Effects of Property Recovery Incentives and Social Interaction on Self-Evacuation Decisions in Natural Disasters: An Agent-Based Modelling Approach
Agent-based simulation finds targeting well-connected households is more effective than just increasing recovery funds.
A new AI-powered study provides crucial insights for disaster response planning, revealing that strategic social targeting is more effective than blanket financial incentives. Researchers Made Krisnanda, Raymond Chiong, Yang Yang, and Kirill Glavatskiy developed an agent-based model (ABM) that simulates household evacuation decisions during natural disasters. The model employs evolutionary game theory, where household agents compete for limited shared resources like property recovery funds and coordination services, mimicking real-world scarcity during crises.
The team tested four distinct scenarios modeling different government prioritization strategies for distributing incentives. Their key finding was counterintuitive: the impact of financial incentives diminishes with both increasing funding value and broader household prioritization. This indicates a clear 'optimal level' of government support beyond which additional spending yields minimal practical benefit. More critically, the model identified that the overall community evacuation rate is heavily dependent on the underlying social network structure. The research pinpointed 'community influencers'—households with high social connectivity (node degree). Prioritizing these influencers for support and information led to significant, discontinuous jumps in the overall evacuation rate. Conversely, prioritizing less-connected households could actually impede collective evacuation efforts.
These findings shift the paradigm from purely economic interventions to socio-strategic ones. For emergency managers and policymakers, the study provides a computational framework to simulate and optimize evacuation campaigns before disaster strikes. It underscores that in resource-constrained scenarios, intelligently leveraging the existing social fabric through key influencers is a more powerful lever than simply increasing the recovery fund budget. This agent-based approach allows for testing policy efficacy in a virtual environment, potentially saving lives and resources when real disasters occur.
- Agent-based model using evolutionary game theory found diminishing returns on evacuation rates when simply increasing property recovery funds.
- Prioritizing 'community influencers' (highly connected households) caused significant jumps in overall evacuation rates, while focusing on low-connectivity agents hindered it.
- The structure of the underlying social network was a more critical factor for collective evacuation than the absolute value of government incentives.
Why It Matters
Provides governments with a data-driven model to optimize limited disaster response budgets by targeting key social connectors, not just spending more.