A climate stress-testing methodology for climate extreme events -related systemic risks in national production networks.

Recent events, such as the COVID-19 pandemic, underscore how localised disruptions can trigger far-reaching economic impacts through supply chain dependencies, extending indirect economic and social damages well beyond affected areas. Despite the growing recognition for the role of interdependencies on shock propagation, current models lack the granularity needed to understand and mitigate the propagation of climate shocks through interconnected supply networks.

Against this backdrop, our study proposes a firm-level climate stress-testing methodology for forecasting indirect social and economic damages arising from disruptions in production networks.

We first develop a firm-level agent-based model to simulate climate risk contagion within national supply chains. The model represents inter-firm production linkages and allows for heterogeneous behavioural responses under alternative assumptions regarding firm-level recovery dynamics, input specificity, and substitution possibilities following climate shocks. We then evaluate our model performance by comparing simulated impacts with observed indirect economic damages associated with the July 2021 and October 2022 flood events in Costa Rica. Using comprehensive administrative data from the Central Bank of Costa Rica’s electronic invoicing system, we reconstruct inter-firm transaction volumes and generate a detailed representation of the national production network. The resulting dataset is uniquely granular, combining full firm coverage (all firms being legally required to issue electronic invoices) with high temporal resolution based on monthly aggregation, allowing us to compare the model performance both at the regional and national level as well as the firm-level.

Our contribution is twofold. First, by conducting multiple simulations under alternative assumptions for a given climate extreme scenario, we explicitly account for uncertainty in the estimation of indirect economic impacts. This scenario-based approach allows us to assess the sensitivity of indirect damage estimates to key modeling assumptions. Second, by quantifying indirect impacts at the firm level and enabling aggregation at the city, district, and regional scales, the model delivers a high degree of spatial and economic granularity. The exceptional resolution of the underlying dataset allows policymakers to identify regions, firms, and communities that are most vulnerable to indirect damages associated with extreme weather events, thereby supporting more targeted and effective adaptation and risk-management strategies.