Tropical cyclones are a major source of climate-related losses, destroying capital and disrupting production. Global assessments often treat bottom-up damage estimation and macroeconomic impacts separately, limiting our ability to trace how asset-level damages translate into world wide production and GDP losses. We develop a framework that combines (i) asset-resolved TC direct damages from exposure data and simulated TC tracks with (ii) an agent-based model that propagates shocks through inter-industry linkages. Losses are measured relative to a stable counterfactual baseline, enabling a decomposition into direct and indirect effects. Under SSP5–8.5, expected global direct damages remain modest but approximately double in the long term, while tail risk increases sharply: a historical 1,000-year loss level becomes a ∼200-year event in the near term and a ∼20-year event by 2075–2100. Production losses shift upward similarly, and indirect losses account for more than half of total production losses on average, with strong sectoral and regional heterogeneity. Network propagation amplifies impacts in exposed regions and generates material spillovers, so similar aggregate damages can produce very different GDP outcomes depending on the sectoral-geographic pattern of impacts. These findings have direct implications for the design of adaptation measures and for estimating climate-related financial risk.
