AI-Driven Shock-Oriented Data Envelopment Analysis for Assessing Economic Resilience in Short-Term Wars

This study develops an AI-driven, shock-oriented Data Envelopment Analysis (DEA) framework to evaluate the economic resilience of decision-making units under short-term wars. The research addresses the challenge of assessing real-time efficiency losses caused by abrupt, nonlinear shocks, which conventional methods often fail to capture. A comprehensive dataset encompassing key economic indicators, including labor, capital, energy, fiscal resources, logistical capacity, and sectoral outputs, was analyzed across pre-war, wartime, and immediate post-war phases. Artificial intelligence models, comprising autoencoders and Long Short-Term Memory networks, were applied to detect the magnitude, direction, and timing of instantaneous shocks, which were then integrated into a shock-adjusted DEA model. Efficiency scores revealed that output-constrained units suffered the most substantial performance declines, whereas input-constrained units were able to mitigate losses through adaptive resource reallocation. High baseline efficiency was associated with faster post-war recovery, emphasizing the importance of pre-conflict operational robustness. Sensitivity analysis demonstrated that even minor variations in shock intensity significantly affect resilience outcomes, underscoring the necessity for precise, high-frequency data monitoring. The proposed framework provides a quantitative, real-time tool for identifying vulnerabilities, guiding resource allocation, and monitoring recovery trajectories, offering actionable insights for policymakers and planners tasked with maintaining economic stability during short-term conflicts.

کلیدواژه ها:

Adaptive capacity ، Economic resilience ، Shock-oriented DEA ، Short-term war ، AI-based shock detection