An entropy-based approach for disaster risk assessment and humanitarian logistics operations planning in Colombia

Abstract

Purpose: This paper aims to design a vulnerability assessment model considering the multidimensional and systematic approach to disaster risk and vulnerability. This model serves to both risk mitigation and disaster preparedness phases of humanitarian logistics. Design/methodology/approach: A survey of 27,218 households in Pueblo Rico and Dosquebradas was conducted to obtain information about disaster risk for landslides, floods and collapses. We adopted a cross entropy-based approach for the measure of disaster vulnerability (Kullback–Leibler divergence), and a maximum-entropy estimation for the reconstruction of risk a priori categorization (logistic regression). The capabilities approach of Sen supported theoretically our multidimensional assessment of disaster vulnerability. Findings: Disaster vulnerability is shaped by economic, such as physical attributes of households, and health indicators, which are in specific morbidity indicators that seem to affect vulnerability outputs. Vulnerability is heterogeneous between communities/districts according to formal comparisons of Kullback–Leibler divergence. Nor social dimension, neither chronic illness indicators seem to shape vulnerability, at least for Pueblo Rico and Dosquebradas. Research limitations/implications: The results need a qualitative or case study validation at the community/district level. Practical implications: We discuss how risk mitigation policies and disaster preparedness strategies can be driven by empirical results. For example, the type of stock to preposition can vary according to the disaster or the kind of alternative policies that can be formulated on the basis of the strong relationship between morbidity and disaster risk. Originality/value: Entropy-based metrics are not widely used in humanitarian logistics literature, as well as empirical data-driven techniques.