NGFS Scenario Explorer

A climate finance tool that overlays NGFS Phase V scenarios on investment portfolios and reveals how damage function specification choice shifts estimated drawdowns by 30-50%.

Key Insight

Climate scenario analysis typically focuses on which pathway the world follows (Net Zero vs. Current Policies). But the choice of damage function (how temperature translates to economic loss) is equally consequential and far less discussed. This tool makes that visible.

Toggle between three well-known damage functions at any NGFS scenario and watch portfolio drawdown estimates shift by 30-50%:

Damage Function	At 3°C Anomaly	Character
Kalkuhl-Wenz (2020)	~1.8% GDP	Conservative, panel regression on levels
Howard-Sterner (2017)	~10.3% GDP	Meta-analytic, includes catastrophic damages
Burke-Hsiang-Miguel (2015)	~14%+ GDP	Growth-rate channel, compounds over time

NGFS Phase V Scenarios

Six pathways across three IAMs (REMIND-MAgPIE, GCAM 6.0, MESSAGEix-GLOBIOM):

• Net Zero 2050: orderly, 1.5°C aligned
• Below 2°C: orderly, <2°C
• Divergent Net Zero: disorderly, regional divergence
• Delayed Transition: disorderly, sudden post-2030 tightening
• NDCs: hot house, ~2.5-3°C
• Current Policies: hot house, ~3+°C

Setup

cd ngfs-scenario-explorer
python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"

Usage

Streamlit App

streamlit run app/app.py

Upload a portfolio CSV or use the included sample. Select scenarios, toggle damage functions, and explore the results.

Portfolio CSV Format

ticker,name,sector,weight,market_value
XOM,Exxon Mobil Corp,Energy,0.08,80000
AAPL,Apple Inc,Technology,0.12,120000

Required columns: ticker, name, sector, weight Optional: market_value

Sector tags are fuzzy-matched to GICS sectors (e.g., "Tech" → Information Technology, "Oil & Gas" → Energy).

Python API

from ngfs.portfolio import parse_portfolio
from ngfs.iiasa_client import IIASAClient
from ngfs.scenario_engine import compute_all_damage_functions

portfolio = parse_portfolio("data/sample/sample_portfolio.csv")
client = IIASAClient()
trajectories = client.get_temperature_trajectories()

matrix = compute_all_damage_functions(portfolio, trajectories, years=[2050])
comparison = matrix.damage_function_comparison(year=2050)
print(comparison)

Tests

pytest

Architecture

src/ngfs/
  damage_functions.py   # Three damage functions with full implementations
  iiasa_client.py       # IIASA API client (stub with synthetic data)
  portfolio.py          # CSV parsing, GICS sector mapping
  scenario_engine.py    # Core: apply damage fn to scenario → sector drawdowns
  visualization.py      # Plotly charts for Streamlit

app/
  app.py                # Streamlit interface

data/
  cache/                # Parquet-cached IIASA data
  sample/               # Sample portfolio CSV

References

• Kalkuhl, M. & Wenz, L. (2020). J. Environ. Econ. Manag., 103, 102360.
• Burke, M., Hsiang, S. M., & Miguel, E. (2015). Nature, 527, 235-239.
• Howard, P. H. & Sterner, T. (2017). Environ. Resource Econ., 68, 197-225.
• NGFS Phase V Scenarios: https://www.ngfs.net/ngfs-scenarios-portal/