Module 4 · Lesson 4.6
ASEAN Climate Vulnerability
Integrated Risk Assessment and Adaptation Finance
33Slides
5Key Formulas
8Charts
1Case Study
5References
01
Overview
Where This Lesson Fits
This is the sixth and final lesson in Module 4: Green Finance Risk Management. It is the capstone that integrates all tools from Lessons 4.1–4.5 — climate risk taxonomy, TCFD/TNFD frameworks, scenario analysis, carbon metrics, and stress testing — and applies them to the ASEAN region with a focus on adaptation finance.
Pedagogical arc: “Apply everything you learned to your own region. Make it real.” Climate risk is not abstract — it is typhoons in the Philippines, Mekong Delta salinity intrusion in Vietnam, and Bangkok subsidence in Thailand. This lesson grounds the entire module in the lived experience of ASEAN economies.
Learning Outcomes
- Apply the Risk = Hazard x Exposure x Vulnerability framework to three ASEAN countries
- Decompose ND-GAIN vulnerability scores into Exposure, Sensitivity, and Adaptive Capacity
- Quantify the ASEAN adaptation finance gap and evaluate instruments to close it
- Design a panel regression to estimate climate event impacts on bank NPLs
- Evaluate Philippines typhoon belt banking sector resilience using integrated assessment
Prerequisites
Lessons 4.1–4.5 required. This capstone lesson assumes familiarity with climate risk taxonomy, TCFD pillars, NGFS scenarios, carbon metrics (WACI, CVaR), and stress testing (PD/LGD adjustments, $\Delta CAR$).
02
Slide Deck
Lecture Slides (33 Slides)
Download Slide Deck (PDF)03
Foundation Level
ASEAN Climate Risk Profile
The Risk = Hazard x Exposure x Vulnerability Framework
Climate risk is not a single number. It is the product of three components: Hazard (the climate event itself — typhoons, floods, sea-level rise), Exposure (what is in harm's way — people, assets, infrastructure), and Vulnerability (the capacity to cope, adapt, and recover).
Climate Risk Identity
$$\text{Risk} = \text{Hazard} \times \text{Exposure} \times \text{Vulnerability}$$Why ASEAN Is a Climate Hotspot
- Tropical location: Maximum exposure to sea-level rise, intensifying cyclones, and monsoon variability
- 150,000+ km coastline: Hundreds of millions live within 50 km of the coast across the region
- Agriculture 10–25% of GDP: Direct climate sensitivity far exceeding developed-economy averages (1–3%)
- Rapid urbanization: Megacities (Bangkok, HCMC, Manila) built on subsiding deltas and floodplains
Three-Country Comparison
| Dimension | Vietnam | Thailand | Philippines |
|---|---|---|---|
| Primary hazard | Sea-level rise, flooding | Subsidence, drought | Typhoons, storm surge |
| Exposed population | 18M (Mekong Delta) | 12M (Bangkok metro) | 20M+ (typhoon belt) |
| Agri GDP share | ~12% | ~8% | ~10% |
| Annual disaster cost | ~1.5% GDP | ~1.0% GDP | ~3.2% GDP |
Key InsightThe Philippines faces the highest per-GDP disaster losses in ASEAN at 3.2%, driven primarily by typhoon frequency (20+ per year). Vietnam faces the greatest long-term structural threat from Mekong Delta sea-level rise. Thailand's risk is more concentrated geographically in the Bangkok metropolitan area.
Vietnam: Mekong Delta and Urban Flooding
Mekong Delta: The Rice Bowl at Risk
- 18 million people depend on the delta for livelihoods
- Produces ~50% of Vietnam's rice and 60% of aquaculture output
- Projected 30–40 cm sea-level rise by 2050 (IPCC AR6 SSP2-4.5)
- Salinity intrusion already reaches 70–90 km inland during dry season, destroying crops
Ho Chi Minh City: Urban Flood Exposure
- 8 million+ residents in flood-prone zones, with rapid densification on reclaimed wetland
- Land subsidence at 2–3 cm/year due to groundwater extraction, compounding sea-level rise
- Combined effect: effective sea-level rise of 5–6 cm/year in worst-affected districts
- 2024 flooding displaced 100,000+ and caused $500M+ in infrastructure damage
Hanoi: Riverine Flood Risk
Increasing Red River flood frequency threatens industrial zones north and west of the city. Return periods for major floods projected to compress from 50 years to 20–25 years by mid-century.
