## tarea2024039 — PREFACE and Chapter 13

## Source details

**Canonical URL:** [tarea2024039 — PREFACE and Chapter 13](https://www.imf.org/-/media/files/publications/tar/2024/english/tarea2024039.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/tar/2024/english/tarea2024039.pdf.md)
- [Structured JSON version](/-/media/files/publications/tar/2024/english/tarea2024039.pdf.json)

---

### PREFACE — mission, scope, and deliverables
- Bangladesh Bank (BB) requested Technical Assistance (TA) to develop a framework for assessing climate-related risks in the financial sector of Bangladesh, focusing on physical risk from river and coastal floods and impacts on the banking sector.
- IMF MCM mission dates: in-person mission between March 5 and March 9, 2023.
- Mission team: Agnese Carella (MCMFR) and Javier Uruñuela López (MCMFS).
- Key counterparties met: Governor Mr. Abdur Rouf Talukder; Mr. Md. Ala Uddin (Director, Financial Stability Department - FSD, BB); Mr. Kamrul Hoque Maruf (Joint Secretary, Financial Institutions Division - FID, Ministry of Finance - MoF); Mr. Md. Rafiqul Islam (Joint Director, Bangladesh Bureau of Statistics - BBS); Dr. Farhina Ahmed (Secretary, Ministry of Environment, Forest and Climate Change - MoEFCC); Mr. Iqbal Abdullah Harun (Additional Secretary, MoEFCC); Mr. Mohammad Eamin Khan (Deputy Secretary, Ministry of Agriculture - MoA); Dr. Kisinger Chakma (Deputy Secretary, Ministry of Disaster Management and Relief - MoDMR); Mr. Malik Fida A. Khan (Executive Director, Center for Environmental and Geographic Information Services – CEGIS); Mr. Bhuiya Md. Tamim Al Hossain (Associate Specialist, CEGIS); Mr. Sardar Udoy Raihan (Sub-Divisional Engineer, Flood Forecasting Warning Center – FFWC, Ministry of Water Resources - MoWR); and other staff across departments from the Ministries and BB.
- Mission deliverables:
  - (i) Propose a risk assessment framework tailored to available data, focusing on floods’ impact on banks.
  - (ii) Take stock of financial and climate data availability and identify data gaps.
  - (iii) Set up collaboration and data sharing mechanisms, considering legal obligations and confidentiality, between MoEFCC, MoWR, MoDMR, MoA and financial sector experts in BB (FSD, SD, SFD, CIB), MoF and BBS.
- Follow-up mission: A follow-up mission will be organized by March 2024 subject to sufficient progress on recommendations; it will focus on mapping the impact of hazards on the banking sector using a selected approach.
- Definitions / notes:
  - Thana (third-order administrative level): 495; first-order divisions: 8; second-order districts: 64.
  - Gridded GDP: distribution of GDP into geographical grids defined in a specific EPSG.

### EXECUTIVE SUMMARY — key findings and scope
- Objective: Develop a suitable framework for assessing climate-related risks in the financial sector of Bangladesh, focusing on physical risk from river and coastal floods and impacts on the banking sector.
- Vulnerability: Bangladesh is highly vulnerable to physical risk from river and coastal floods, which already substantially affect population (particularly poorest and most vulnerable groups), the economy, social development, and could have consequences for financial stability.
- RSF access and obligations:
  - Bangladesh was granted access to the Resilience and Sustainability Facility (RSF).
  - Under the RSF arrangement: BB is expected to conduct and publish climate stress testing for the overall financial system.
  - BB must update the Guidelines on Stress Testing for banks and financial institutions to include climate change considerations by end-December 2024.
  - The data for the TA scope is only available for banks.
- Preferred analytical approaches proposed:
  - Preferred micro approach: borrower-level assessment using geo-located bank exposures data.
  - Alternative micro approach: bank exposure data at Thana administrative level.
  - Macro approach: exposure based on gridded GDP.

