The Propagation of shocks from Natural Disasters through Supply Chains

Due to globalization and digitization, supply chains have become increasingly more evolving, complex, and interconnected. Recent geopolitical tensions, the covid-19 pandemic, and natural disasters have resulted in unexpected global supply chain shocks. Such shocks can propagate upstream and downstream and cause substantial indirect effects to geographically distant firms that are indirectly affected (Inoue & Todo, 2019; D.

Wang et al., 2021). Systematically underestimating the economic risks of natural disasters could lead to inefficient use of resources and poor shock recoveries.

Therefore, the main objective of my proposed research is to accurately model the shock propagation that follows from climate-related natural disasters, like floods and earthquakes. Modelling supply chain networks and the shocks that propagate through the networks expose the direct and indirect economic and infrastructural vulnerabilities that result from shocks.

I seek to leverage advanced computational methods and different data sources to create a network through which firms relate to critical infrastructures, like ports, airports, roads, railways, or telecommunications infrastructure. This would allow for a more thorough explanation of why climate risks propagate in the ways they do. The ultimate practical implementations would be to design more effective forward-looking policy interventions that can optimally mitigate risks or proactively support firms or infrastructures that are (in)directly affected by a shock.

Supply chains can be represented using weighted, directed graphs, where the nodes represent firms, edges represent supplier-customer relationships, and edge weights denote the value or volume of transactions. In production networks, geographical dependence matters for link predictions (Mungo et al., 2023), and their vulnerability to climate change developments or natural disasters.

The latter has been studied extensively for ports (Verschuur, Pant, et al., 2022). Moreover, several efforts have been made to incorporate transport networks and supply chains, focusing specifically on the integration of road networks with production networks (Colon et al., 2021).

However, a large-scale granular network that contains firm-level data and land, maritime, and airport transport routes is non-existent yet (Verschuur, Pant, et al., 2022). Hence, an aim of this research would be to create firm-level networks that map on what infrastructures trade relations between firms in the networks would likely depend, i.e. via ports, airports, railways, etc.

This could also extend the earlier analyses to inter-regional studies. As a more flexible and granular alternative to input-output (I-O) models, Agent-Based Models (ABMs) can be used to flexibly model complex economic systems. The micro-level interactions, structures and agents are modelled using highly granular data, advanced network analyses, and agent-based computer simulations.

A barrier to the application of ABMs in supply chains is the sensitivity of firm-level data and an unlikely assembly of a complete dataset soon (Wichmann et al., 2020). Therefore, there is a need to reconstruct supply chain dependencies using graph reconstruction techniques, for instance link prediction or network inference.