Financial ImpactVietnamese banks hold ~$45B in Mekong Delta agricultural loans and ~$30B in HCMC real estate loans. A 30 cm sea-level rise scenario would affect collateral values across both portfolios, with salinity intrusion reducing agricultural land values by 20–40% in affected provinces.
Thailand: Subsidence, Drought, and Coral Loss
Bangkok: Sinking City
- Land subsidence at 1.5–2 cm/year (groundwater pumping + weight of built environment)
- Average elevation only 1.5 m above sea level — lower than many coastal cities globally
- Coastal flood risk compounded by Chao Phraya River basin drainage patterns
- 2011 floods: $46.5B damage benchmark, 800+ factories inundated in industrial belt
Northeast Thailand: Drought Amplification
- Drought frequency projected to double by 2050 under SSP3-7.0 scenarios
- Isaan region: 20M people, rice-dependent livelihoods, limited irrigation infrastructure
- Agricultural NPLs historically spike 1.5–2x during severe drought years
Andaman and Gulf Coasts: Coral and Tourism
- Coral reef loss projected at 30–50% by 2050 under 2°C warming
- Marine tourism contributes ~6% of Thai GDP; reef degradation threatens dive tourism revenue
- Coastal erosion accelerating at 1–5 m/year on southern beaches
Financial ImpactThai banks' exposure to Bangkok metropolitan real estate exceeds THB 2 trillion ($~$58B). The BOT's 2011-flood-anchored stress tests project that a repeat event under current climate conditions would produce 2.5x the original loss, given higher asset values and continued subsidence.
Philippines: Typhoon Belt Exposure
The Typhoon Corridor
- 20+ typhoons per year enter the Philippine Area of Responsibility; 8–9 make landfall
- Visayas-Mindanao corridor bears the heaviest impact, with Leyte and Samar most exposed
- Typhoon Haiyan (2013): 6,300+ deaths, $2.9B in damages — the calibration reference event
- Category 4–5 typhoon frequency projected to increase 10–20% by mid-century
Economic Vulnerability
- 3.2% GDP annual disaster losses — among the highest globally (World Bank 2024)
- Remittance-dependent communities: OFW remittances ($36B/year) fund post-disaster recovery, creating fragile resilience
- Insurance penetration only ~1.5% of GDP, leaving most losses uninsured
Compound Risks
The Philippines faces compound volcanic, seismic, and climate risks. A typhoon striking during or after a volcanic eruption (lahar flows) or earthquake creates cascading damage multipliers that standard single-hazard models miss entirely.
Banking SectorBSP data shows NPL ratios in typhoon-affected provinces spike 2–3 percentage points within 6 months of a major landfall, with recovery taking 12–18 months. Banks with concentrated Visayas exposure face the greatest asset quality volatility.
Adaptation Finance: Why Mitigation Is Not Enough
The Locked-In Warming Problem
Even under the most aggressive mitigation scenarios, global temperatures will reach 1.5°C above pre-industrial levels by the 2030s (IPCC AR6). This warming is already locked in by past emissions. For ASEAN, this means: adaptation is not optional — it is the immediate priority.
The Adaptation Finance Gap
ASEAN Adaptation Gap
$$\text{Adaptation Gap} = \text{Needs} - \text{Mobilized} \approx \$40\text{B} - \$5\text{B} = \$35\text{B/year}$$ASEAN needs approximately $40 billion per year in adaptation investment (UNEP 2023), but only ~$5 billion is currently mobilized from all sources — public, private, and multilateral combined.
Instruments to Close the Gap
| Instrument | Mechanism | ASEAN Example |
|---|---|---|
| Cat bonds | Transfer tail risk to capital markets | Philippines IBRD cat bond ($225M, 2019) |
| Resilience bonds | Fund infrastructure hardening, lower insurance costs | Pilot stage in Vietnam (ADB-backed) |
| GCF grants | Concessional climate finance | $300M+ approved for ASEAN projects |
| Parametric insurance | Automatic payout on trigger (e.g., wind speed) | SEADRIF (ASEAN disaster risk facility) |
Just Transition Imperative
Adaptation finance must reach the most vulnerable — smallholder farmers, coastal fishing communities, urban informal settlements. Without targeted channeling, adaptation investments risk benefiting only those with existing financial access.
