A Hybrid Multi-Criteria Decision-Making Model for Assessing Social Vulnerability to Natural Disasters

Reducing the harm caused by natural disasters requires an accurate, defensible assessment of where social vulnerability is concentrated. Existing social vulnerability studies rely predominantly on conventional statistical techniques, such as principal component analyses and simple additive indices, and rarely derive indicator weights and an overall ranking within a single, internally consistent decision framework. This study addresses that gap by proposing a hybrid multi-criteria decision-making (MCDM) model that combines hierarchical data envelopment analysis (H-DEA) with the reference ideal method (RIM). H-DEA objectively derives the relative importance of four dimensions and twelve indicators of social vulnerability directly from the data, avoiding the subjectivity inherent in expert-elicited weights, while RIM ranks alternatives against a decision-maker-defined reference ideal range rather than an extreme value, thereby mitigating the rank-reversal problem associated with conventional distance-based methods. The model is applied to 161 townships and districts in Taiwan using government-sourced mitigation data covering exposure, mitigation and preparedness, response capacity and recovery. The results show that recovery capacity—in particular, local public finance and social support—contributes most to social vulnerability, followed by mitigation and preparedness. The townships and districts of highest vulnerabilities are heavily concentrated in Pingtung County. A comparison against H-DEA alone, TOPSIS and VIKOR, using Spearman’s rank correlation, confirms the stability of the proposed ranking, while a comparison against an unweighted average demonstrates the practical value of objective weighting. The findings offer local governments and disaster-management authorities an evidence-based basis for prioritizing mitigation investment.