### Table 1 — Key recommendations (summary of priorities and timeframes)
- General
  - BB and Ministries: Lay out a climate risk assessment plan defining time, scope, objectives, outcomes, contributors, beneficiaries, and division of tasks (Para 21).
  - Priority: High; Timeframe: Short term.
- Risk Framework
  - BB and Ministries: Select the risk framework among options proposed by the TA after evaluating pros and cons (Para 22).
  - Priority: High; Timeframe: Short term.
  - BB and CEGIS: Design a stylized climate scenario tailored to available data with focus on floods; refine to capture tail risks, country-specific damage factors, and changes in data availability (Para 24, 25).
  - Priority: High; Timeframe: Medium term.
  - BB: Link the climate scenario to assessment of financial sector resilience using NPL inflow rate approach or other credit risk modeling as defined for macroprudential solvency stress test (Para 26).
  - Priority: High; Timeframe: Medium term.
  - BB and CEGIS: Test the proposed framework; start with a basic assessment and then expand (Para 25).
  - Priority: Medium; Timeframe: Medium term.
- Data Assessment
  - BB: Close identified data gaps: (i) start data collection and validation for value of exposed assets (corporate and household physical assets at market value, LTVs at current prices, collateral value, firms’ balance sheets, financial statements); (ii) produce longer and more granular time series for credit risk modeling (NPL inflow rate, stock, write-offs, and other NPL outflows) as recommended by the macroprudential stress testing TA (Para 26).
  - Priority: High; Timeframe: Medium term.
  - BB and Ministries: Establish regular cross-validation framework for data quality and consistency across agencies (Para 27).
  - Priority: High; Timeframe: Medium term.
  - BB: Increase granularity of asset location information (location of firms’ assets and houses bought using a mortgage loan at individual household level, in addition to Thana-level) (Para 27).
  - Priority: High; Timeframe: Long term.
  - CEGIS, BBS, MoA, MoDMR: Start data collection and collaboration to develop Bangladesh or regional-specific damage functions, possibly differentiating by assets and regions (Para 27).
  - Priority: Medium; Timeframe: Long term.
- Cooperation between stakeholders
  - BB and Ministries: Establish a framework to coordinate and manage collaboration and data-sharing between financial sector experts (BB and MoF) and climate experts (Ministries) (Para 28).
  - Priority: High; Timeframe: Medium term.
- Timeframe definitions: Short term: up to 6 months; Medium term: 6 to 12 months; Long term: more than 12 months.

### I–III — introduction, background, and TA objectives (summary)
- TA series planned through end-2024; current mission is a scoping mission (March 5-9, 2023).
- Macro context: war in Ukraine, COVID-19 recovery interruption, rising commodity prices, supply chain disruptions, slowdown in external demand, widening current account deficit, depletion of FX reserves, and tighter global financial conditions.
- Climate trends: Average temperatures rising; extreme rainfall and tropical cyclones likely to intensify; sea-level rise expected — increasing frequency/severity of floods and disproportionate effects on poor and vulnerable groups.
- Estimated impacts:
  - Damages from extreme flood events could account for up to 6 percent of the country’s annual GDP in a “hot house world” scenario.
  - For an extreme flood event, financial losses could be as high as 5 percent of total outstanding loans (see 2022 Article IV Staff Report).
- RSF financing: Bangladesh granted SDR 1 billion (about US$1.4 billion) under the RSF to help address climate financing needs and strengthen institutional setup, mobilize private climate finance, and enhance financial sector resilience.
- TA pillars:
  - Pillar (i): Propose risk assessment options and analytical tools based on data availability, timeframe, and inter-agency collaboration.
  - Pillar (ii): Assess financial and climate data availability, legal/confidentiality issues, data sharing options, and data gaps.
  - Pillar (iii): Establish coordination mechanisms between climate experts (ministries) and financial sector experts (BB).
- RSF deliverables: Support BB to meet RSF condition to conduct and publish climate stress testing and update stress testing Guidelines by end-December 2024.