Scale of the ChallengeThe 8:1 gap ratio (needs vs. mobilized) means ASEAN must increase adaptation finance flows by nearly an order of magnitude. Blended finance structures that use concessional capital to de-risk private investment are the most promising pathway to scale.
04
Intermediate Level
ND-GAIN Vulnerability Decomposition
Two-Dimensional Framework
The ND-GAIN Country Index measures climate vulnerability and readiness along two independent dimensions:
ND-GAIN Components
$$\text{Vulnerability} = f(\text{Exposure}, \text{Sensitivity}, \text{Adaptive Capacity})$$$$\text{Readiness} = f(\text{Economic}, \text{Governance}, \text{Social})$$ASEAN Scores (2024 Data)
| Country | Vulnerability | Readiness | ND-GAIN Rank |
|---|---|---|---|
| Thailand | 0.42 | 0.47 | ~70 |
| Vietnam | 0.44 | 0.39 | ~95 |
| Philippines | 0.48 | 0.38 | ~110 |
The Vietnam Paradox
Vietnam has the region's strongest GDP growth (~6.5% annually) yet scores lower on readiness than Thailand. Why? Governance and institutional capacity lag economic growth. Climate policy implementation requires bureaucratic capacity, data systems, and regulatory enforcement that take decades to build — GDP growth alone does not create adaptive capacity.
InterpretationHigh vulnerability + low readiness = maximum urgency. The Philippines occupies this quadrant most acutely: highest hazard exposure (typhoons), lowest insurance penetration, and weakest governance scores among the three countries. This is where adaptation finance is most needed and hardest to deploy.
Country-Level Integrated Risk Frameworks
Vietnam: Dual-Basin Risk
- Mekong Delta: Salinity intrusion advancing 5–10 km/decade, affecting shrimp ponds and rice paddies simultaneously
- HCMC flooding: Combination of tidal surge, river overflow, and drainage system failure during monsoon
- Financial transmission: Agricultural loan defaults cascade through rural cooperative banks to state-owned commercial banks (Agribank, BIDV)
Thailand: Industrial Corridor + Agricultural Heartland
- Chao Phraya Basin: Flood risk to Bangkok and the northern industrial belt (electronics, automotive supply chains)
- Eastern Seaboard: Industrial zone with petrochemical, steel, and automotive plants facing coastal inundation risk
- Financial transmission: Manufacturing supply chain disruption affects corporate loan books at major commercial banks (Bangkok Bank, SCB, Kasikorn)
Philippines: Multi-Hazard Convergence
- Visayas typhoon corridor: Concentrated destruction path affecting agriculture, infrastructure, and housing
- Metro Manila flood basin: Pasig-Marikina river system floods affecting 5M+ residents and commercial centers
- Financial transmission: Rural bank closures post-typhoon, microfinance portfolio losses, MSME loan defaults in affected regions
Cross-Country PatternAll three countries show the same transmission chain: climate event → physical asset damage → borrower income loss → loan default → bank capital erosion. The difference is the hazard type (flood vs. typhoon vs. subsidence) and the banking system structure (state-owned vs. commercial vs. rural cooperative).
Adaptation Finance Gap Analysis
The 8:1 Gap
Finance Gap Ratio
$$\text{Finance Gap Ratio} = \frac{\text{Needed}}{\text{Mobilized}} = \frac{40}{5} = 8:1$$For every dollar of adaptation finance currently flowing to ASEAN, eight dollars are needed. This is the largest climate finance gap of any sector globally.
Breakdown by Sector
| Sector | Need ($B/yr) | Mobilized ($B/yr) | Gap ($B/yr) |
|---|---|---|---|
| Agriculture | 12 | 1.5 | 10.5 |
| Infrastructure | 14 | 2.0 | 12.0 |
| Coastal protection | 8 | 0.8 | 7.2 |
| Urban resilience | 6 | 0.7 | 5.3 |
Why the Gap Persists
- Revenue problem: Adaptation projects generate avoided losses, not positive cash flows — hard to attract private capital
- Measurement problem: No standardized metric for adaptation effectiveness comparable to tCO₂e for mitigation
- Horizon mismatch: Adaptation benefits accrue over 20–50 years; private capital demands 5–10 year returns
Infrastructure DominanceInfrastructure accounts for $12B/year of the gap — the largest single sector. This includes flood barriers, drainage upgrades, climate-resilient roads, and elevated housing. These are classic public goods with high upfront costs and long payback periods, making blended finance essential.
