Establishment-level Simulation of Supply Chain Disruption: The case of the Great East Japan Earthquake

Author Name INOUE Hiroyasu (University of Hyogo / RIKEN Center for Computational Science) / OKUMURA Yoshihiro (Kansai University) / TORAYASHIKI Tetsuya (Hyogo Earthquake Memorial 21st Century Research Institute) / TODO Yasuyuki (Faculty Fellow, RIETI)
Creation Date/NO. June 2022 22-E-059
Research Project Research on relationships between economic and social networks and globalization
Download / Links


This paper simulates the economic loss resulting from supply chain disruptions triggered by the Great East Japan Earthquake (GEJE) in 2011, applying data on firm-level supply chains and establishment-level attributes to an agent-based model. In particular, we improve previous studies on this issue in the following four ways by modifying the model and data and thus by estimating more accurate parameter values. First, our model incorporates more parameters, some of which vary across sectors, than the previous models. Second, our data can identify the damage to production facilities in the disaster-hit regions more accurately, using establishment-level census and survey data and geographic information system (GIS) data on the GEJE and subsequent tsunami. Third, the use of the establishment-level data enables us to capture supply chains between non-headquarter establishments in the disaster-hit regions and establishments in other regions, even though we cannot capture the whole network at the establishment level. Finally, we incorporate power outages after the GEJE that exacerbated the supply chain disruption, particularly for a few weeks immediately after the GEJE. We find that our extended method can greatly improve the capability of replicating the actual economic outcomes after the GEJE, and this improvement is mostly due to the last three improvements, and not because of the use of more parameters. Our method can be applied to predict the economic effect of future disasters, such as the Nankai Trough earthquake, on each region more accurately.