How do banks conduct climate scenario analysis for physical risk, and what methodologies translate climate events into financial losses?

Climate scenario analysis for physical risk assesses how acute events (hurricanes, floods, wildfires) and chronic changes (sea-level rise, temperature increase) affect a bank's asset values, loan performance, and capital adequacy. Unlike traditional stress testing, climate scenarios extend over decades and require geospatial data integration.\n\nMethodological Framework:\n\n`mermaid\ngraph TD\n A[\"Climate Scenario
(NGFS, IPCC pathways)\"] --> B[\"Hazard Modeling
Flood maps, wildfire zones,
heat stress projections\"]\n B --> C[\"Exposure Mapping
Geolocate collateral
to hazard zones\"]\n C --> D[\"Vulnerability Assessment
Damage functions per
property/asset type\"]\n D --> E[\"Financial Impact
LTV deterioration, PD increase,
LGD adjustment\"]\n E --> F[\"Portfolio Loss Estimation
Expected/unexpected losses
under scenario\"]\n F --> G[\"Capital & Strategy
Concentration limits,
pricing adjustments\"]\n`\n\nScenario Pathways (NGFS):\n\n| Scenario | Temperature by 2100 | Physical Risk Level |\n|---|---|---|\n| Net Zero 2050 | +1.5C | Lower physical, higher transition |\n| Below 2C | +1.7C | Moderate both |\n| Current Policies | +3.0C+ | Highest physical risk |\n| Delayed Transition | +1.8C | High transition, moderate physical |\n\nWorked Example:\nPeninsula Savings Bank has a $4.2 billion residential mortgage portfolio concentrated in coastal areas. Under a \"Current Policies\" scenario:\n\nStep 1 -- Hazard assessment:\n- 12% of collateral properties ($504M) are in FEMA 100-year flood zones\n- Climate models project 100-year flood frequency increasing to 1-in-30-year events by 2050\n\nStep 2 -- Damage estimation:\n- Expected flood damage for affected properties: 28% of property value (based on depth-damage functions)\n- Projected losses on collateral: $504M x 0.28 = $141.1 million\n\nStep 3 -- Credit impact:\n- LTV increases from 72% (origination) to 100%+ for flooded properties\n- Projected PD increase: 2.1% baseline to 8.7% for affected segment (underwater borrowers)\n- LGD increases from 25% to 52% (damaged collateral)\n\nStep 4 -- Portfolio loss:\n- Expected loss on affected segment: $504M x 0.087 x 0.52 = $22.8 million\n- Unaffected portfolio: $3,696M x 0.021 x 0.25 = $19.4 million\n- Total expected loss: $42.2 million (vs. $26.5M baseline = 59% increase)\n\nData Challenges:\n- Property-level geolocation requires matching loan records to precise coordinates\n- Damage functions vary by construction type, elevation, flood depth, and duration\n- Insurance coverage reduces bank losses but may become unavailable in high-risk zones\n- Non-linear tipping points (ice sheet collapse, ecosystem failure) are poorly modeled\n- Horizon mismatch: climate scenarios span 30-80 years while bank risk horizons are 1-5 years\n\nRegulatory Expectations:\nSupervisors (ECB, Bank of England, Fed) increasingly require banks to demonstrate climate scenario capability, even if results are not yet tied to Pillar 1 capital. The focus is on governance, methodology, and strategic integration.\n\nExplore climate risk analysis in our FRM Part II course.