Blended Finance for Resilience Infrastructure
The Blended Finance Model
Blended finance combines concessional capital (grants, below-market loans from development banks) with commercial capital (private equity, institutional investors) to make adaptation projects bankable. The concessional layer absorbs first losses, making the commercial layer's risk-return profile acceptable.
Typical Structure
| Tranche | Provider | Return | Risk Absorption |
|---|---|---|---|
| First-loss (5–15%) | GCF, bilateral donors | 0% | Absorbs initial losses |
| Mezzanine (15–35%) | DFIs (ADB, IFC, AfD) | 2–4% | Subordinated to senior |
| Senior (50–80%) | Commercial banks, pension funds | 5–8% | Protected by lower tranches |
Risk Mitigation Instruments
- Partial credit guarantees: DFI guarantees covering 50–80% of principal, reducing lender risk
- Currency hedging facilities: Absorb FX risk on USD-denominated climate bonds in local-currency economies
- Technical assistance grants: Fund project preparation, feasibility studies, and monitoring systems
How Development Banks De-Risk Private Investment
The ADB's ASEAN Catalytic Green Finance Facility ($1B+ pipeline) demonstrates the model: $1 of concessional capital mobilizes $3–5 of private capital through structured risk-sharing. The key metric is leverage ratio — dollars of private capital mobilized per dollar of public subsidy.
ASEAN ExampleVietnam's Mekong Delta climate resilience project (ADB, $200M) uses a blended structure: GCF first-loss grant ($30M), ADB concessional loan ($70M), and commercial bank participation ($100M). The first-loss layer enabled private bank participation at risk-adjusted returns equivalent to standard infrastructure lending.
NDC Commitments: Vietnam, Thailand, Philippines
Nationally Determined Contributions Comparison
| Dimension | Vietnam | Thailand | Philippines |
|---|---|---|---|
| Net-zero target | 2050 | 2065 | None (aspirational 75% reduction) |
| Unconditional reduction | 15.8% by 2030 | 30% by 2030 | 2.71% by 2030 |
| Conditional reduction | 43.5% (with intl. support) | 40% (with intl. support) | 72.29% (with intl. support) |
| Key sector | Energy (coal phase-down) | Energy + transport | Energy + forestry |
| Adaptation focus | Mekong Delta, agriculture | Water, agriculture | Disaster risk reduction |
Ambition Gaps
- Vietnam: Ambitious 2050 target but coal still 30%+ of energy mix; gap between pledge and current trajectory ~15 percentage points
- Thailand: 2065 net-zero is 15 years behind Vietnam; unconditional pledge is strongest but implementation lags
- Philippines: Lowest unconditional ambition (2.71%), reflecting prioritization of development over mitigation; adaptation receives more policy attention
Connection to Paris Agreement
All three countries' unconditional NDCs are insufficient for 1.5°C alignment (Climate Action Tracker). Only with full conditional commitments and international financial support would trajectories approach 2°C compatibility. This reinforces the adaptation imperative: the warming gap between pledges and Paris targets must be met with adaptation investment.
Implementation RealityThe gap between NDC pledges and implementation is 30–50% across all three countries (UNEP 2023). Conditional targets depend on international climate finance that has not yet materialized at the required scale, making the $35B/year adaptation gap a direct consequence of unmet mitigation financing promises.
05
PhD Extension
Panel Regression: Climate Events and NPLs
Panel Fixed Effects Model
$$NPL_{it} = \alpha + \beta_1 \cdot ClimateEvent_{it} + \beta_2 \cdot Controls_{it} + \gamma_i + \delta_t + \varepsilon_{it}$$Specification Details
- $i$ = country (or province), $t$ = quarter
- $\gamma_i$ = country fixed effects (absorb time-invariant country characteristics)
- $\delta_t$ = time fixed effects (absorb global shocks common to all countries)
- $ClimateEvent_{it}$ = count or intensity of climate disasters in country $i$, quarter $t$
- $Controls_{it}$ = GDP growth, interest rates, exchange rate, commodity prices
Identification Strategy
Climate events are plausibly exogenous to bank-level decisions — typhoons do not strike because banks made bad loans. This natural experiment logic provides credible identification of $\beta_1$ as the causal effect of climate disasters on NPL ratios.
