Lesson 4.3: Scenario Analysis and Climate Modeling

01

Overview

Where This Lesson Fits

This is the third of six lessons in Module 4: Green Finance Risk Management. Building on the risk taxonomy established in Lesson 4.1 and the TCFD/TNFD frameworks from Lesson 4.2, this lesson teaches how to model climate futures using NGFS scenarios and Integrated Assessment Models (IAMs).

Pedagogical arc: “With risk identified and frameworks established, now model the future.” TCFD mandates scenario analysis; this lesson gives you the tools to execute it.

Learning Outcomes

Explain why scenarios are used instead of predictions for climate-financial analysis
Describe the NGFS three-scenario framework (Orderly, Disorderly, Hot House World) and IEA complementary scenarios
Trace the scenario-to-financial-impact transmission chain from climate pathway to portfolio losses
Generate Monte Carlo fan charts and sensitivity tornado diagrams for ASEAN country data
Critically evaluate Integrated Assessment Models (IAMs) and their damage function assumptions

Prerequisites

Lesson 4.1 (Climate Risk Taxonomy) and Lesson 4.2 (TCFD/TNFD Frameworks) recommended but not required.

02

Slide Deck

Lecture Slides (33 Slides)

Download Slide Deck (PDF)

03

Foundation Level

Why Scenarios Instead of Predictions?

Three Reasons Predictions Fail

Non-linear dynamics: Climate systems involve tipping points, feedback loops, and cascading failures that defy linear extrapolation
Policy dependence: Outcomes depend on carbon pricing, phase-out timelines, and regulatory choices not yet made
Deep uncertainty: We cannot assign meaningful probabilities to 30-year climate-economy outcomes

What Scenarios Provide

Structured exploration of plausible futures
Stress testing of portfolios against multiple pathways
Identification of vulnerabilities across scenarios
Decision-making under uncertainty (robust strategies that work across scenarios)

ASEAN ContextThe Philippines experiences 20+ typhoons annually. No single prediction can capture the range of intensification trajectories under different warming pathways—scenarios explore all of them.

NGFS Three-Scenario Framework

The Network for Greening the Financial System (NGFS) provides the global standard for climate scenarios in financial supervision, adopted by 130+ central banks.

Three Scenario Families

Orderly (1.5–2°C): Early, smooth transition. Carbon pricing rises gradually. Low physical risk, low transition risk. Best-case pathway.
Disorderly (1.5–2°C): Late, abrupt transition. Carbon price shock around 2030. Low physical risk but HIGH transition risk. Most financially disruptive for transition-exposed portfolios.
Hot House World (3°C+): Limited climate action. Catastrophic physical risk. Low transition risk. Most damaging for ASEAN due to geographic exposure.

Key Scenario Variables

Carbon price, GDP impact, temperature pathway, energy mix, technology cost curves, and policy stringency.

ASEAN InsightFor ASEAN portfolios, Hot House World is typically the worst scenario—the opposite of developed markets where Disorderly is most disruptive. This is because ASEAN faces disproportionate physical risk from typhoons, flooding, and heat stress.

IEA Energy Scenarios (NZE, APS, STEPS)

Three IEA Scenarios

Net Zero Emissions by 2050 (NZE): All energy pledges met plus additional action to reach net zero
Announced Pledges Scenario (APS): Only existing government commitments honored
Stated Policies Scenario (STEPS): Only policies currently in force—no new commitments

How IEA Complements NGFS

NGFS provides macro-financial variables (GDP, carbon price, temperature). IEA provides energy-sector granularity (fuel mix, power generation, investment needs, technology deployment). Together they give both the “big picture” and the “sector detail.”

ASEAN ContextUnder NZE, fossil fuels drop to ~20% of primary energy by 2050; under STEPS they remain at ~60%. The difference is $130+ trillion in cumulative energy investment (IEA, 2024).

Scenario-to-Financial-Impact Chain

Every climate scenario analysis in finance follows a five-stage transmission chain:

Climate Pathway — Temperature, precipitation, extreme weather frequency
Macro Variables — GDP impact, carbon price, energy prices, commodity prices
Sector Shocks — Sector-specific output changes, employment shifts, technology disruption
Counterparty Impact — Revenue decline, cost increase, asset impairment
Portfolio Losses — PD increase, LGD increase, collateral devaluation, mark-to-market losses

ASEAN ExampleFor Philippine agriculture: Hot House World → +2.5°C by 2050 → GDP deviation -8% → Rice yield decline 15% → Agricultural revenue drops → NPL rate rises from 8% to ~16%.