### IV — framework design and methodological choices (three alternatives)
- Objective: estimate banking sector losses from damages caused by more frequent/intense hazards in the future; requires hazard projections, exposure, and vulnerability data.
- Key definitions:
  - Hazard: potential occurrence of a natural or human-induced physical event characterized by frequency (return periods) and geographically explicit intensity distribution.
  - Exposure: presence of population, livelihoods, ecosystems, services, resources, infrastructures, economic/social/cultural assets that could be adversely affected, including banks’ exposure to affected assets.
  - Vulnerability: propensity to be adversely affected, depending on sensitivity to damage and adaptive capacity.
- Proposed alternatives:
  - First-best micro approach (preferred/first best): borrower-level assessment using geo-located exposures (longitude and latitude).
    - Map flood event grids to exposures’ location by longitude and latitude.
    - Estimate damages by applying the flood depth-damage global function (Huizinga et al., 2017) damage rate to the value of exposed assets at each location.
    - Quantify impact on banks’ balance sheets using NPL inflow rate approach or other credit risk modeling developed in macroprudential solvency stress test context.
  - Second-best micro approach (Thana-level aggregation):
    - Map flood event grids to banks’ exposures aggregated at Thana.
    - Upscale damage rate grids within a Thana by taking mean, median, or a percentile value of the grids.
    - Optionally use land cover datasets (e.g., built-up grids) to split grids within a Thana.
  - Third-best macro approach (gridded GDP mapping):
    - Map flood event grids to gridded GDP by upscaling high-resolution grids to lower-resolution grids.
    - Use country-level damages from global damage function to calibrate a macro model to evaluate impact of physical risk shocks on macro and financial variables and effects on banks’ credit and market risks.

### IV.B — data assessment (outcomes)
- Hazards
  - Flood hazard data will come from CEGIS under all framework alternatives.
  - Historical flood maps with 100 m resolution and flood events (depth and extent) are already available.
  - Projections under SSP1-2.6 and SSP5-8.5 can be generated over the medium term, relying on existing hydrological models for river and coastal floods.
  - Projections will include river and coastal floods, considering 4 to 5 different events and then more, in actual conditions and future climate scenarios.
  - CORDEX (6 GCM-RCM) and CHELSA (5 GCM) climate data availability noted; CORDEX data available for RCP 4.5 & 8.5 up to 2100 and CHELSA for SSP1-2.6 & 5-8.5.
- Exposures
  - First-best micro: Confidential individual bank loan data with borrowers’ addresses and geo-location available at BB’s CIB; high-granularity location data currently not available publicly and subject to legal/confidentiality constraints.
  - Second-best micro: Individual bank loans by Thana available from BB’s SD with no restrictions.
  - Macro: Gridded GDP publicly available from LitPop, Murakami, or other sources.
- Vulnerability and damage estimation
  - Initial reliance on global flood depth-damage function from Huizinga et al. (2017).
  - Bangladesh- or regional-specific damage functions should be developed by CEGIS as damage information becomes available (sources: BBS, MoA, MoDMR and others).
  - Damage data would be mapped into effects on bank exposure to assess impacts on banks’ health.
- Identified main data gaps:
  - Exposure data for borrower-level asset values, LTVs, collateral values, firms’ balance sheets, financial statements.
  - Longer and more granular time series for credit risk modeling (NPL inflow rate, stock, write-offs, other NPL outflows).

### IV.C — inter-agency collaboration (outcomes)
- Main data owners/providers and roles
  - CEGIS: climate data and generation of hazard projections and events.
  - BBS: hazards damages and survey data validation; validation of damage information collected by MoA and MoDMR.
  - BB: financial data (CIB for confidential loan-level data; SD for Thana-level aggregated exposures); FSD and SD to lead data collection/validation on value of exposed assets and credit time series.
  - Other stakeholders: MoEFCC, MoWR, MoA, MoDMR to be involved at all stages for data applicability and capacity building.
- Recommended coordination mechanisms
  - Authorities should decide on coordination framework: high-level climate risk committee (possibly joint by the MoEFCC and the BB) or informal forum/workshops.
  - Ministries (ideally MoEFCC) and BB should lead given objectives and data ownership.
  - Strong collaboration between CEGIS and BB to design climate scenarios and iterate on damage estimation (micro and macro approaches).
- Legal and confidentiality considerations: implementation of first-best micro approach may be delayed due to legal issues and validation requirements; Thana-level aggregation allows timelier assessment with no confidentiality restrictions.

