Why a small probability of terror generates a large macroeconomic impact

Abstract

It has been shown that terror activities have had a substantial macroeconomic impact. This work presents a macroeconomic model showing quantitatively that an increase in the probability of terror-induced death such as that observed in Israel in 2001–2004 is consistent with the documented contraction of economic activity associated with the impact of terror. The model includes fear of terrorism, represented as a higher level of probability perception than the cumulative-prospect theory expects, and a local government that can supply the public with a security good. The effectiveness of the production of the security good plays an important role in determining the terror-induced loss of production to the economy.

Keywords:

• Terrorism
• Macroeconomic impact

JEL Classification:

• H56

Acknowledgements

This paper was written as part of my PhD dissertation under the dedicated supervision of Prof Daniel Tsiddon and Prof Zvi Eckstein. I thank Prof Edi Karni, Prof Zvi Hercowitz and Prof Itzhak Zilcha for their helpful comments and the participants in Tel Aviv University Macro Workshop. The comments and suggestions by an anonymous referee are also gratefully acknowledged.

Notes

¹ The paper referenced is similar to ours with two main differences: Shieh et al. (2005) do not assume that government expenditure on direct anti-terrorism activities is determined optimally and they do not use any fear mechanism.

² Although this is not a strong assumption, but it is made to prevent a situation in which the population diminishes. Any population growth rate exceeding the terror rate achieves this goal; the assumption is made only to simplify the mathematical calculations.

³ In which a sure $30 is preferred over an 80% chance of obtaining $45 whereas a 20% chance of obtaining $45 is preferred over a 25% chance of obtaining $30 (Tversky and Fox 1995).

⁴ We do not assume that all government expenditures are optimally determined. Therefore, to simplify the model we assume that the agent gains no utility from any government expenditure other than direct anti-terrorism expenditure.

⁵ This is the first-order condition when the government and the agents use the same level of terror (alternatives 1 and 3). In the other alternative, the first-order condition is more complicated and does not have an analytical solution.

⁶ Estimations such as those in Cohen and Hsein (2003) found that A is 1–1.5 per year. Since each period in the OLG model accounts for 30 years, 120 means 1.17 per year.

⁷ This is based on a subjective time preference of 0.97 per period; therefore, it is about 0.4 for 30 years.

⁸ In order to show that ω(d)  =  d, note that . Then if ln d  =  x we have exp x  =  d and hence ω(d)  =  exp ( ln d)  =  exp x  =  d.

⁹ Even though terrorism leads the economy to a lower steady state, it rarely causes market contraction but ‘settles’ for causing a lower growth rate.

¹⁰