Recent disruptions in global spatial economic networks highlight the need for models and theories that help explain when and how large-scale failures occur in these systems. This study addresses this problem by examining dragon kings (DKs)—ultra-extreme events that lie beyond the tail of power law and other statistical distributions. Although DKs appear in many systems, they remain understudied in spatial economic networks. We introduce a framework grounded in economic and complex-systems principles to predict DK failures in input-output trade networks. Using a cascading failure model, we simulate how shocks propagate through a multilayered network and generate avalanche-size distributions from which DKs are identified. We then combine these outcomes with network indicators to train machine learning classifiers that distinguish DK from non-DK failures. This approach is applied to the 2019 OECD inter-country input–output network. The analysis advances theoretical understanding of how network structure and dynamics increase the likelihood of DK failures and points to the features that heighten systemic vulnerability to such events. We conclude with recommendations for strengthening the efficiency and resilience of spatial economic networks, along with directions for future research.
Schintler et al. (2026) studied this question.