Data Sources
| Variable | Source | Coverage |
|---|---|---|
| Climate events | EM-DAT International Disaster Database | 1990–present, global |
| NPL ratios | National banking supervisory data (BSP, BOT, SBV) | Quarterly, country-level |
| Controls | World Bank WDI, IMF IFS | Quarterly/annual |
Research QuestionDoes $\beta_1$ differ across ASEAN countries? Is the NPL impact of typhoons in the Philippines larger than the impact of floods in Vietnam, controlling for event severity? What is the lag structure — do NPL spikes peak at 1 quarter, 2 quarters, or later?
GIS Spatial Risk Modeling
Overlaying Climate Hazard and Financial Exposure
GIS spatial risk modeling combines climate hazard layers (flood zones, typhoon tracks, sea-level rise projections) with financial exposure data (bank branch networks, loan portfolio locations, collateral property coordinates) to create spatially explicit risk maps.
Layer Stack
| Layer | Data Source | Resolution |
|---|---|---|
| Flood risk zones | UNEP Global Flood Map, national agencies | 30m–90m grid |
| Typhoon tracks | IBTrACS (NOAA) | 6-hourly positions |
| SLR projections | Climate Central CoastalDEM | 90m elevation model |
| Bank branches | Central bank registry + geocoding | Point locations |
| Collateral locations | Land title registries, property databases | Parcel-level |
QGIS Methodology
- Import climate hazard rasters and financial exposure point/polygon layers
- Spatial join: assign hazard intensity values to each bank branch and collateral property
- Aggregate: compute portfolio-weighted hazard exposure by bank, district, or sector
- Visualize: generate heat maps showing concentration of climate-exposed financial assets
Heat Map Interpretation
The output is a spatial concentration map: red = high overlap between climate hazard and financial exposure (e.g., bank branches in flood zones with large loan books). This directly identifies which portfolios need climate stress testing first and where adaptation investment has the highest financial protection value.
Research QuestionCan spatial correlation between bank branch density and climate hazard intensity predict future NPL concentrations? How does branch-level flood exposure correlate with loan-level default rates after controlling for borrower characteristics?
Difference-in-Differences: Disclosure Mandates
Difference-in-Differences Estimator
$$\hat{\delta} = (\bar{Y}_{treat,post} - \bar{Y}_{treat,pre}) - (\bar{Y}_{control,post} - \bar{Y}_{control,pre})$$Research Design
- Treatment: Countries/banks that adopted mandatory climate disclosure (e.g., Thailand's BOT guidelines 2023)
- Control: Countries/banks not yet subject to mandatory disclosure
- Outcome ($Y$): Climate risk provisioning, green loan share, stock price volatility, credit ratings
- Pre/post: Before and after disclosure mandate implementation
Parallel Trends Assumption
The core identifying assumption: in the absence of the disclosure mandate, treatment and control groups would have followed parallel trends in the outcome variable. Testing requires plotting pre-treatment trends and running event-study specifications with leads and lags.
Staggered Adoption Design
ASEAN regulators adopted climate disclosure at different times (MAS 2022, BOT 2023, BSP 2024, SBV planned 2025), creating a staggered treatment design. This enables more robust identification using Callaway & Sant'Anna (2021) or Sun & Abraham (2021) estimators that handle heterogeneous treatment timing.
Research QuestionDoes mandatory climate disclosure lead banks to increase climate risk provisions? Or do banks engage in cosmetic compliance (better reporting without portfolio adjustment)? Can the DiD design distinguish between real risk management improvement and reporting-only responses?
Adaptation Finance Effectiveness: IV Approaches
The Endogeneity Problem
Does adaptation finance reduce climate losses, or does it flow to countries that already have better institutions and would have lower losses anyway? OLS estimates of adaptation finance effectiveness are biased because finance allocation is endogenous to country characteristics that also affect climate outcomes.