### V — recommendations (prioritized actions, selected highlights)
- Adopt and implement a climate risk assessment plan (High priority, Short term).
- Choose risk framework among proposed options after evaluating pros/cons (High priority, Short term).
- Design and iteratively refine a stylized climate scenario focused on floods, capturing tail risks and country-specific damage factors (High priority, Medium term).
- Link climate scenarios to financial sector resilience assessments using NPL inflow rates or other credit risk models (High priority, Medium term).
- Test and expand the proposed framework beginning with a basic assessment (Medium priority, Medium term).
- Close data gaps: collect and validate exposed asset values and improve credit risk time series (High priority, Medium term).
- Establish regular cross-validation for data quality and consistency across agencies (High priority, Medium term).
- Increase spatial granularity of asset location data to household level for mortgages and firm assets (High priority, Long term).
- Develop Bangladesh or regional-specific damage functions through inter-agency data collection and collaboration (Medium priority, Long term).
- Establish a coordination and data-sharing framework between financial and climate stakeholders (High priority, Medium term).

### Implementation timelines and operational notes
- Pilot and short-term activities
  - Start with a few events and low granularity; expand events and granularity over time.
  - Hazard projections and initial scenarios to be produced by CEGIS; 4 to 5 events initially.
- Medium-term (6-12 months)
  - Produce river and coastal floods for 4 to 5 events including actual and future climate scenarios (SSP1-2.6 and SSP5-8.5).
  - BB to collect collateral value, LTV, and value of exposed assets; produce longer and more granular credit time series.
  - MoA and MoDMR to provide natural disaster damage information for validation by BBS.
- Long-term (more than 12 months)
  - Generate further events, other hazards, and a Bangladesh-specific damage function by sector.
- Timelines and deliverables noted in mission context
  - IMF MCM in-person TA mission during March 5-9, 2023.
  - Proposed publication of a climate risk stress-testing exercise required by end-December 2024 to support RSF conditionality.
  - Possible follow-up missions by March 2024 once sufficient progress is achieved; proposed regular virtual check-ins ideally once every 3 months.

### VI — conclusions (scoping mission outputs)
- The scoping mission proposed a prioritized, data-driven path to operationalize physical climate risk assessment for Bangladesh’s banking sector, led by BB with CEGIS as the main climate data partner.
- Short-, medium-, and long-term actions aim to close data gaps, build inter-agency collaboration, and produce climate stress testing consistent with RSF requirements by end-December 2024.
- A follow-up mission by March 2024 will focus on implementing mapping and impact analysis once sufficient progress on recommendations is achieved.

*Source: IMF Technical Assistance Report – Bangladesh Climate Risk Analysis (content unit: tarea2024039, chapter 13)*

### PREFACE _________________________________________________________________________________________ 5

### tarea2024039 - PREFACE

### PREFACE
- Bangladesh Bank (BB, the central bank of Bangladesh) requested Technical Assistance (TA) to develop a suitable framework for assessing climate-related risks in the financial sector of Bangladesh.
- IMF’s Monetary and Capital Markets (MCM) Department conducted an in-person mission between March 5 and March 9, 2023, to help the authorities set up a physical climate risk assessment framework.
- Mission team: Agnese Carella (MCMFR) and Javier Uruñuela López (MCMFS).
- Key meetings: Governor Mr. Abdur Rouf Talukder; Mr. Md. Ala Uddin (Director, Financial Stability Department - FSD, BB); Mr. Kamrul Hoque Maruf (Joint Secretary, Financial Institutions Division - FID, Ministry of Finance - MoF); Mr. Md. Rafiqul Islam (Joint Director, Bangladesh Bureau of Statistics - BBS); Dr. Farhina Ahmed (Secretary, Ministry of Environment, Forest and Climate Change - MoEFCC); Mr. Iqbal Abdullah Harun (Additional Secretary, MoEFCC); Mr. Mohammad Eamin Khan (Deputy Secretary, Ministry of Agriculture - MoA); Dr. Kisinger Chakma (Deputy Secretary, Ministry of Disaster Management and Relief - MoDMR); Mr. Malik Fida A. Khan (Executive Director, Center for Environmental and Geographic Information Services – CEGIS); Mr. Bhuiya Md. Tamim Al Hossain (Associate Specialist, CEGIS); Mr. Sardar Udoy Raihan (Sub-Divisional Engineer, Flood Forecasting Warning Center – FFWC, Ministry of Water Resources - MoWR); and other staff across departments from the Ministries and BB.
- Mission acknowledged cooperation from BB (including FSD, Statistics Department (SD), Sustainable Finance Department (SFD), and Credit Information Bureau (CIB)), MoEFCC staff, and CEGIS.

