在复合极端条件下评估天气预报的真实世界经济价值：一个特定于决策的框架

Policy Research Working Paper Assessing The Real-World Economic Valueof Weather Forecasts under Compounding Extremes A Decision-Specific Framework Leonardo OlivettiGabriele MessoriPaolo AvnerStéphane Hallegatte Urban, Disaster Risk Management,Resilience and Land Global Department & Policy Research Working Paper11407 Abstract Assessing the real-world economic value of weather fore-casts remains challenging, particularly in the context ofhigh-impact extreme events. Although meteorologicalskill has improved substantially in recent years—driven bysteady advances in physics-based models and impressivebreakthroughs in artificial intelligence-based forecasting—operational evaluations still focus primarily on standardskill metrics, with limited consideration of how improve-ments in meteorological skill translate into economic value.This study proposes a flexible framework to assess the eco- economic value of leading physics-based and data-drivenforecasting systems from the European Centre for Medi-um-Range Weather Forecasts. The value of forecasts ishighly sensitive to assumptions about compounding losses,penalty structures, and prevention costs, which often sub-stantially alter conclusions drawn from meteorological skillalone. For instance, in some cities in Southern Europe, thehigher sensitivity of the physics-based Integrated ForecastSystem high-resolution model (IFS HRES) makes it bettersuited when protection costs are small relative to poten- This paper is a product of the Urban, Disaster Risk Management, Resilience and Land Global Department and the ClimateGroup. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution todevelopment policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at leonardo.olivetti@geo.uu.se, gabriele.messori@geo.uu.se,pavner@worldbank.org, and shallegatte@worldbank.org. A verified reproducibility package for this paper is available at The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about developmentissues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry thenames of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely thoseof the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and Assessing The Real-World Economic Value ofWeather Forecasts under Compounding Leonardo Olivetti1,2,3, Gabriele Messori1,2,4, Paolo Avner5,6, andSt´ephane Hallegatte1Department of Earth Sciences, Uppsala University2Swedish Centre for Impacts of Climate Extremes (climes)3Centre of Natural Hazards and Disaster Science (CNDS)4Department of Meteorology, Stockholm University5World Bank Group6Global Facility for Disaster Reduction and Recovery (GFDRR) JELClassi ication:Q54,Q51,Q55 Keywords:Weatherforecasting;Disasterriskreduction;Compoundrisk;Economicvalueofweatherforecasts;Artificialintelligence 1Introduction Weather forecasts play a crucial role in numerous societal applications. They arefundamental to early-warning systems and disaster mitigation strategies (WorldMeteorological Organization, 2022) and are of central importance to key sectorssuch as agriculture, energy, and insurance (e.g., Kron et al., 2019).This vi-tal role is reflected in the continuous pursuit of higher forecast quality, whichhas led to steady improvements in numerical forecasting systems over the pastdecades (Bauer et al., 2015; ECMWF, 2025). More recently, data-driven fore- Traditionally, improvements in forecasting systems have been assessed pri-marily through meteorological skill metrics. While these metrics directly relateto the primary goal of weather forecasts – namely predicting future weather –they provide only an indirect link to the real-world utility of forecasts.Clas-sical economic evaluation frameworks (Thompson and Brier, 1955; Nelson andWinter, 1960; Murphy, 1969), such as the Relative Economic Value (REV),offer a bridge between forecast skill and user benefit. They enable the assess-ment of forecast value in terms of economic impact—for example by quantifying However, classic economic evaluation frameworks typically rely on the as-sumption of fixed costs and losses that are independent of the sequence orfrequency of weather events.This assumption is problematic for high-impactweather extremes. Losses associated with extreme events often scale non-linearly,compounding over time if multiple extremes occur in close succession (Worouand Messori, 2025; J¨ager et al., 2025). This effect is further aggravated in thecase of repeated forecast failures for the compounding extremes, potentiallyleading to overwhelmed infrastructure, strained emergency resp