Instrumental Variables
- Instrument 1: Distance to nearest GCF regional office (affects access to funding, plausibly uncorrelated with climate outcomes)
- Instrument 2: Historical UN General Assembly voting alignment with major donor countries (affects bilateral aid flows, not climate outcomes directly)
IV Validity Conditions
| Condition | Requirement | Test |
|---|---|---|
| Relevance | Instrument predicts adaptation finance | First-stage F-statistic > 10 |
| Exclusion restriction | Instrument affects outcome only through finance | Economic argument (not testable) |
| Independence | Instrument uncorrelated with error term | Overidentification test (Hansen J) |
Expected Findings
IV estimates of adaptation finance effectiveness are typically larger than OLS estimates, suggesting that OLS is biased downward: adaptation finance flows disproportionately to harder-to-help countries, attenuating the measured effect. The true causal impact of adaptation investment on loss reduction may be 2–3x larger than naive correlations suggest.
Research QuestionCan the IV approach be applied at the sub-national level within ASEAN countries? Province-level variation in adaptation spending and climate outcomes would provide more statistical power and policy-relevant estimates of where adaptation investment generates the highest returns.
Research Frontier: Five Open Questions
Open Research Questions for ASEAN Climate Finance
- ASEAN-Specific Damage Functions: Current integrated assessment models (DICE, PAGE) use global damage functions calibrated on temperate-zone data. ASEAN needs region-specific functions that capture tropical cyclone intensification, monsoon variability, and delta system vulnerability. Can ML methods trained on EM-DAT + national loss data improve upon functional-form assumptions?
- Cross-Border Climate Risk Transmission: Thai banks lend to Vietnamese agriculture; Singaporean banks finance Philippine infrastructure. How do climate shocks in one ASEAN country transmit through cross-border financial linkages? What is the systemic amplification multiplier for the ASEAN banking network?
- Adaptation Finance Additionality: Does multilateral adaptation finance crowd in or crowd out domestic investment? Is the dollar of GCF funding truly additional, or does it substitute for government spending that would have occurred anyway? Measuring additionality requires counterfactual estimation — what would have happened without the intervention.
- Optimal Cat Bond Trigger Design for ASEAN Perils: Cat bonds require parametric triggers (wind speed, rainfall, earthquake magnitude). For ASEAN's compound hazard environment, what trigger combinations minimize basis risk while maintaining actuarial feasibility? Can multi-peril triggers cover typhoon + flood + volcanic ash simultaneously?
- Just Transition Metrics for Financial Sector Climate Policies: As ASEAN banks divest from carbon-intensive sectors, who bears the cost? Can we measure the distributional impact of climate-motivated credit tightening on smallholders, MSMEs, and informal workers? What metrics should regulators track to ensure the green transition does not deepen inequality?
Connecting to Module 4Each of these questions draws on tools from earlier lessons: damage functions (L3 scenario analysis), cross-border risk (L5 second-round effects), cat bonds (L4 CVaR), panel methods (L6 NPL regression). The capstone integrates the entire module's analytical toolkit into frontier research agenda for ASEAN.