### EXECUTIVE SUMMARY — Key findings and scope
- Objective: Develop a suitable framework for assessing climate-related risks in the financial sector of Bangladesh, focusing on physical risk from river and coastal floods and impacts on the banking sector.
- Vulnerability: Bangladesh is highly vulnerable to physical risk from river and coastal floods, which already substantially affect population (particularly poorest and most vulnerable groups), the economy, social development, and could have consequences for financial stability.
- RSF access: Bangladesh was granted access to the Resilience and Sustainability Facility (RSF). Under the RSF arrangement:
  - BB is expected to conduct and publish climate stress testing for the overall financial system.
  - BB must update the Guidelines on Stress Testing for banks and financial institutions to include climate change considerations by end-December 2024.
  - The data for the TA scope is only available for banks.
- Mission deliverables:
  - (i) Propose a risk assessment framework tailored to available data, focusing on floods’ impact on banks.
  - (ii) Take stock of financial and climate data availability and identify data gaps.
  - (iii) Set up collaboration and data sharing mechanisms, considering legal obligations and confidentiality, between MoEFCC, MoWR, MoDMR, MoA and financial sector experts in BB (FSD, SD, SFD, CIB), MoF and BBS.
- Approaches proposed:
  - Preferred micro approach: borrower-level assessment using geo-located bank exposures data.
  - Alternative micro approach: bank exposure data at Thana administrative level.
  - Macro approach: exposure based on gridded GDP.
- Follow-up mission: A follow-up mission will be organized by March 2024 subject to sufficient progress on recommendations; it will focus on mapping the impact of hazards on the banking sector using a selected approach.
- Definitions / notes:
  - Thana (third-order administrative level): 495; first-order divisions: 8; second-order districts: 64.
  - Gridded GDP: distribution of GDP into geographical grids defined in a specific EPSG.

### Table 1 — Key Recommendations (summary of priorities and timeframes)
- General
  - BB and Ministries: Lay out a climate risk assessment plan defining time, scope, objectives, outcomes, contributors, beneficiaries, and division of tasks (Para 21).
  - Priority: High; Timeframe: Short term.
- Risk Framework
  - BB and Ministries: Select the risk framework among options proposed by the TA after evaluating pros and cons (Para 22).
  - Priority: High; Timeframe: Short term.
  - BB and CEGIS: Design a stylized climate scenario tailored to available data with focus on floods; refine to capture tail risks, country-specific damage factors, and changes in data availability (Para 24, 25).
  - Priority: High; Timeframe: Medium term.
  - BB: Link the climate scenario to assessment of financial sector resilience using NPL inflow rate approach or other credit risk modeling as defined for macroprudential solvency stress test (Para 26).
  - Priority: High; Timeframe: Medium term.
  - BB and CEGIS: Test the proposed framework; start with a basic assessment and then expand (Para 25).
  - Priority: Medium; Timeframe: Medium term.
- Data Assessment
  - BB: Close identified data gaps: (i) start data collection and validation for value of exposed assets (corporate and household physical assets at market value, LTVs at current prices, collateral value, firms’ balance sheets, financial statements); (ii) produce longer and more granular time series for credit risk modeling (NPL inflow rate, stock, write-offs, and other NPL outflows) as recommended by the macroprudential stress testing TA (Para 26).
  - Priority: High; Timeframe: Medium term.
  - BB and Ministries: Establish regular cross-validation framework for data quality and consistency across agencies (Para 27).
  - Priority: High; Timeframe: Medium term.
  - BB: Increase granularity of asset location information (location of firms’ assets and houses bought using a mortgage loan at individual household level, in addition to Thana-level) (Para 27).
  - Priority: High; Timeframe: Long term.
  - CEGIS, BBS, MoA, MoDMR: Start data collection and collaboration to develop Bangladesh or regional-specific damage functions, possibly differentiating by assets and regions (Para 27).
  - Priority: Medium; Timeframe: Long term.
- Cooperation between stakeholders
  - BB and Ministries: Establish a framework to coordinate and manage collaboration and data-sharing between financial sector experts (BB and MoF) and climate experts (Ministries) (Para 28).
  - Priority: High; Timeframe: Medium term.
- Timeframe definitions: Short term: up to 6 months; Medium term: 6 to 12 months; Long term: more than 12 months.

