Modelling post-disaster recovery process of industrial sectors: A case study of 2016 Kumamoto earthquakes

Abstract

Understanding the post-disaster recovery process of industrial sectors is critical for ensuring a quicker recovery and estimating economic losses instigated by disasters. However, the recovery process of firms is complex, multidimensional, and the impact of uncertainty and relevant determiners on recovery has not been adequately addressed in past studies. Therefore, this research developed a recovery function for a firm's post-disaster recovery process, which integrated the impact of initial damage rates and lifeline service supply status. To model the stochastic recovery process of industrial sectors, this research establishes a multi-state semi-Markov modelling framework that considers both the sojourn time before production capacity state transitions and the probability of state transition in each production capacity state. Besides, the impacts of firm size and sector type on recovery are also acknowledged and compared. After the 2016 Kumamoto earthquakes in Japan, the proposed model was integrated into the firms' recovery process, and the results were consistent with the actual observed dataset and recovery tendency. The proposed model can evaluate the recovery probability at any post-disaster time but is conditional on the initial production capacity rate and lifeline service availability of different industrial sectors. Such a model can contribute by providing empirical evidence to decision-makers and business managers about the systematisation of recovery strategies, as well as the prediction of business recovery processes in case of future incidents.