Authors: Peter B. Dixon, Michael Jerie, Maureen T. Rimmer and Glyn Wittwer
The Terrorism Risk Assessment (TRA) groups in the Department of Homeland Security assess millions of terrorism scenarios defined by location, agent (e.g. nuclear device), and delivery method (e.g. car bomb). For each scenario they estimate deaths, injuries, property damage, clean-up and health expenses, visitor discouragement, and other damage dimensions. The TRA groups translate damages into economic measures, e.g. loss of GDP. Previously they used an input-output (I-O) model. Here we replace I-O with computable general equilibrium (CGE).
Solving CGE models is computationally time-consuming and requires specialist skills. For the TRA groups this creates two challenges: feasibility and security. A model that cannot be solved in less than a fraction of a second is infeasible for analyzing millions of scenarios. The TRAs can rely only on people with high security clearances, limiting the possibilities for obtaining specialist advice.
Our approach to these challenges was to use a CGE model to estimate elasticities that show the sensitivity of economic variables to direct damage effects of events occurring in different regions. For example, we supplied the TRA groups with CGE-based estimates of the percentage effect on national welfare of destruction of 1 per cent of the capital stock in congressional district NY14.
Our elasticity approach meets both challenges. First, for any given terrorism scenario specified by a location and a vector of direct damage shocks, the TRA groups can use the elasticities in linear equations to estimate in nanoseconds the implications for a wide range of economic variables. Second, as outside contractors, we have no need for access to sensitive information on specific shock vectors and target regions.
We describe how we used a dynamic, multi-regional CGE model, USAGE-TERM, to estimate the elasticities.
JEL classification: C68, F52, R13.
Key words: terrorism assessment, multi-regional, dynamic CGE.
Working Paper Number G-280 can be downloaded in PDF format. To print this you will need the Adobe Acrobat Reader.
Go to working papers page