### I. INTRODUCTION — mission scope and sequencing
- TA series planned through end-2024; current mission is a scoping mission (March 5-9, 2023).
- Follow-up mission will provide hands-on training after sufficient progress on scoping recommendations.

### II. BACKGROUND — macro context and climate vulnerability
- Macroeconomic pressures: war in Ukraine, COVID-19 recovery interruption, rising commodity prices, supply chain disruptions, slowdown in external demand, widening current account deficit, depletion of FX reserves, and tighter global financial conditions.
- Structural challenges: growth and productivity, social and development spending, tax revenue mobilization, export scale and diversification, FDI inflows, climate investment.
- Climate trends: Average temperatures rising; extreme rainfall and tropical cyclones likely to intensify; sea-level rise expected — increasing frequency/severity of floods and disproportionate effects on poor and vulnerable groups.
- Estimated impacts:
  - Damages from extreme flood events could account for up to 6 percent of the country’s annual GDP in a “hot house world” scenario.
  - For an extreme flood event, financial losses could be as high as 5 percent of total outstanding loans (see 2022 Article IV Staff Report).
- RSF financing: Bangladesh granted SDR 1 billion (about US$1.4 billion) under the RSF to help address climate financing needs and strengthen institutional setup, mobilize private climate finance, and enhance financial sector resilience.
- Macroprudential context: a macroprudential solvency stress testing TA (September 2022) recommended practical solutions for scenario-based stress testing and assessment of data and satellite models; current climate TA leverages those tools.

### III. OBJECTIVES — pillars of the TA
- Pillar (i): Propose risk assessment options and analytical tools based on data availability, timeframe, and inter-agency collaboration.
- Pillar (ii): Assess financial and climate data availability, legal/confidentiality issues, data sharing options, and data gaps.
- Pillar (iii): Establish coordination mechanisms between climate experts (ministries) and financial sector experts (BB).
- RSF deliverables: Support BB to meet RSF condition to conduct and publish climate stress testing and update stress testing Guidelines by end-December 2024.

### IV. OUTCOMES — framework design and methodological choices
- Objective of physical risk analysis: estimate banking sector losses from damages caused by more frequent/intense hazards in the future; requires data on hazard projections, exposure, and vulnerability.
- Key definitions:
  - Hazard: potential occurrence of a natural or human-induced physical event characterized by frequency (return periods) and geographically explicit intensity distribution.
  - Exposure: presence of population, livelihoods, ecosystems, services, resources, infrastructures, economic/social/cultural assets that could be adversely affected, including banks’ exposure to affected assets.
  - Vulnerability: propensity to be adversely affected, depending on sensitivity to damage and adaptive capacity.
- Three alternatives proposed (based on current and future data availability):
  - First two: micro approach (borrower-level assessment) — either highest granularity (preferred/first best) or aggregated by Thana (second best), preserving geographic differentiation of banks.
  - Third: macro approach — country-level assessment focusing on aggregate economy-wide effects.
- Preferred option — micro approach with borrower geo-location:
  - Map flood event grids to exposures’ location by longitude and latitude.
  - Estimate damages by applying the flood depth-damage global function (Huizinga et al., 2017) damage rate to the value of exposed assets at each location.
  - Quantify impact on banks’ balance sheets using NPL inflow rate approach or other credit risk modeling developed in macroprudential solvency stress test context.
- Second-best micro approach — Thana-level mapping:
  - Map flood event grids to banks’ exposures aggregated at Thana.
  - Upscale damage rate grids within a Thana by taking mean, median, or a percentile value of the grids.
  - Optionally use land cover datasets (e.g., built-up grids) to split grids within a Thana.
- Third-best macro approach — gridded GDP mapping:
  - Map flood event grids to gridded GDP by upscaling high-resolution grids to lower-resolution grids with similar considerations.
  - Use country-level damages from global damage function to calibrate a macro model to evaluate impact of physical risk shocks on macro and financial variables and effects on banks’ credit and market risks (approach similar to physical risk assessment from typhoons in the Philippines).

### IV.B Data Assessment (outcomes referenced)
- Identified main data gaps: exposure data for borrower-level asset values, LTVs, collateral values, firms’ balance sheets, financial statements; longer and more granular time series for credit risk modeling (NPL inflow rate, stock, write-offs, other NPL outflows).
- Proposed main stakeholder for climate data: CEGIS for flood projections.
- Emphasis on improving asset location granularity (household-level mortgage locations and firms’ asset locations) to enable preferred micro approach.