06
Quantitative Lab
Lab: Integrated ASEAN Climate Vulnerability Dashboard
Task:Build an integrated ASEAN climate vulnerability dashboard combining spatial, index, regression, and finance gap data
Method:Python (pandas, geopandas, statsmodels, matplotlib)
Output:4-panel dashboard: flood zone map, ND-GAIN visualization, panel regression output, adaptation finance gap chart
Sub-Tasks
- QGIS flood zone mapping: Overlay ASEAN flood hazard raster with bank branch point data; export spatial join results to GeoJSON for Python visualization
- ND-GAIN visualization: Plot vulnerability vs. readiness scatter for ASEAN countries; annotate Vietnam, Thailand, Philippines with ND-GAIN scores and trend arrows
- Panel regression: Estimate the NPL model using 10-year quarterly data for 3 countries; report $\hat{\beta}_1$, standard errors, and country fixed effects
- Adaptation finance gap chart: Stacked bar chart showing sector-level needs vs. mobilized, with gap highlighted in red overlay
Python Pseudocode (Panel Regression Setup)
import pandas as pd
import statsmodels.api as sm
from linearmodels.panel import PanelOLS
# Load and structure panel data
df = pd.read_csv('asean_climate_npl.csv')
df = df.set_index(['country', 'quarter'])
# Specify model with country + time FE
model = PanelOLS(
df['npl_ratio'],
df[['climate_events', 'gdp_growth', 'interest_rate']],
entity_effects=True,
time_effects=True
)
results = model.fit(cov_type='clustered',
cluster_entity=True)
print(results.summary)
import statsmodels.api as sm
from linearmodels.panel import PanelOLS
# Load and structure panel data
df = pd.read_csv('asean_climate_npl.csv')
df = df.set_index(['country', 'quarter'])
# Specify model with country + time FE
model = PanelOLS(
df['npl_ratio'],
df[['climate_events', 'gdp_growth', 'interest_rate']],
entity_effects=True,
time_effects=True
)
results = model.fit(cov_type='clustered',
cluster_entity=True)
print(results.summary)
Expected Regression Output
$$\hat{\beta}_1 \approx 1.8\text{--}2.5 \text{ (pp increase in NPL per climate event)}$$Data Sources
07
Case Study: Philippines — Typhoon Belt Banking Sector Resilience
Context: 3.2% GDP Annual Disaster Losses
The Philippines experiences the highest per-GDP disaster losses in ASEAN. The banking sector operates in a chronic hazard environment where major typhoon impacts are not tail risks but annual near-certainties. This case study applies the full Module 4 toolkit to assess banking sector resilience.
Part 1: Typhoon Projections
- 20+ typhoons enter the Philippine Area of Responsibility per year; 8–9 make landfall
- Category 4–5 frequency projected to increase 10–20% by mid-century (IPCC AR6 WG1)
- Reference event: Typhoon Haiyan (2013) — 6,300+ deaths, $2.9B damages, 4.1M displaced
- Climate change contribution: warmer sea surface temperatures fuel more intense peak winds and heavier rainfall
Part 2: Portfolio Mapping
| Region | Bank Exposure | Typhoon Frequency | Risk Rating |
|---|---|---|---|
| Visayas | PHP 450B | Very High | Critical |
| Mindanao (North) | PHP 280B | High | Elevated |
| Metro Manila | PHP 2.8T | Moderate | Elevated |
| Luzon (Central) | PHP 680B | High | Elevated |
Part 3: NPL Impact Analysis
- NPL ratios in typhoon-affected provinces spike 2–3 percentage points within 6 months of a major landfall
- Recovery to pre-typhoon NPL levels takes 12–18 months on average
- Rural and cooperative banks suffer disproportionately: NPL spikes of 5–8pp vs. 1–2pp for universal banks
- Agriculture and MSME portfolios drive ~70% of post-typhoon NPL increases
Part 4: PCIP Evaluation
The Philippine Crop Insurance Program (PCIP) covers ~2.5M farmers but pays out only ~PHP 15,000 per claim ($~$270) — far below actual crop replacement costs. This underinsurance gap means climate losses transmit directly to bank loan books rather than being absorbed by the insurance layer.
Key Numbers
Haiyan reference: 6,300+ deaths, $2.9B damages, 4.1M displaced
NPL spikes: 2–3pp within 6 months, recovery 12–18 months
Insurance gap: PCIP payout $270 vs. crop replacement cost $800–1,200
Banking sector resilience depends on closing the insurance gap and diversifying geographic concentration
NPL spikes: 2–3pp within 6 months, recovery 12–18 months
Insurance gap: PCIP payout $270 vs. crop replacement cost $800–1,200
Banking sector resilience depends on closing the insurance gap and diversifying geographic concentration
Policy Implications
- Parametric insurance scaling: SEADRIF and PCIP expansion could reduce direct transmission of climate losses to bank balance sheets
- Geographic diversification mandates: BSP could impose concentration limits on typhoon-belt lending to reduce systemic portfolio vulnerability
- Adaptation-linked lending: Preferential capital treatment for loans financing climate-resilient infrastructure (hardened buildings, elevated housing, mangrove restoration)
- Stress testing integration: BSP's emerging climate stress test framework should incorporate typhoon frequency projections and NPL lag structures from historical data
Discussion Questions
- Why does the Philippines have the highest per-GDP disaster losses despite decades of typhoon experience? What prevents institutional learning?