Key Scenario Variables

Six variables form the minimum inputs for any credible climate scenario analysis:

Variable	Financial Channel	Example Impact
Carbon price	Cost of emissions → margin compression	$250/tCO₂ by 2050 (NZE)
GDP growth	Aggregate demand → credit quality	Philippines: -8% under Hot House
Temperature pathway	Physical damage frequency/severity	Typhoon intensity +25%
Energy mix	Stranded assets vs green investment	Fossil share: 60% → 20%
Technology cost	Competitive advantage shifts	Solar 90% cheaper since 2010
Policy stringency	Regulatory compliance costs	Carbon border adjustments

04

Intermediate Level

NGFS Scenario Explorer Tool

Step-by-Step Guide

Navigate to data.ene.iiasa.ac.at/ngfs
Select scenario family (Phase IV, 2024 vintage)
Choose model (REMIND-MAgPIE recommended for ASEAN)
Select region (Southeast Asia or individual countries)
Choose variables (GDP, carbon price, temperature, energy mix)
Download CSV for analysis in Python or Excel

The NGFS Scenario Explorer is free and open-access. All data used in the Quantitative Lab comes from this tool.

Scenario Selection for ASEAN

Why ASEAN Differs from Developed Markets

Developed markets: High transition risk (fossil assets, heavy industry). Disorderly scenario is worst.

ASEAN markets: HIGH physical risk (tropical exposure, typhoons, agriculture-dependent). Hot House World is worst.

GDP Impact by 2050

Country	Hot House	Disorderly	Orderly
Philippines	-8.0%	-4.0%	-1.0%
Vietnam	-6.0%	-3.5%	-1.0%
Thailand	-5.0%	-3.0%	-0.5%

Calibrating Sectoral Shocks (PD/LGD)

PD Adjustment

Probability of Default Shift

$$\Delta PD_{sector} = \beta_{sector} \times \Delta GDP_{scenario}$$

Where $\beta_{sector}$ is the sector sensitivity coefficient. Example: Agriculture in Philippines under Hot House: $\Delta GDP = -8\%$, $\beta_{agriculture} = 1.5$, so $\Delta PD = 12\%$ increase.

LGD Adjustment

Climate-Adjusted Loss Given Default

$$LGD_{climate} = LGD_{base} + (1 - Insurance\_Coverage) \times Physical\_Damage\_Rate$$

Example: Agricultural loan in Philippine typhoon corridor. $LGD_{base} = 45\%$, insurance = 30%, damage rate = 25%. $LGD_{climate} = 45\% + 0.70 \times 25\% = 62.5\%$.

Key InsightLGD adjustment for physical risk is often the largest single impact for ASEAN portfolios, larger than PD shifts.

Monte Carlo Extensions

Fan charts visualize the full range of scenario uncertainty. The central path is the median of 10,000 simulated paths—not a prediction.

Key Percentile Bands

25th–75th percentile: The “likely” range (darker shading)
10th–90th percentile: The “plausible” range (lighter shading)
Risk management should focus on the tails, not the center

The fan widens over time as uncertainty compounds—the width of the fan is the message, not the central path.

Philippine GDP ExampleUnder NGFS Disorderly: median reaches -4% by 2050, but the 90th percentile tail shows GDP loss could reach -8% or worse under extreme outcomes.

Worked Example: Philippine Bank Exposures

Portfolio Composition

Agriculture 25%, Manufacturing 20%, Real Estate 15%, Energy 10%, Services 30%.

Disorderly Scenario PD Adjustments

Sector	PD Increase	Driver
Energy	+15%	Transition dominated
Agriculture	+8%	Physical + transition
Manufacturing	+5%	Energy cost pass-through
Real Estate	+3%	Physical: flood risk
Services	+2%	GDP elasticity

Portfolio-weighted PD increase = 5.6%

05

PhD Extension

Integrated Assessment Models (IAMs)

IAMs are the computational engines behind NGFS scenarios. Four major models:

Model	Scope	Key Sectors	Primary Strength
DICE (Nordhaus/Yale)	Global, aggregate	All (simplified)	Intuition & pedagogy
REMIND-MAgPIE (PIK)	Regional (12)	Energy + land use	ASEAN granularity
GCAM (PNNL)	Regional (32)	Energy + water + land	Technology richness
MESSAGE (IIASA)	Global/Regional	Energy systems	SDG integration

Research QuestionHow do different IAM structural assumptions (top-down vs bottom-up) affect the estimated financial impact of climate scenarios for ASEAN emerging markets?