### IV.C Inter-Agency Collaboration (outcomes referenced)
- Proposed mechanisms ideally involving BB and MoEFCC to coordinate collaboration and capacity development between climate and financial sector experts.
- Need to account for legal obligations and confidentiality constraints in data sharing among MoEFCC, MoWR, MoDMR, MoA, BB (FSD, SD, SFD, CIB), MoF and BBS.

### V. RECOMMENDATIONS — prioritized actions (selected highlights)
- Adopt and implement a climate risk assessment plan (High priority, Short term).
- Choose risk framework among proposed options after evaluating pros/cons (High priority, Short term).
- Design and iteratively refine a stylized climate scenario focused on floods, capturing tail risks and country-specific damage factors (High priority, Medium term).
- Link climate scenarios to financial sector resilience assessments using NPL inflow rates or other credit risk models (High priority, Medium term).
- Test and expand the proposed framework beginning with a basic assessment (Medium priority, Medium term).
- Close data gaps: collect and validate exposed asset values and improve credit risk time series (High priority, Medium term).
- Establish regular cross-validation for data quality and consistency across agencies (High priority, Medium term).
- Increase spatial granularity of asset location data to household level for mortgages and firm assets (High priority, Long term).
- Develop Bangladesh or regional-specific damage functions through inter-agency data collection and collaboration (Medium priority, Long term).
- Establish a coordination and data-sharing framework between financial and climate stakeholders (High priority, Medium term).

### VI. CONCLUSIONS (scoping mission outputs)
- The scoping mission proposed a prioritized, data-driven path to operationalize physical climate risk assessment for Bangladesh’s banking sector, led by BB with CEGIS as the main climate data partner, and outlined short-, medium-, and long-term actions to close data gaps, build inter-agency collaboration, and produce climate stress testing consistent with RSF requirements by end-December 2024.
- A follow-up mission by March 2024 will focus on implementing mapping and impact analysis once sufficient progress on recommendations is achieved.

*IMF | Technical Assistance Report – Bangladesh Climate Risk Analysis*

### 13. The choice of the framework should aim at striking the right balance between timeliness

### 13. The choice of the framework should aim at striking the right balance between timeliness and quality of the analysis

### Framework trade-offs and key findings
- First-best micro approach (high-granularity exposures)
  - Requires access to high-granularity exposure data (value and location of firms’ and households’ physical assets).
  - Implementation might be delayed due to legal issues and validation requirements.
  - Allows for geographical differentiation of banks’ exposure and higher precision.
- Second-best micro approach (Thana-level aggregation)
  - Uses exposures aggregated at the Thana level (available from BB’s SD).
  - Allows a timelier assessment with no confidentiality restrictions.
  - Lower resolution may reduce precision and potentially lead to upward-biased estimates of damages and bank losses because floods are highly localized.
- Macro approach (gridded GDP)
  - Subject to fewer information constraints; data mostly retrievable from public sources (gridded GDP from LitPop, Murakami, or others).
  - Uses macro financial scenario shocks to physical capital based on gridded GDP and the global vulnerability curve from Huizinga et al. (2017).
  - Requires macro modeling skills to translate shocks to physical capital into macroeconomic impacts; currently lacking and results do not capture highly locational and sectoral nature of climate risk.
- Common data prerequisite for micro options
  - Data collection on the value of corporates and households’ physical assets is required and could further delay implementation.

### Data assessment — hazards, exposures, vulnerability
- Hazards
  - Flood hazard data will come from CEGIS under all framework alternatives.
  - Historical flood maps with 100 m resolution and flood events (depth and extent) are already available.
  - Projections under SSP1-2.6 and SSP5-8.5 can be generated over the medium term, relying on existing hydrological models for river and coastal floods.
  - Projections will include river and coastal floods, considering 4 to 5 different events and then more, in actual conditions and future climate scenarios.
  - CORDEX (6 GCM-RCM) and CHELSA (5 GCM) climate data availability noted; CORDEX data available for RCP 4.5 & 8.5 up to 2100 and CHELSA for SSP1-2.6 & 5-8.5.
- Exposures
  - First-best micro: Confidential individual bank loan data with borrowers’ addresses and geo-location available at BB’s CIB; high-granularity location data currently not available publicly and subject to legal/confidentiality constraints.
  - Second-best micro: Individual bank loans by Thana available from BB’s SD with no restrictions.
  - Macro: Gridded GDP publicly available from LitPop, Murakami, or other sources.
- Vulnerability and damage estimation
  - Initial reliance on global flood depth-damage function from Huizinga et al. (2017).
  - Bangladesh- or regional-specific damage functions should be developed by CEGIS as damage information becomes available (sources: BBS, MoA, MoDMR and others).
  - Damage data would be mapped into effects on bank exposure to assess impacts on banks’ health.

