CoPS/IMPACT Working Paper Number G-349

Title: Constructing a Destructive Events Tool using Small Rectangular Areas, Computable General Equilibrium Modelling and Neural Networks

Authors: Peter Dixon, Michael Jerie, Dean Mustakinov, Maureen T. Rimmer, Nicholas Sheard, Florian Schiffmann and Glyn Wittwer

Abstract

This paper describes a destructive events tool (DET) for anticipating the national and regional economic effects of a destructive event occurring at any latitude/longitude in a country. The event is characterized by areas of complete destruction and evacuation. The event could be a natural disaster, major industrial accident, or terrorist attack.

The key ingredient for a DET is data showing population and employment by industry in small rectangular areas (SRAs). In the Poland DET, motivating the paper, there are 600,000 SRAs, each 0.5 sq km. This spatial resolution greatly improves the accuracy of the estimation of the economic impacts of events where physical impacts vary substantially across small areas.

The second ingredient is an economic model with sufficient regional/industrial definition to translate shocks at an SRA level into implications at the sub-national and national levels. This requirement is met by a multi-regional computable general equilibrium (CGE) model.

The final ingredient is an approximation for the model's reduced form. This is necessary so that the DET can be applied by organizations, without in-house CGE expertise, that need quick turnaround in a secure environment. We implement an approximation method for CGE reduced forms based on Neural Networks.

JEL classification: C81, C68, C45, H84

Keywords: Destructive events tool; Small rectangular areas; Multi-regional computable general equilibrium models; Neural network approximations to reduced forms



Working Paper Number G-349 can be downloaded in PDF format. To print this you will need the Adobe Acrobat Reader.


Go to working papers page