- How should BSP design its climate stress test differently from the BOT framework, given the Philippines' distinct hazard profile (typhoons vs. floods)?
- What is the optimal balance between parametric insurance (fast payout, basis risk) and traditional indemnity insurance (accurate payout, slow processing) for ASEAN typhoon risk?
- How do remittance flows ($36B/year) affect banking sector resilience to climate shocks? Do they substitute for or complement formal insurance?
08
Key Concepts
Glossary
- Risk (H x E x V)
- Climate risk as the product of Hazard (the event), Exposure (what is at risk), and Vulnerability (capacity to cope); the foundational IPCC risk framework
- ND-GAIN Index
- Notre Dame Global Adaptation Initiative Country Index measuring vulnerability and readiness to climate change on a 0–1 scale
- Adaptation finance gap
- The difference between adaptation investment needed and currently mobilized; approximately $35B/year for ASEAN ($40B needed vs. $5B mobilized)
- Cat bond
- Catastrophe bond — a capital markets instrument that transfers tail climate/disaster risk from sponsors (governments, insurers) to investors
- Parametric insurance
- Insurance that pays out automatically when a measurable trigger is breached (e.g., wind speed > 200 km/h), without loss adjustment
- Resilience bond
- A bond where proceeds fund climate resilience infrastructure, with reduced insurance premiums as a co-benefit to bondholders
- GCF
- Green Climate Fund — the primary multilateral fund for climate finance in developing countries, capitalized at $10B+
- Adaptation Fund
- UNFCCC fund specifically for adaptation projects in developing countries, funded by a share of CDM proceeds and voluntary contributions
- Just transition
- Ensuring that the shift to a green economy does not disproportionately burden vulnerable workers, communities, and countries
- Blended finance
- Combining concessional (public/donor) capital with commercial capital using structured risk-sharing to make climate projects bankable
- NDC
- Nationally Determined Contribution — a country's self-defined climate pledge under the Paris Agreement, covering mitigation and adaptation targets
- Difference-in-differences
- Econometric method estimating causal effects by comparing changes in outcomes between treatment and control groups before and after an intervention
- Panel fixed effects
- Regression approach using entity and time fixed effects ($\gamma_i + \delta_t$) to control for unobserved heterogeneity in panel data
- Instrumental variable
- A variable correlated with the endogenous regressor but uncorrelated with the error term, used to identify causal effects when OLS is biased
- Damage function
- A function mapping climate variables (temperature, sea level) to economic losses; central to integrated assessment models (DICE, PAGE)
- PCIP
- Philippine Crop Insurance Program — government insurance for farmers covering typhoon, flood, and drought losses (limited payout ~PHP 15,000)
- Salinity intrusion
- Inland advance of saltwater into freshwater systems due to sea-level rise and reduced river flows, destroying crops and contaminating aquifers
- Compound risk
- Simultaneous or sequential occurrence of multiple hazards (e.g., typhoon + volcanic eruption) creating cascading damage exceeding individual hazard impacts
- Cross-border risk transmission
- Propagation of climate financial shocks across national borders through trade, lending, and investment linkages in interconnected banking systems
- Additionality
- The extent to which climate finance produces outcomes that would not have occurred without the intervention; key metric for evaluating GCF and bilateral aid effectiveness
09
References
Key References
- 2023 ADB. Climate Risk Country Profile: Southeast Asia. Asian Development Bank.
- 2022 Cevik, S. & Jalles, J.T. “This Changes Everything: Climate Shocks and Sovereign Bonds.” Energy Economics, 107, 105856.
- 2025 EM-DAT. The International Disaster Database. Centre for Research on the Epidemiology of Disasters (CRED).
- 2024 GCF. Portfolio Performance Report. Green Climate Fund.
- 2024 World Bank. Climate and Development Report: East Asia and Pacific. World Bank Group.
Supplementary References
- 2024 ND-GAIN. Country Index Technical Report. Notre Dame Global Adaptation Initiative.
- 2024 BSP. Annual Report on Climate Risk in the Philippine Financial System. Bangko Sentral ng Pilipinas.
- 2024 BOT. Climate Risk Assessment Report. Bank of Thailand.
- 2022 IPCC. AR6 WGII Chapter 10: Asia. Intergovernmental Panel on Climate Change.
10