### Inter-agency collaboration — roles and coordination
- Identified main data owners/providers and roles
  - CEGIS: climate data and generation of hazard projections and events.
  - BBS: hazards damages and survey data validation; validation of damage information collected by MoA and MoDMR.
  - BB: financial data (CIB for confidential loan-level data; SD for Thana-level aggregated exposures); FSD and SD to lead data collection/validation on value of exposed assets and credit time series.
  - Other stakeholders: MoEFCC, MoWR, MoA, MoDMR to be involved at all stages for data applicability and capacity building.
- Recommended coordination mechanisms
  - Authorities should decide on coordination framework: high-level climate risk committee (possibly joint by the MoEFCC and the BB) or informal forum/workshops.
  - Ministries (ideally MoEFCC) and BB should lead given objectives and data ownership.
  - Strong collaboration between CEGIS and BB to design climate scenarios and iterate on damage estimation (micro and macro approaches).

### Recommendations (selection by theme)
- General
  - Authorities should lay out a climate risk assessment plan specifying time, scope, initial focus on river and coastal floods, objectives, contributors, division of tasks, and monitoring.
- Risk framework
  - Authorities should decide among the three proposed options and may start with a stylized framework (short term: few events and low granularity) and refine over time.
  - Implementation should be led by stakeholders providing the data; others engaged for capacity development.
  - Pilot climate scenario should begin with a few events and be refined to account for tail risks, Bangladesh-specific damage factors, and other hazards as data allow.
  - BB should define mapping from climate scenario to financial sector resilience consistent with macroprudential solvency stress testing approaches.
- Data assessment
  - BB FSD and SD should:
    - (i) start data collection and validation of value of exposed assets (market value of physical assets held by corporates and households to which banks have exposures; LTV at current prices; value of collateral);
    - (ii) produce longer and more granular time series for credit risk modeling (NPL inflow rate, stock, write-offs, and other NPL outflows).
  - Medium-term: increase granularity of asset location data to longitude and latitude (in addition to Thana level).
  - BB, CEGIS, BBS, MoA, MoDMR should collaborate to develop Bangladesh- or regional-specific damage functions by sector/region; CEGIS to develop country-specific damage function using validated damage information and BB input on vulnerable sectors.
- Coordination mechanisms
  - Authorities to decide on framework for coordination, collaboration, and data sharing; consider formal high-level committee or recurring forums/workshops.

### Implementation timelines and operational notes
- Pilot and short-term activities
  - Start with a few events and low granularity; expand events and granularity over time.
  - Hazard projections and initial scenarios to be produced by CEGIS; 4 to 5 events initially.
- Medium-term (6-12 months)
  - Produce river and coastal floods for 4 to 5 events including actual and future climate scenarios (SSP1-2.6 and SSP5-8.5).
  - BB to collect collateral value, LTV, and value of exposed assets; produce longer and more granular credit time series.
  - MoA and MoDMR to provide natural disaster damage information for validation by BBS.
- Long-term (more than 12 months)
  - Generate further events, other hazards, and a Bangladesh-specific damage function by sector.
- Timelines and deliverables noted in mission context
  - IMF MCM in-person TA mission during March 5-9, 2023.
  - Proposed publication of a climate risk stress-testing exercise required by end-December 2024 to support RSF conditionality.
  - Possible follow-up missions by March 2024 once sufficient progress is achieved; proposed regular virtual check-ins ideally once every 3 months.

*Source: IMF Technical Assistance Report – Bangladesh Climate Risk Analysis (content unit: tarea2024039, chapter 13)*

---


_Source: https://www.imf.org/-/media/files/publications/tar/2024/english/tarea2024039.pdf_