Damage Function Specification

Standard Damage Function

$$D(T) = \frac{\alpha T^2}{1 + \beta T^2}$$

Where $D$ = damage as fraction of GDP, $T$ = temperature increase above pre-industrial, and $\alpha$, $\beta$ are calibration parameters.

Divergence Above 3°C

Nordhaus (conservative): ~8% GDP loss at 4°C
Weitzman (catastrophic): ~25% GDP loss at 4°C
Howard & Sterner (updated meta-analysis): ~15% GDP loss at 4°C

The choice of damage function can change estimated financial impact by 3× or more at the same temperature.

Research QuestionCan machine learning approaches improve damage function estimation by incorporating high-frequency disaster data from ASEAN typhoon records?

The Discount Rate Debate

Stern vs Nordhaus

Parameter	Stern (2006)	Nordhaus (2017)
Discount rate ($\delta$)	1.4%	4.5%
Social Cost of Carbon	~$200+/tCO₂	~$35/tCO₂
Policy implication	Urgent, aggressive action	Gradual, cost-optimized
Ethical basis	Future = present	Market-observed rates

The discount rate choice alone accounts for a 2–5× difference in the social cost of carbon. For ASEAN, the debate is less relevant because physical damages arrive in the near term (typhoons, flooding), not in 2100.

IAM Critiques

Weitzman (2009): Fat-Tailed Catastrophic Risk

IAMs assume thin-tailed damage distributions, but climate tipping points create fat tails. Even a small probability of 6°C+ warming should dominate cost-benefit analysis. The “dismal theorem”: under fat tails, expected damages are infinite.

Pindyck (2013): “Models of Doom”

IAMs create an illusion of precision with deeply uncertain parameters. Damage functions are essentially calibrated guesses. IAMs tell us nothing useful about optimal policy because results are sensitive to arbitrary assumptions.

Despite these critiques, IAMs remain the best available tool for structured scenario analysis—the alternative is no quantitative framework at all.

Research QuestionHow should ASEAN regulators weight IAM outputs given known limitations in physical risk modeling for tropical regions?

06

Quantitative Lab

Lab: ASEAN Climate Scenario Dashboard

Task:Download NGFS data for Vietnam, Thailand, Philippines under 3 scenarios Method:Python notebook — Monte Carlo fan charts + sensitivity tornado diagrams Output:3-country × 3-scenario comparison dashboard

Step-by-Step Instructions

Download: Extract GDP, carbon price, energy mix, and temperature data from NGFS Scenario Explorer for three ASEAN countries
Clean: Align time series to common 2025–2060 horizon; handle missing values
Model: Calibrate sector sensitivity coefficients ($\beta$) from historical data
Simulate: Run 10,000 Monte Carlo paths per country-scenario combination
Visualize: Generate fan charts (percentile bands) and tornado diagrams (parameter sensitivity)
Interpret: Compare across countries; identify which parameters drive the most uncertainty

Python Pseudocode

import numpy as np

import matplotlib.pyplot as plt

np.random.seed(42)

years = np.arange(2025, 2061)

n_paths = 10000

# Generate Monte Carlo paths

drift = -4.0 / 25  # -4% GDP by 2050

sigma = 0.28

paths = np.cumsum(np.random.normal(drift, sigma, (n_paths, len(years))), axis=1)

# Compute percentile bands

p10 = np.percentile(paths, 10, axis=0)

p50 = np.percentile(paths, 50, axis=0)

p90 = np.percentile(paths, 90, axis=0)

# Plot fan chart

plt.fill_between(years, p10, p90, alpha=0.15)

plt.plot(years, p50, lw=2.5)

Data Sources

NGFS Scenario Explorer EM-DAT Disaster Database IEA Data Portal

07

Case Study: Philippines — Typhoon Exposure Under Climate Scenarios

Climate Context

The Philippines experiences 20+ typhoons annually. Climate models project intensity increases of 10–30% by 2050. Super Typhoon Haiyan (2013) caused $12.9 billion in damages and affected 14.1 million people. Agricultural lending represents 15–20% of Philippine bank portfolios, while insurance penetration is approximately 5%.

