> For the complete documentation index, see [llms.txt](https://docs.alphageo.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.alphageo.ai/methodology/methodology-docs/clima-metrics-climate-gdp-methodology.md). # Clima-Metrics (Climate GDP) - Methodology ### Methodology overview AlphaGeo's Climate GDP Module translates physical climate hazard intensities into sector-disaggregated GDP losses at the subnational level, combining six hazards, three sectors, and 800+ subnational units with endogenous structural transformation within a single, internally consistent framework. Unlike traditional climate risk assessments that operate at the national level and apply uniform damage functions across the entire economy, the Climate GDP Module resolves where damage falls, which sectors absorb it, and how multiple hazards interact when they strike the same capital stock simultaneously. The framework is empirically calibrated using subnational economic data, downscaled daily climate projections, and AlphaGeo's proprietary and granular adaptation and resilience scores. This enables GDP loss estimates that are physically bounded, sectorally disaggregated, and operationally robust. The module produces commercially actionable outputs for real estate portfolio risk assessment, insurance and reinsurance underwriting, infrastructure project finance, sovereign credit analysis, and multilateral climate finance allocation — at a level of subnational, sector-specific, and multi-hazard specificity that goes beyond what aggregate national models currently offer.
### The Data Gap Climate change is already repricing assets, widening sovereign credit spreads, and reshaping insurance loss distributions, but the data that financial markets need to quantify and act on this risk does not yet exist at the right resolution. Three structural gaps in existing climate-GDP models create this blind spot: * **National-level models mask subnational variation.** Within a single country, province-level GDP losses can differ by more than 20 percentage points. A national average treats every asset in that country identically systematically mispricing risk at the level that investment decisions are made. * **Static economic structures distort long-term risk.** Models that hold sector shares constant overstate exposure in economies undergoing structural transformation and fail to capture the different climate sensitivities of agriculture, manufacturing, and services. * **Additive multi-hazard models miss compound non-linearity.** For one-third of countries studied, combined losses exceed the arithmetic sum of individual hazard losses. An additive model systematically underprices risk in precisely the economies most likely to generate correlated catastrophic losses in the same reporting year. ### What This Module Provides Every climate-exposed investment decision rest on the same question: how much of this location's economic output is at risk, and how much is already protected? The Climate GDP Module answers it by delivering economy-wide GDP loss estimates at the subnational level, disaggregated by hazard, sector, and time horizon, with both unadapted and resilience-adjusted outputs at every GID\_1 unit and country level. **Six hazards covered:** Heat · Flash Flood · Riverine Flood · Coastal Flood · Wind · Drought · Combined Multi-Hazard **Three sectors:** Agriculture · Manufacturing · Services All outputs are empirically calibrated and structured for direct use in econometric analysis, climate stress testing, and financial risk modelling. The GDP figures in this module represent modelled hazard exposure, the share of economic output structurally at risk from each climate hazard, not observed losses. Each figure should be read as: the share of GDP at risk if hazard exposure fully materialises under the modelled scenario.
OutputDescription
GDP loss %Combined and per-hazard GDP loss, unadapted and adapted, at subnational level.
Sectoral breakdownAgri / manufacturing / services share of loss per hazard
Resilience dividendPercentage points protected by adaptation infrastructure
National aggregatesGRP-weighted country-level scores for all metrics
Adaptation gap rankingCountries ranked by adaptation efficiency gap
Global snapshotHighest-exposed country rankings by hazards
### Methodology The framework combines hazard- and sector-specific damage functions, dynamic sector shares, and AlphaGeo’s observed resilience scores to estimate both baseline GDP loss and resilience-adjusted GDP loss. The difference between the two is the resilience dividend that represents the GDP loss prevented by adaptation infrastructure. The framework generates sector-weighted, bottom-up GDP loss estimates at the subnational level. The module is built on five methodological advances: * **Six hazards, one model.** Heat, flash flood, riverine flood, coastal flood, wind, and drought — modelled simultaneously at subnational resolution. * **Bounded damage functions**. 18 sigmoid curves (6 hazards × 3 sectors) — physically realistic, threshold-aware, stable beyond the estimation sample. * **Sectors that evolve.** Agricultural, manufacturing, and service shares shift over time as economies develop eliminating the overstatement of agricultural damage that fixed-share models produce at long horizons. * **Compound interactions captured.** Hazards are combined multiplicatively, reflecting that flood, heat, and drought hitting the same economy simultaneously cause more damage than the sum of their parts. * **Real adaptation, not assumptions.** Resilience dividends are derived from AlphaGeo's observed resilience scores per hazard and location — not a uniform flat discount applied equally everywhere. Five sequential modules operationalize this concept, from feature computation to adaptation calibration, producing subnational GDP-at-risk estimates across six hazards, three sectors, and three time horizons. 1. **Feature Computation** Six hazard indices and economic features are computed at GID\_1 subnational resolution. Climate hazard features are derived from NASA NEX-GDDP-CMIP6 (daily downscaled, 0.25°), WRI Aqueduct Floods (1 km), and the STORM synthetic tropical cyclone dataset. Economic features — sectoral GRP per capita for agriculture, manufacturing, and services — are drawn from the DOSE database (Wenz et al., 2023). All features are spatially aggregated via H3 hexagonal indexing against GADM v4.1 boundaries.
