf - 66. Another potential survey could address the lack of information on cyber risk. It could cover, for example

### Data gaps and proposed cyber-risk survey
- Proposed survey items:
  - Number of cyber incidents with the impact exceeding the defined threshold,
  - Total losses related to cyber incidents,
  - The information could be potentially broken down by type of cyber incidents, e.g., malware attack, phishing attack, insider threat, etc.
- Any other information/survey that would address the existing data gap based on the risk identified should be further considered. This could reflect any potential new emerging risks or any existing data gaps beyond the mentioned examples.

### Mission objectives and overall findings
- The CARTAC technical assistance mission aimed at strengthening the capacity of the BOJ to analyze and assess financial stability. It addressed:
  - the structure, communication, organizational and operational setup, and the FSR's content,
  - data sources and an analytical toolbox.
- The mission sought to offer suggestions on every topic that could impact the report, both directly and indirectly, by enhancing the quality of the FSR through improved analysis and risk assessments.

### Governance, communication, and FSR design
- Financial stability should be viewed as equally important as monetary policy and supervision.
- The BOJ should enhance internal communication and processes covering interaction among relevant departments/divisions.
- The prepared communication strategy should complement the internal FSR production plan by promoting the report as the key communication tool for financial stability.
- The FSR could be further streamlined and more forward-looking by:
  - focusing on key risk drivers,
  - reducing nominal figures that could be replaced/complemented by relative indicators, or moving them to annexes,
  - communicating financial stability indicators together with the key aspects of the existing regulatory framework to allow readers to interpret the numbers.
- Credit risk and market risk analysis should especially be improved:
  - use granular banking data to monitor credit risk models for different segments of loans, such as mortgages, consumer loans, or loans to industry-specific sectors,
  - analyses of the impact of interest rate risk should be elaborated for market risk, including key information on duration in regular reporting.

### Analytical toolkit and modelling recommendations
- The BOJ must enhance the financial stability analytical toolkit to project important factors such as NPLs, credit growth, and net interest rate revenue for banks via econometric models.
- The mission explained the Bayesian Model Averaging (BMA) technique as appropriate for short time series, noting:
  - BMA's adaptability in considering various models aids in managing uncertainty in parameter estimates and forecasts,
  - BMA tackles problems like variable selection and dynamic relationships, making it useful with limited information.
- The mission provided the R script with an initial estimate for each of the five sectoral credit risk models:
  - mortgages to households,
  - personal loans to households,
  - corporate loans – tourism,
  - corporate loans – construction,
  - other corporate loans.
- The BOJ should become familiar with the provided R script for the sectoral credit risk models and further work on improvements to the initial estimates.

### Non-bank sector, emerging risks, and institutional roles
- The BOJ must improve non-bank analyses and regularly report and assess new emerging risks, especially climate and cyber risks.
- Climate and cyber risk must be a regular component of the financial stability department's activity.
- Institutional responsibilities:
  - The FSC is in charge of overseeing the non-banking industry,
  - The BOJ is the macroprudential authority.
- The anticipated twin-peek model should assist in enhancing the financial stability analyses and assessment for the insurance and pension industries.

### Stress testing and scenario analysis
- The BOJ should implement a multi-factor, multiperiod solvency stress testing tool for deposit-taking institutions.
- The current simple static methodology for stress testing should be replaced with a dynamic approach, allowing projection of the impact of different macroeconomic scenarios on institutions' balance sheets over time.
- Implementation could be facilitated via the foreseen CARTAC technical assistance mission on stress testing.

### Data management, infrastructure, and interim measures
- The BOJ should organize all data in one data warehouse and fully utilize them to support the financial stability analytical toolkit.
- A mirroring database duplicate of the original database should be set up for analytical purposes.
- As data should be analyzed directly at the server, the statistical mirroring database must allow analytical tools (e.g., Power BI, R) to process large data.
- Any data management system selected must offer full flexibility to carry out intricate analyses not anticipated during setup.
- The BOJ should consider creating a centralized credit register as a rich data source for research.
- Meanwhile, surveys conducted with the industry may fill the current data gaps.

*IMF | CARTAC Jamaica*

---


_Source: https://www.imf.org/-/media/files/publications/tar/2024/english/tarea2024090-print-pdf.pdf_