On the Determinants of Terrorism Risk Concern in Europe

Abstract

We investigate whether differences in terrorism risk are mirrored on terrorism risk perception across European countries for the period 2003–2007. We find that the average propensity for terrorism risk concern is affected by actual risk levels. Country and individual heterogeneity contribute substantially to the variation of observed risk perception. Singles and individuals with white collar jobs are less likely to mention terrorism as one of the most pressing issues their country faces, while political positioning towards the right makes it more likely to be concerned about terrorism. As far as competing risks are concerned, we find that the likelihood terrorism is mentioned, decreases with spending on pensions as a percentage of GDP being higher. Finally, based on the Bayesian framework, we also examined the formation of terrorism risk perceptions, and decompose the observed country-level time series of terrorism activity into a long- and a short-run component. We concluded that the observed risk perception variation is only explained by the long-term trend of terrorism activity countries face.

Keywords:

• Probit
• Survey data
• Terrorism risk concern
• Time series decomposition

Notes

1 We thank an anonymous referee for this suggestion.

2 For this exercise to be meaningful, each proxy should be a significant determinant of the probability that its respective concern is mentioned. For that purpose, we estimated a probit model for each concern using its potential driver as covariate. For all cases, drivers were statistically significant. Unfortunately, we could not find an adequate driver proxy for housing, which led us to exclude it from the analysis.

3 Respondents are shown a card listing the following: crime, public transport, economic situation, rising prices/inflation, taxation, unemployment, terrorism, defence/foreign affairs, housing, immigration, health care system, educational system, pensions, protecting the environment, others (spontaneous) DK.

4 Eurobarometer issue numbers ZA: 3938, 4229, 4414, 4526, 4565.

5 Given the size of our data-set, we use the 1% significance level. However, for the reader's convenience, the tables also flag coefficients that are significant at the 5 as well as the 10% levels.