HazardFeatureSource
HeatDays above regional 90th-percentile TmaxNASA NEX-GDDP-CMIP6
Flash FloodDays above regional 95th-percentile precipitationNASA NEX-GDDP-CMIP6
Riverine FloodRiverine Inundation depth at 100-and-1000-year return periodWRI Aqueduct
Coastal FloodCoastal Inundation depth at 100-and-1000-year return periodWRI Aqueduct
Wind / StormHurricane maximum speed with high wind days as compound multiplierSTORM +NASA Dataset
DroughtConsecutive dry days, hot days above 40C and Water Stress as compound multiplierNASA + WRI Aqueduct
2. **Sector-Hazard Damage Curves** 18 sigmoid damage curves (6 hazards × 3 sectors) map hazard intensity to fractional GRP loss. Each curve has three independently sourced parameter layers: onset and half thresholds estimated from historical climate distributions; maximum loss ceilings from the empirical damage literature and beta scaling coefficients estimated from panel regression. Sigmoid curves are used rather than quadratic functions because they are bounded, threshold-aware, and stable beyond the estimation sample. 3. **Sector Share Projection** Sector shares are not held fixed at base-year values. They are projected forward using Chenery-Syrquin structural transformation equations fitted to the economic data, as economies grow richer, agricultural shares decline, manufacturing rises then plateaus, and services expand. This structural evolution changes which hazards dominate long-run losses and ensures damage estimates reflect how economies develop over time. **4. Compound Hazard Aggregation** Sector-weighted GDP losses are computed for each hazard, then combined using a multiplicative survival rule that captures compound non-linearity without arithmetic impossibilities. The estimation is bottom-up: losses are computed at the subnational GID\_1 level, then aggregated to the national level using GRP-weighted averaging ensuring economically larger provinces contribute proportionally. 5. **Adaptation and Resilience Dividend** [Adaptation capacity ](https://docs.alphageo.ai/adaptation-data-hub/methodology-2-3-global-adaptation-layer)is sourced from the AlphaGeo repository, hazard-specific resilience scores, constructed from observed physical infrastructure including flood barriers, coastal defences, drainage networks, nature-based solutions, water storage facilities, and urban heat resilience indicators such as building density and green coverage. The adapted loss is computed as: ### Use Cases **Real Estate:** Real estate investors apply uniform national scores to all assets within a country, masking subnational variation. This framework links assets to GID\_1 subnational risk score with hazard decomposition and adaptation efficiency, enabling differentiated pricing across a portfolio rather than a single national adjustment. **Insurance and Reinsurance:** Identifies compound exposure where single-peril models underprice correlated losses and distinguishes business interruption (flash flood) from asset destruction (riverine flood), improving underwriting precision and cat bond pricing. **Infrastructure, Development Finance and Sovereign Markets:** Converts a binary climate risk flag into a quantified, multi-horizon calculation: hazard decomposition, adaptation efficiency, and residual GDP at risk aligned to bond maturities for sovereign credit and infrastructure stress testing. **Multilateral Development Banks and Climate Finance:** Ranks countries by exposure and adaptation efficiency gap identifying where resilience investment generates the highest marginal return, directly actionable for fund allocation. **Research and Climate Economics:** Provides subnational GDP loss estimates usable as an input, benchmark, or cross-country dataset for climate-growth research, damage function calibration, and adaptation effectiveness studies. ### **Conclusion** The physical consequences of climate change are not equally distributed, not sectorally uniform, not temporally static, and not linearly additive. Every model that assumes otherwise, and every financial decision made with that model is answering the wrong question. The Climate GDP Module resolves climate risk at the level where real decisions are made, the asset, the province, the sector rather than the country average that obscures more than it reveals. It answers not whether a country faces climate risk, but exactly how much GDP is at risk, from which hazards, in which sectors, in which provinces, and in which year and what existing infrastructure already protects. The findings reflect what the data shows: that the most exposed economies benefit least from existing adaptation, that compound multi-hazard losses are systematically missed by single-peril models, and that subnational variation within a single country can be large enough to render national averages commercially unreliable. These are structural features of climate risk not edge cases, and they change how portfolios are constructed, how bonds are priced, and where adaptation investment flows. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://docs.alphageo.ai/methodology/methodology-docs/clima-metrics-climate-gdp-methodology.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.