no19-24

[24] Designing and Revising Reconstruction Policies Based on Dynamic Population Movement Predictions after Large-Scale Disasters

○Rena Koseki¹, Eiji Hato² (1. Nippon Koei Co., Ltd., 2. The University of Tokyo)

Keywords:Reconstruction, Recursive Logit model, Location choice, Network Design Problem

We aim to model the long-term evacuation process as a recursive path choice problem under uncertainty and establish a framework for modifying the reconstruction policy by optimizing the Spatio-temporal allocation of assistance policies by dynamic Network Design Problem. We also attempt to reduce estimation bias by modeling the interaction of local communities during the recovery period as the expected number of people at the next time step (externality term) and estimating it endogenously through structural estimation. The empirical analysis shows a positive externality during the reconstruction process, suggesting that out-migration promotes further out-migration and the usefulness of collective relocation projects. We formulate a network design problem to optimize rent subsidies on a Spatio-temporal network, and case studies confirmed its usability and importance in revising recovery plans.