The Task

Model agricultural loan default rates under NGFS Hot House and Disorderly scenarios, incorporating projected typhoon intensity increases and cropping pattern shifts.

Wind-Damage Relationship

Cubic Damage Function

$$Damage \propto WindSpeed^3$$

A 25% increase in typhoon intensity implies approximately 95% more damage (1.25³ ≈ 1.95). This cubic relationship means small changes in intensity produce large changes in financial losses.

Default Rate Projections

Metric	Current	Hot House (+25%)	Disorderly (+15%)
Agricultural NPL rate	8%	~15.6%	~12.2%
Additional heat stress	—	+2–3%	+1–2%

Discussion Questions

How does the cubic wind-damage relationship amplify small changes in typhoon intensity into large financial losses?
Why does insurance penetration of ~5% make the Philippines uniquely vulnerable compared to developed markets?
What portfolio adjustments would you recommend to BSP (Bangko Sentral ng Pilipinas) based on these scenario results?

08

Key Concepts

Glossary

Carbon price trajectory: The projected path of carbon pricing ($/tCO₂) over time under a given scenario
Climate sensitivity: The equilibrium temperature change resulting from a doubling of atmospheric CO₂ concentration
Damage function: Mathematical relationship mapping temperature increase to economic output loss: $D(T) = \alpha T^2 / (1 + \beta T^2)$
Deep uncertainty: Condition where probabilities cannot be meaningfully assigned to outcomes—distinguishes climate from conventional risk
Delayed Transition: NGFS scenario where climate action is postponed until ~2030, then implemented abruptly
Discount rate: Rate at which future costs/benefits are reduced to present value; central to the Stern-Nordhaus debate
Fan chart: Visualization showing percentile bands of Monte Carlo simulation paths, illustrating scenario uncertainty
Hot House World: NGFS scenario with limited climate action, resulting in 3°C+ warming and catastrophic physical risk
Integrated Assessment Model: Computational model linking climate, economy, and energy systems (e.g., DICE, REMIND-MAgPIE, GCAM, MESSAGE)
IEA scenarios: Energy-sector scenarios: NZE (Net Zero), APS (Announced Pledges), STEPS (Stated Policies)
LGD adjustment: Climate-adjusted Loss Given Default accounting for physical damage to collateral and insurance gaps
Monte Carlo simulation: Method generating thousands of random paths to quantify the distribution of possible outcomes
NGFS: Network for Greening the Financial System; provides the global standard climate scenarios for 130+ central banks
Orderly transition: NGFS scenario with early, gradual climate action achieving 1.5–2°C with manageable transition costs
Scenario analysis: Structured exploration of plausible futures; distinct from prediction, which claims to know the single outcome
Sector sensitivity ($\beta$): Coefficient mapping macro-level GDP changes to sector-level PD shifts: $\Delta PD = \beta \times \Delta GDP$
Social Cost of Carbon: Estimated total economic damage per tonne of CO₂ emitted; ranges from $35 (Nordhaus) to $200+ (Stern)
Tornado diagram: Sensitivity chart showing which parameters have the largest impact on model output, ranked by magnitude
Transmission chain: Five-stage pathway: climate → macro → sector → counterparty → portfolio losses

09

References

Key References

2024 NGFS. NGFS Climate Scenarios for Central Banks and Supervisors (Phase IV). Network for Greening the Financial System.
2017 Nordhaus, W. “Revisiting the Social Cost of Carbon.” Proceedings of the National Academy of Sciences, 114(7), 1518–1523.
2009 Weitzman, M. “On Modeling and Interpreting the Economics of Catastrophic Climate Change.” Review of Economics and Statistics, 91(1), 1–19.
2024 IEA. World Energy Outlook 2024. International Energy Agency.
2023 ADB. Climate Risk and Adaptation in the Philippines Financial Sector. Asian Development Bank.

Supplementary References

2017 Howard, P. & Sterner, T. “Few and Not So Far Between: A Meta-Analysis of Climate Damage Estimates.” Environmental and Resource Economics, 68(1), 197–225.
2013 Pindyck, R.S. “Climate Change Policy: What Do the Models Tell Us?” Journal of Economic Literature, 51(3), 860–872.
2024 Potsdam Institute for Climate Impact Research. REMIND-MAgPIE Documentation.

10

Downloads

Materials

Download Slide Deck (PDF)

Data Resources

NGFS Scenario Explorer EM-DAT Disaster Database IEA Data Portal